anza-xyz-agave/wen-restart/src/wen_restart.rs

//! The `wen-restart` module handles automatic repair during a cluster restart

use {
    crate::{
        heaviest_fork_aggregate::{HeaviestForkAggregate, HeaviestForkAggregateResult},
        last_voted_fork_slots_aggregate::{
            LastVotedForkSlotsAggregate, LastVotedForkSlotsAggregateResult,
            LastVotedForkSlotsEpochInfo, LastVotedForkSlotsFinalResult,
        },
        solana::wen_restart_proto::{
            self, ConflictMessage, GenerateSnapshotRecord, HeaviestForkAggregateRecord,
            HeaviestForkRecord, LastVotedForkSlotsAggregateFinal,
            LastVotedForkSlotsAggregateRecord, LastVotedForkSlotsEpochInfoRecord,
            LastVotedForkSlotsRecord, State as RestartState, WenRestartProgress,
        },
    },
    agave_snapshots::{
        paths::{
            get_highest_full_snapshot_archive_slot, get_highest_incremental_snapshot_archive_slot,
        },
        snapshot_archive_info::SnapshotArchiveInfoGetter,
    },
    anyhow::Result,
    log::*,
    prost::Message,
    solana_clock::{Epoch, Slot},
    solana_gossip::{
        cluster_info::{ClusterInfo, GOSSIP_SLEEP_MILLIS},
        restart_crds_values::RestartLastVotedForkSlots,
    },
    solana_hash::Hash,
    solana_ledger::{
        ancestor_iterator::AncestorIterator,
        blockstore::Blockstore,
        blockstore_processor::{process_single_slot, ConfirmationProgress, ProcessOptions},
        leader_schedule_cache::LeaderScheduleCache,
    },
    solana_pubkey::Pubkey,
    solana_runtime::{
        accounts_background_service::AbsStatus,
        bank::Bank,
        bank_forks::BankForks,
        snapshot_bank_utils::{
            bank_to_full_snapshot_archive, bank_to_incremental_snapshot_archive,
        },
        snapshot_controller::SnapshotController,
        snapshot_utils::purge_all_bank_snapshots,
    },
    solana_shred_version::compute_shred_version,
    solana_svm_timings::ExecuteTimings,
    solana_time_utils::timestamp,
    solana_vote::vote_transaction::VoteTransaction,
    std::{
        collections::{HashMap, HashSet},
        fs::{read, File},
        io::{Cursor, Write},
        path::{Path, PathBuf},
        str::FromStr,
        sync::{
            atomic::{AtomicBool, Ordering},
            Arc, RwLock,
        },
        thread::sleep,
        time::{Duration, Instant},
    },
};

// If >42% of the validators have this block, repair this block locally.
const REPAIR_THRESHOLD: f64 = 0.42;
// When counting Heaviest Fork, only count those with no less than
// 67% - 5% - (100% - active_stake) = active_stake - 38% stake.
// 67% is the supermajority threshold (2/3), 5% is the assumption we
// made regarding how much non-conforming/offline validators the
// algorithm can tolerate.
const HEAVIEST_FORK_THRESHOLD_DELTA: f64 = 0.38;
// The coordinator print new stats every 10 seconds.
const COORDINATOR_STAT_PRINT_INTERVAL_SECONDS: u64 = 10;

#[derive(Debug, PartialEq)]
pub enum WenRestartError {
    BankHashMismatch(Slot, Hash, Hash),
    BlockNotFound(Slot),
    BlockNotFull(Slot),
    BlockNotFrozenAfterReplay(Slot, Option<String>),
    BlockNotLinkedToExpectedParent(Slot, Option<Slot>, Slot),
    ChildStakeLargerThanParent(Slot, u64, Slot, u64),
    Exiting,
    FutureSnapshotExists(Slot, Slot, String),
    GenerateSnapshotWhenOneExists(Slot, String),
    GenerateSnapshotWhenDisabled,
    HeaviestForkOnLeaderOnDifferentFork(Slot, Slot),
    MalformedLastVotedForkSlotsProtobuf(Option<LastVotedForkSlotsRecord>),
    MalformedProgress(RestartState, String),
    MissingLastVotedForkSlots,
    MissingSnapshotInProtobuf,
    NotEnoughStakeAgreeingWithUs(Slot, Hash, HashMap<(Slot, Hash), u64>),
    UnexpectedState(wen_restart_proto::State),
}

impl std::fmt::Display for WenRestartError {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        match self {
            WenRestartError::BankHashMismatch(slot, expected, actual) => {
                write!(
                    f,
                    "Bank hash mismatch for slot: {slot} expected: {expected} actual: {actual}"
                )
            }
            WenRestartError::BlockNotFound(slot) => {
                write!(f, "Block not found: {slot}")
            }
            WenRestartError::BlockNotFull(slot) => {
                write!(f, "Block not full: {slot}")
            }
            WenRestartError::BlockNotFrozenAfterReplay(slot, err) => {
                write!(f, "Block not frozen after replay: {slot} {err:?}")
            }
            WenRestartError::BlockNotLinkedToExpectedParent(slot, parent, expected_parent) => {
                write!(
                    f,
                    "Block {slot} is not linked to expected parent {expected_parent} but to \
                     {parent:?}"
                )
            }
            WenRestartError::ChildStakeLargerThanParent(
                slot,
                child_stake,
                parent,
                parent_stake,
            ) => {
                write!(
                    f,
                    "Block {slot} has more stake {child_stake} than its parent {parent} with \
                     stake {parent_stake}"
                )
            }
            WenRestartError::Exiting => write!(f, "Exiting"),
            WenRestartError::FutureSnapshotExists(slot, highest_slot, directory) => {
                write!(
                    f,
                    "Future snapshot exists for slot: {slot} highest slot: {highest_slot} in \
                     directory: {directory}",
                )
            }
            WenRestartError::GenerateSnapshotWhenOneExists(slot, directory) => {
                write!(
                    f,
                    "Generate snapshot when one exists for slot: {slot} in directory: {directory}",
                )
            }
            WenRestartError::GenerateSnapshotWhenDisabled => {
                write!(f, "Generate snapshot when snapshots are disabled")
            }
            WenRestartError::HeaviestForkOnLeaderOnDifferentFork(
                coordinator_heaviest_slot,
                should_include_slot,
            ) => {
                write!(
                    f,
                    "Heaviest fork on coordinator on different fork: heaviest: \
                     {coordinator_heaviest_slot} does not include: {should_include_slot}",
                )
            }
            WenRestartError::MalformedLastVotedForkSlotsProtobuf(record) => {
                write!(f, "Malformed last voted fork slots protobuf: {record:?}")
            }
            WenRestartError::MalformedProgress(state, missing) => {
                write!(f, "Malformed progress: {state:?} missing {missing}")
            }
            WenRestartError::MissingLastVotedForkSlots => {
                write!(f, "Missing last voted fork slots")
            }
            WenRestartError::MissingSnapshotInProtobuf => {
                write!(f, "Missing snapshot in protobuf")
            }
            WenRestartError::NotEnoughStakeAgreeingWithUs(slot, hash, block_stake_map) => {
                write!(
                    f,
                    "Not enough stake agreeing with our slot: {slot} hash: {hash}\n \
                     {block_stake_map:?}",
                )
            }
            WenRestartError::UnexpectedState(state) => {
                write!(f, "Unexpected state: {state:?}")
            }
        }
    }
}

impl std::error::Error for WenRestartError {}

// We need a WenRestartProgressInternalState so we can convert the protobuf written in file
// into internal data structure in the initialize function. It should be easily
// convertible to and from WenRestartProgress protobuf.
#[derive(Debug, PartialEq)]
pub(crate) enum WenRestartProgressInternalState {
    Init {
        last_voted_fork_slots: Vec<Slot>,
        last_vote_bankhash: Hash,
    },
    LastVotedForkSlots {
        last_voted_fork_slots: Vec<Slot>,
        aggregate_final_result: Option<LastVotedForkSlotsFinalResult>,
    },
    FindHeaviestFork {
        aggregate_final_result: LastVotedForkSlotsFinalResult,
        my_heaviest_fork: Option<HeaviestForkRecord>,
    },
    HeaviestFork {
        my_heaviest_fork_slot: Slot,
        my_heaviest_fork_hash: Hash,
    },
    GenerateSnapshot {
        my_heaviest_fork_slot: Slot,
        my_snapshot: Option<GenerateSnapshotRecord>,
    },
    Done {
        slot: Slot,
        hash: Hash,
        shred_version: u16,
    },
}

pub(crate) fn send_restart_last_voted_fork_slots(
    cluster_info: Arc<ClusterInfo>,
    last_voted_fork_slots: &[Slot],
    last_vote_bankhash: Hash,
) -> Result<LastVotedForkSlotsRecord> {
    cluster_info.push_restart_last_voted_fork_slots(last_voted_fork_slots, last_vote_bankhash)?;
    Ok(LastVotedForkSlotsRecord {
        last_voted_fork_slots: last_voted_fork_slots.to_vec(),
        last_vote_bankhash: last_vote_bankhash.to_string(),
        shred_version: cluster_info.my_shred_version() as u32,
        wallclock: timestamp(),
    })
}

pub(crate) fn aggregate_restart_last_voted_fork_slots(
    wen_restart_path: &PathBuf,
    wait_for_supermajority_threshold_percent: u64,
    cluster_info: Arc<ClusterInfo>,
    last_voted_fork_slots: &Vec<Slot>,
    bank_forks: Arc<RwLock<BankForks>>,
    blockstore: Arc<Blockstore>,
    wen_restart_repair_slots: Arc<RwLock<Vec<Slot>>>,
    exit: Arc<AtomicBool>,
    progress: &mut WenRestartProgress,
) -> Result<LastVotedForkSlotsFinalResult> {
    let root_bank = bank_forks.read().unwrap().root_bank();
    let root_slot = root_bank.slot();
    let mut last_voted_fork_slots_aggregate = LastVotedForkSlotsAggregate::new(
        root_bank.clone(),
        REPAIR_THRESHOLD,
        last_voted_fork_slots,
        &cluster_info.id(),
    );
    if let Some(aggregate_record) = &progress.last_voted_fork_slots_aggregate {
        for (key_string, message) in &aggregate_record.received {
            if let Err(e) =
                last_voted_fork_slots_aggregate.aggregate_from_record(key_string, message)
            {
                error!("Failed to aggregate from record: {e:?}");
            }
        }
    } else {
        progress.last_voted_fork_slots_aggregate = Some(LastVotedForkSlotsAggregateRecord {
            received: HashMap::new(),
            final_result: None,
        });
    }
    let mut cursor = solana_gossip::crds::Cursor::default();
    let mut is_full_slots = HashSet::new();
    let mut old_progress = WenRestartProgress::default();
    loop {
        if exit.load(Ordering::Relaxed) {
            return Err(WenRestartError::Exiting.into());
        }
        let start = timestamp();
        for new_last_voted_fork_slots in cluster_info.get_restart_last_voted_fork_slots(&mut cursor)
        {
            let from = new_last_voted_fork_slots.from.to_string();
            match last_voted_fork_slots_aggregate.aggregate(new_last_voted_fork_slots) {
                LastVotedForkSlotsAggregateResult::Inserted(record) => {
                    progress
                        .last_voted_fork_slots_aggregate
                        .as_mut()
                        .unwrap()
                        .received
                        .insert(from, record);
                }
                LastVotedForkSlotsAggregateResult::DifferentVersionExists(
                    old_record,
                    new_record,
                ) => {
                    info!(
                        "Different LastVotedForkSlots message exists from {from}: {old_record:#?} \
                         vs {new_record:#?}"
                    );
                    progress.conflict_message.insert(
                        from,
                        ConflictMessage {
                            old_message: format!("{old_record:?}"),
                            new_message: format!("{new_record:?}"),
                        },
                    );
                }
                LastVotedForkSlotsAggregateResult::AlreadyExists => (),
            }
        }
        // Because all operations on the aggregate are called from this single thread, we can
        // fetch all results separately without worrying about them being out of sync. We can
        // also use returned iterator without the vector changing underneath us.
        let active_percent = last_voted_fork_slots_aggregate.min_active_percent();
        let mut filtered_slots: Vec<Slot>;
        {
            filtered_slots = last_voted_fork_slots_aggregate
                .slots_to_repair_iter()
                .filter(|slot| {
                    if *slot <= &root_slot || is_full_slots.contains(*slot) {
                        return false;
                    }
                    if blockstore.is_full(**slot) {
                        is_full_slots.insert(**slot);
                        false
                    } else {
                        true
                    }
                })
                .cloned()
                .collect();
        }
        filtered_slots.sort();
        if progress != &old_progress {
            info!(
                "Active peers: {} Slots to repair: {:?}",
                active_percent, &filtered_slots
            );
            write_wen_restart_records(wen_restart_path, progress)?;
            old_progress = progress.clone();
        }
        if filtered_slots.is_empty()
            && active_percent >= wait_for_supermajority_threshold_percent as f64
        {
            *wen_restart_repair_slots.write().unwrap() = vec![];
            break;
        }
        {
            *wen_restart_repair_slots.write().unwrap() = filtered_slots;
        }
        let elapsed = timestamp().saturating_sub(start);
        let time_left = GOSSIP_SLEEP_MILLIS.saturating_sub(elapsed);
        if time_left > 0 {
            sleep(Duration::from_millis(time_left));
        }
    }
    Ok(last_voted_fork_slots_aggregate.get_final_result())
}

fn is_over_stake_threshold(
    epoch_info_vec: &[LastVotedForkSlotsEpochInfo],
    epoch: Epoch,
    stake: &u64,
) -> bool {
    epoch_info_vec
        .iter()
        .find(|info| info.epoch == epoch)
        .is_some_and(|info| {
            let threshold = info
                .actively_voting_stake
                .checked_sub((info.total_stake as f64 * HEAVIEST_FORK_THRESHOLD_DELTA) as u64)
                .unwrap();
            stake >= &threshold
        })
}

// Verify that all blocks with at least (active_stake_percnet - 38%) of the stake form a
// single chain from the root, and use the highest slot in the blocks as the heaviest fork.
// Please see SIMD 46 "gossip current heaviest fork" for correctness proof.
pub(crate) fn find_heaviest_fork(
    aggregate_final_result: LastVotedForkSlotsFinalResult,
    bank_forks: Arc<RwLock<BankForks>>,
    blockstore: Arc<Blockstore>,
    exit: Arc<AtomicBool>,
) -> Result<(Slot, Hash)> {
    let root_bank = bank_forks.read().unwrap().root_bank();
    let root_slot = root_bank.slot();
    let mut slots = aggregate_final_result
        .slots_stake_map
        .iter()
        .filter(|(slot, stake)| {
            **slot > root_slot
                && is_over_stake_threshold(
                    &aggregate_final_result.epoch_info_vec,
                    root_bank.epoch_schedule().get_epoch(**slot),
                    stake,
                )
        })
        .map(|(slot, _)| *slot)
        .collect::<Vec<Slot>>();
    slots.sort();

    // The heaviest slot we selected will always be the last of the slots list, or root if the list is empty.
    let heaviest_fork_slot = slots.last().map_or(root_slot, |x| *x);

    let mut expected_parent = root_slot;
    for slot in &slots {
        if exit.load(Ordering::Relaxed) {
            return Err(WenRestartError::Exiting.into());
        }
        if let Ok(Some(block_meta)) = blockstore.meta(*slot) {
            if block_meta.parent_slot != Some(expected_parent) {
                if expected_parent == root_slot {
                    error!(
                        "First block {slot} in repair list not linked to local root {root_slot}, \
                         this could mean our root is too old"
                    );
                } else {
                    error!(
                        "Block {slot} in blockstore is not linked to expected parent from Wen \
                         Restart {expected_parent} but to Block {:?}",
                        block_meta.parent_slot
                    );
                }
                return Err(WenRestartError::BlockNotLinkedToExpectedParent(
                    *slot,
                    block_meta.parent_slot,
                    expected_parent,
                )
                .into());
            }
            if !block_meta.is_full() {
                return Err(WenRestartError::BlockNotFull(*slot).into());
            }
            expected_parent = *slot;
        } else {
            return Err(WenRestartError::BlockNotFound(*slot).into());
        }
    }
    let heaviest_fork_bankhash = find_bankhash_of_heaviest_fork(
        heaviest_fork_slot,
        slots,
        blockstore.clone(),
        bank_forks.clone(),
        &exit,
    )?;
    info!("Heaviest fork found: slot: {heaviest_fork_slot}, bankhash: {heaviest_fork_bankhash:?}");
    Ok((heaviest_fork_slot, heaviest_fork_bankhash))
}

fn check_slot_smaller_than_intended_snapshot_slot(
    slot: Slot,
    intended_snapshot_slot: Slot,
    directory: &Path,
) -> Result<()> {
    match slot.cmp(&intended_snapshot_slot) {
        std::cmp::Ordering::Greater => Err(WenRestartError::FutureSnapshotExists(
            intended_snapshot_slot,
            slot,
            directory.to_string_lossy().to_string(),
        )
        .into()),
        std::cmp::Ordering::Equal => Err(WenRestartError::GenerateSnapshotWhenOneExists(
            slot,
            directory.to_string_lossy().to_string(),
        )
        .into()),
        std::cmp::Ordering::Less => Ok(()),
    }
}

// Given the agreed upon slot, add hard fork and rehash the corresponding bank, then
// generate new snapshot. Generate incremental snapshot if possible, but generate full
// snapshot if there is no full snapshot or snapshot generation is turned off (in this
// case the incremental snasphot based on the full snapshot is incorrect).
//
// We don't use set_root() explicitly, because it may kick off snapshot requests, we
// can't have multiple snapshot requests in progress. In bank_to_snapshot_archive()
// everything set_root() does will be done (without bank_forks setting root). So
// when we restart from the snapshot bank on my_heaviest_fork_slot will become root.
pub(crate) fn generate_snapshot(
    bank_forks: Arc<RwLock<BankForks>>,
    snapshot_controller: &SnapshotController,
    abs_status: &AbsStatus,
    genesis_config_hash: Hash,
    my_heaviest_fork_slot: Slot,
) -> Result<GenerateSnapshotRecord> {
    let new_root_bank;
    {
        let my_bank_forks = bank_forks.read().unwrap();
        let old_root_bank = my_bank_forks.root_bank();
        if !old_root_bank
            .hard_forks()
            .iter()
            .any(|(slot, _)| slot == &my_heaviest_fork_slot)
        {
            old_root_bank.register_hard_fork(my_heaviest_fork_slot);
        }
        // my_heaviest_fork_slot is guaranteed to have a bank in bank_forks, it's checked in
        // find_bankhash_of_heaviest_fork().
        match my_bank_forks.get(my_heaviest_fork_slot) {
            Some(bank) => new_root_bank = bank.clone(),
            None => {
                return Err(WenRestartError::BlockNotFound(my_heaviest_fork_slot).into());
            }
        }
        let mut banks = vec![&new_root_bank];
        let parents = new_root_bank.parents();
        banks.extend(parents.iter());
    }

    // Snapshot generation calls AccountsDb background tasks (flush/clean/shrink).
    // These cannot run concurrent with each other, so we must shutdown
    // AccountsBackgroundService before proceeding.
    abs_status.stop();
    info!("Waiting for AccountsBackgroundService to stop");
    while abs_status.is_running() {
        std::thread::yield_now();
    }

    let snapshot_config = snapshot_controller.snapshot_config();
    let mut directory = &snapshot_config.full_snapshot_archives_dir;
    // Calculate the full_snapshot_slot an incremental snapshot should depend on. If the
    // validator is configured not the generate snapshot, it will only have the initial
    // snapshot on disk, which might be too old to generate an incremental snapshot from.
    // In this case we also set full_snapshot_slot to None.
    let full_snapshot_slot = if snapshot_config.should_generate_snapshots() {
        get_highest_full_snapshot_archive_slot(directory)
    } else {
        None
    };
    // In very rare cases it's possible that the local root is not on the heaviest fork, so the
    // validator generated snapshot for slots > local root. If the cluster agreed upon restart
    // slot my_heaviest_fork_slot is less than the current highest full_snapshot_slot, that means the
    // locally rooted full_snapshot_slot will be rolled back. this requires human inspection。
    //
    // In even rarer cases, the selected slot might be the latest full snapshot slot. We could
    // just re-generate a new snapshot to make sure the snapshot is up to date after hard fork,
    // but for now we just return an error to keep the code simple.
    let new_snapshot_path = if let Some(full_snapshot_slot) = full_snapshot_slot {
        check_slot_smaller_than_intended_snapshot_slot(
            full_snapshot_slot,
            my_heaviest_fork_slot,
            directory,
        )?;
        directory = &snapshot_config.incremental_snapshot_archives_dir;
        if let Some(incremental_snapshot_slot) =
            get_highest_incremental_snapshot_archive_slot(directory, full_snapshot_slot)
        {
            check_slot_smaller_than_intended_snapshot_slot(
                incremental_snapshot_slot,
                my_heaviest_fork_slot,
                directory,
            )?;
        }
        bank_to_incremental_snapshot_archive(
            &snapshot_config.bank_snapshots_dir,
            &new_root_bank,
            full_snapshot_slot,
            Some(snapshot_config.snapshot_version),
            &snapshot_config.full_snapshot_archives_dir,
            &snapshot_config.incremental_snapshot_archives_dir,
            snapshot_config.archive_format,
        )?
        .path()
        .display()
        .to_string()
    } else {
        info!(
            "Can't find full snapshot, generating full snapshot for slot: {my_heaviest_fork_slot}"
        );
        bank_to_full_snapshot_archive(
            &snapshot_config.bank_snapshots_dir,
            &new_root_bank,
            Some(snapshot_config.snapshot_version),
            &snapshot_config.full_snapshot_archives_dir,
            &snapshot_config.incremental_snapshot_archives_dir,
            snapshot_config.archive_format,
        )?
        .path()
        .display()
        .to_string()
    };
    let new_shred_version =
        compute_shred_version(&genesis_config_hash, Some(&new_root_bank.hard_forks()));
    info!("wen_restart snapshot generated on {new_snapshot_path} base slot {full_snapshot_slot:?}");
    // We might have bank snapshots past the my_heaviest_fork_slot, we need to purge them.
    purge_all_bank_snapshots(&snapshot_config.bank_snapshots_dir);
    Ok(GenerateSnapshotRecord {
        path: new_snapshot_path,
        slot: my_heaviest_fork_slot,
        bankhash: new_root_bank.hash().to_string(),
        shred_version: new_shred_version as u32,
    })
}

// Find the hash of the heaviest fork, if block hasn't been replayed, replay to get the hash.
pub(crate) fn find_bankhash_of_heaviest_fork(
    heaviest_fork_slot: Slot,
    slots: Vec<Slot>,
    blockstore: Arc<Blockstore>,
    bank_forks: Arc<RwLock<BankForks>>,
    exit: &AtomicBool,
) -> Result<Hash> {
    if let Some(hash) = bank_forks
        .read()
        .unwrap()
        .get(heaviest_fork_slot)
        .map(|bank| bank.hash())
    {
        return Ok(hash);
    }
    let root_bank = bank_forks.read().unwrap().root_bank();
    let leader_schedule_cache = LeaderScheduleCache::new_from_bank(&root_bank);
    let replay_tx_thread_pool = rayon::ThreadPoolBuilder::new()
        .thread_name(|i| format!("solReplayTx{i:02}"))
        .build()
        .expect("new rayon threadpool");
    let mut timing = ExecuteTimings::default();
    let opts = ProcessOptions::default();
    // Now replay all the missing blocks.
    let mut parent_bank = root_bank;
    for slot in slots {
        if exit.load(Ordering::Relaxed) {
            return Err(WenRestartError::Exiting.into());
        }
        let saved_bank = bank_forks.read().unwrap().get_with_scheduler(slot);
        let bank_with_scheduler = saved_bank.unwrap_or_else(|| {
            let new_bank = Bank::new_from_parent(
                parent_bank.clone(),
                &leader_schedule_cache
                    .slot_leader_at(slot, Some(&parent_bank))
                    .unwrap(),
                slot,
            );
            bank_forks.write().unwrap().insert_from_ledger(new_bank)
        });
        let bank = if bank_with_scheduler.is_frozen() {
            bank_with_scheduler.clone_without_scheduler()
        } else {
            let mut progress = ConfirmationProgress::new(parent_bank.last_blockhash());
            if let Err(e) = process_single_slot(
                &blockstore,
                &bank_with_scheduler,
                &replay_tx_thread_pool,
                &opts,
                &mut progress,
                None,
                None,
                None,
                &mut timing,
            ) {
                return Err(
                    WenRestartError::BlockNotFrozenAfterReplay(slot, Some(e.to_string())).into(),
                );
            }
            let cur_bank;
            {
                cur_bank = bank_forks
                    .read()
                    .unwrap()
                    .get(slot)
                    .expect("bank should have been just inserted");
            }
            cur_bank
        };
        parent_bank = bank;
    }
    Ok(parent_bank.hash())
}

// Aggregate the heaviest fork at the coordinator.
pub(crate) fn aggregate_restart_heaviest_fork(
    wen_restart_path: &PathBuf,
    cluster_info: Arc<ClusterInfo>,
    bank_forks: Arc<RwLock<BankForks>>,
    exit: Arc<AtomicBool>,
    progress: &mut WenRestartProgress,
) -> Result<()> {
    let root_bank = bank_forks.read().unwrap().root_bank();
    if progress.my_heaviest_fork.is_none() {
        return Err(WenRestartError::MalformedProgress(
            RestartState::HeaviestFork,
            "my_heaviest_fork".to_string(),
        )
        .into());
    }
    let my_heaviest_fork = progress.my_heaviest_fork.clone().unwrap();
    let heaviest_fork_slot = my_heaviest_fork.slot;
    let heaviest_fork_hash = Hash::from_str(&my_heaviest_fork.bankhash)?;
    // Use the epoch_stakes associated with the heaviest fork slot we picked.
    let epoch_stakes = root_bank
        .epoch_stakes(root_bank.epoch_schedule().get_epoch(heaviest_fork_slot))
        .unwrap();
    let total_stake = epoch_stakes.total_stake();
    let mut heaviest_fork_aggregate = HeaviestForkAggregate::new(
        cluster_info.my_shred_version(),
        epoch_stakes,
        heaviest_fork_slot,
        heaviest_fork_hash,
        &cluster_info.id(),
    );
    if let Some(aggregate_record) = &progress.heaviest_fork_aggregate {
        for message in &aggregate_record.received {
            if let Err(e) = heaviest_fork_aggregate.aggregate_from_record(message) {
                // Do not abort wen_restart if we got one malformed message.
                error!("Failed to aggregate from record: {e:?}");
            }
        }
    } else {
        progress.heaviest_fork_aggregate = Some(HeaviestForkAggregateRecord {
            received: Vec::new(),
            total_active_stake: 0,
        });
    }

    let mut cursor = solana_gossip::crds::Cursor::default();
    let mut total_active_stake = 0;
    let mut stat_printed_at = Instant::now();
    let mut old_progress = WenRestartProgress::default();
    loop {
        if exit.load(Ordering::Relaxed) {
            return Ok(());
        }
        let start = timestamp();
        for new_heaviest_fork in cluster_info.get_restart_heaviest_fork(&mut cursor) {
            info!("Received new heaviest fork: {new_heaviest_fork:?}");
            let from = new_heaviest_fork.from.to_string();
            match heaviest_fork_aggregate.aggregate(new_heaviest_fork) {
                HeaviestForkAggregateResult::Inserted(record) => {
                    info!("Successfully aggregated new heaviest fork: {record:?}");
                    progress
                        .heaviest_fork_aggregate
                        .as_mut()
                        .unwrap()
                        .received
                        .push(record);
                }
                HeaviestForkAggregateResult::DifferentVersionExists(old_record, new_record) => {
                    warn!(
                        "Different version from {from} exists old {old_record:#?} vs new \
                         {new_record:#?}"
                    );
                    progress.conflict_message.insert(
                        from,
                        ConflictMessage {
                            old_message: format!("{old_record:?}"),
                            new_message: format!("{new_record:?}"),
                        },
                    );
                }
                HeaviestForkAggregateResult::ZeroStakeIgnored => (),
                HeaviestForkAggregateResult::AlreadyExists => (),
                HeaviestForkAggregateResult::Malformed => (),
            }
        }
        let current_total_active_stake = heaviest_fork_aggregate.total_active_stake();
        if current_total_active_stake > total_active_stake {
            total_active_stake = current_total_active_stake;
            progress
                .heaviest_fork_aggregate
                .as_mut()
                .unwrap()
                .total_active_stake = current_total_active_stake;
        }
        if old_progress != *progress {
            info!(
                "Total active stake: {} Total stake {}",
                heaviest_fork_aggregate.total_active_stake(),
                total_stake
            );
            write_wen_restart_records(wen_restart_path, progress)?;
            old_progress = progress.clone();
        }
        let elapsed = timestamp().saturating_sub(start);
        let time_left = GOSSIP_SLEEP_MILLIS.saturating_sub(elapsed);
        if time_left > 0 {
            sleep(Duration::from_millis(time_left));
        }
        // Print the block stake map after a while.
        if stat_printed_at.elapsed() > Duration::from_secs(COORDINATOR_STAT_PRINT_INTERVAL_SECONDS)
        {
            heaviest_fork_aggregate.print_block_stake_map();
            stat_printed_at = Instant::now();
        }
    }
}

pub(crate) fn repair_heaviest_fork(
    my_heaviest_fork_slot: Slot,
    heaviest_slot: Slot,
    exit: Arc<AtomicBool>,
    blockstore: Arc<Blockstore>,
    wen_restart_repair_slots: Arc<RwLock<Vec<Slot>>>,
) -> Result<()> {
    loop {
        if exit.load(Ordering::Relaxed) {
            return Err(WenRestartError::Exiting.into());
        }
        // Repair all ancestors of heaviest_slot (including itself) which are larger than
        // my_heaviest_fork_slot.
        let to_repair = if blockstore.meta(heaviest_slot).is_ok_and(|x| x.is_some()) {
            AncestorIterator::new_inclusive(heaviest_slot, &blockstore)
                .take_while(|slot| *slot > my_heaviest_fork_slot)
                .filter(|slot| !blockstore.is_full(*slot))
                .collect()
        } else {
            vec![heaviest_slot]
        };
        info!("wen_restart repair slots: {to_repair:?}");
        if to_repair.is_empty() {
            return Ok(()); // All blocks are full
        }
        *wen_restart_repair_slots.write().unwrap() = to_repair;
        sleep(Duration::from_millis(GOSSIP_SLEEP_MILLIS));
    }
}

pub(crate) fn verify_coordinator_heaviest_fork(
    my_heaviest_fork_slot: Slot,
    coordinator_heaviest_slot: Slot,
    coordinator_heaviest_hash: &Hash,
    bank_forks: Arc<RwLock<BankForks>>,
    blockstore: Arc<Blockstore>,
    exit: Arc<AtomicBool>,
    wen_restart_repair_slots: Arc<RwLock<Vec<Slot>>>,
) -> Result<()> {
    repair_heaviest_fork(
        my_heaviest_fork_slot,
        coordinator_heaviest_slot,
        exit.clone(),
        blockstore.clone(),
        wen_restart_repair_slots.clone(),
    )?;
    let root_slot = bank_forks.read().unwrap().root_bank().slot();
    let mut coordinator_heaviest_slot_ancestors: Vec<Slot> =
        AncestorIterator::new_inclusive(coordinator_heaviest_slot, &blockstore)
            .take_while(|slot| slot >= &root_slot)
            .collect();
    coordinator_heaviest_slot_ancestors.sort();
    if !coordinator_heaviest_slot_ancestors.contains(&root_slot) {
        return Err(WenRestartError::HeaviestForkOnLeaderOnDifferentFork(
            coordinator_heaviest_slot,
            root_slot,
        )
        .into());
    }
    if coordinator_heaviest_slot > my_heaviest_fork_slot
        && !coordinator_heaviest_slot_ancestors.contains(&my_heaviest_fork_slot)
    {
        return Err(WenRestartError::HeaviestForkOnLeaderOnDifferentFork(
            coordinator_heaviest_slot,
            my_heaviest_fork_slot,
        )
        .into());
    }
    if coordinator_heaviest_slot < my_heaviest_fork_slot
        && !AncestorIterator::new(my_heaviest_fork_slot, &blockstore)
            .any(|slot| slot == coordinator_heaviest_slot)
    {
        return Err(WenRestartError::HeaviestForkOnLeaderOnDifferentFork(
            coordinator_heaviest_slot,
            my_heaviest_fork_slot,
        )
        .into());
    }
    let my_bankhash = if !coordinator_heaviest_slot_ancestors.is_empty() {
        find_bankhash_of_heaviest_fork(
            coordinator_heaviest_slot,
            coordinator_heaviest_slot_ancestors,
            blockstore.clone(),
            bank_forks.clone(),
            &exit,
        )?
    } else {
        bank_forks
            .read()
            .unwrap()
            .get(coordinator_heaviest_slot)
            .unwrap()
            .hash()
    };
    if my_bankhash != *coordinator_heaviest_hash {
        return Err(WenRestartError::BankHashMismatch(
            coordinator_heaviest_slot,
            my_bankhash,
            *coordinator_heaviest_hash,
        )
        .into());
    }
    Ok(())
}

pub(crate) fn receive_restart_heaviest_fork(
    wen_restart_coordinator: Pubkey,
    cluster_info: Arc<ClusterInfo>,
    exit: Arc<AtomicBool>,
    progress: &mut WenRestartProgress,
) -> Result<(Slot, Hash)> {
    let mut cursor = solana_gossip::crds::Cursor::default();
    loop {
        if exit.load(Ordering::Relaxed) {
            return Err(WenRestartError::Exiting.into());
        }
        for new_heaviest_fork in cluster_info.get_restart_heaviest_fork(&mut cursor) {
            if new_heaviest_fork.from == wen_restart_coordinator {
                info!(
                    "Received new heaviest fork from coordinator: {wen_restart_coordinator} \
                     {new_heaviest_fork:?}"
                );
                let coordinator_heaviest_slot = new_heaviest_fork.last_slot;
                let coordinator_heaviest_hash = new_heaviest_fork.last_slot_hash;
                progress.coordinator_heaviest_fork = Some(HeaviestForkRecord {
                    slot: coordinator_heaviest_slot,
                    bankhash: coordinator_heaviest_hash.to_string(),
                    total_active_stake: 0,
                    wallclock: new_heaviest_fork.wallclock,
                    shred_version: new_heaviest_fork.shred_version as u32,
                    from: new_heaviest_fork.from.to_string(),
                });
                return Ok((coordinator_heaviest_slot, coordinator_heaviest_hash));
            }
        }
        sleep(Duration::from_millis(GOSSIP_SLEEP_MILLIS));
    }
}

pub(crate) fn send_and_receive_heaviest_fork(
    my_heaviest_fork_slot: Slot,
    my_heaviest_fork_hash: Hash,
    config: &WenRestartConfig,
    progress: &mut WenRestartProgress,
    pushfn: impl FnOnce(Slot, Hash),
) -> Result<(Slot, Hash)> {
    if config.cluster_info.id() == config.wen_restart_coordinator {
        pushfn(my_heaviest_fork_slot, my_heaviest_fork_hash);
        Ok((my_heaviest_fork_slot, my_heaviest_fork_hash))
    } else {
        let (coordinator_slot, coordinator_hash) = receive_restart_heaviest_fork(
            config.wen_restart_coordinator,
            config.cluster_info.clone(),
            config.exit.clone(),
            progress,
        )?;
        match verify_coordinator_heaviest_fork(
            my_heaviest_fork_slot,
            coordinator_slot,
            &coordinator_hash,
            config.bank_forks.clone(),
            config.blockstore.clone(),
            config.exit.clone(),
            config.wen_restart_repair_slots.clone().unwrap(),
        ) {
            Ok(()) => pushfn(coordinator_slot, coordinator_hash),
            Err(e) => {
                warn!("Failed to verify coordinator heaviest fork: {e:?}, exit soon");
                pushfn(my_heaviest_fork_slot, my_heaviest_fork_hash);
                // flush_push_queue only flushes the messages to crds, doesn't guarantee
                // sending them out, so we still need to wait for a while before exiting.
                config.cluster_info.flush_push_queue();
                sleep(Duration::from_millis(GOSSIP_SLEEP_MILLIS));
                return Err(e);
            }
        }
        Ok((coordinator_slot, coordinator_hash))
    }
}

#[derive(Clone)]
pub struct WenRestartConfig {
    pub wen_restart_path: PathBuf,
    pub wen_restart_coordinator: Pubkey,
    pub last_vote: VoteTransaction,
    pub blockstore: Arc<Blockstore>,
    pub cluster_info: Arc<ClusterInfo>,
    pub bank_forks: Arc<RwLock<BankForks>>,
    pub wen_restart_repair_slots: Option<Arc<RwLock<Vec<Slot>>>>,
    pub wait_for_supermajority_threshold_percent: u64,
    pub snapshot_controller: Option<Arc<SnapshotController>>,
    pub abs_status: AbsStatus,
    pub genesis_config_hash: Hash,
    pub exit: Arc<AtomicBool>,
}

pub fn wait_for_wen_restart(config: WenRestartConfig) -> Result<()> {
    let (mut state, mut progress) = initialize(
        &config.wen_restart_path,
        config.last_vote.clone(),
        config.blockstore.clone(),
    )?;
    loop {
        state = match state {
            WenRestartProgressInternalState::Init {
                last_voted_fork_slots,
                last_vote_bankhash,
            } => {
                progress.my_last_voted_fork_slots = Some(send_restart_last_voted_fork_slots(
                    config.cluster_info.clone(),
                    &last_voted_fork_slots,
                    last_vote_bankhash,
                )?);
                WenRestartProgressInternalState::Init {
                    last_voted_fork_slots,
                    last_vote_bankhash,
                }
            }
            WenRestartProgressInternalState::LastVotedForkSlots {
                last_voted_fork_slots,
                aggregate_final_result,
            } => {
                let final_result = match aggregate_final_result {
                    Some(result) => result,
                    None => aggregate_restart_last_voted_fork_slots(
                        &config.wen_restart_path,
                        config.wait_for_supermajority_threshold_percent,
                        config.cluster_info.clone(),
                        &last_voted_fork_slots,
                        config.bank_forks.clone(),
                        config.blockstore.clone(),
                        config.wen_restart_repair_slots.clone().unwrap(),
                        config.exit.clone(),
                        &mut progress,
                    )?,
                };
                WenRestartProgressInternalState::LastVotedForkSlots {
                    last_voted_fork_slots,
                    aggregate_final_result: Some(final_result),
                }
            }
            WenRestartProgressInternalState::FindHeaviestFork {
                aggregate_final_result,
                my_heaviest_fork,
            } => {
                let heaviest_fork = match my_heaviest_fork {
                    Some(heaviest_fork) => heaviest_fork,
                    None => {
                        let (slot, bankhash) = find_heaviest_fork(
                            aggregate_final_result.clone(),
                            config.bank_forks.clone(),
                            config.blockstore.clone(),
                            config.exit.clone(),
                        )?;
                        info!("Heaviest fork found: slot: {slot}, bankhash: {bankhash}");
                        HeaviestForkRecord {
                            slot,
                            bankhash: bankhash.to_string(),
                            total_active_stake: 0,
                            wallclock: 0,
                            shred_version: config.cluster_info.my_shred_version() as u32,
                            from: config.cluster_info.id().to_string(),
                        }
                    }
                };
                WenRestartProgressInternalState::FindHeaviestFork {
                    aggregate_final_result,
                    my_heaviest_fork: Some(heaviest_fork),
                }
            }
            WenRestartProgressInternalState::HeaviestFork {
                my_heaviest_fork_slot,
                my_heaviest_fork_hash,
            } => {
                let (slot, hash) = send_and_receive_heaviest_fork(
                    my_heaviest_fork_slot,
                    my_heaviest_fork_hash,
                    &config,
                    &mut progress,
                    |slot, hash| {
                        config
                            .cluster_info
                            .push_restart_heaviest_fork(slot, hash, 0);
                    },
                )?;
                WenRestartProgressInternalState::HeaviestFork {
                    my_heaviest_fork_slot: slot,
                    my_heaviest_fork_hash: hash,
                }
            }
            WenRestartProgressInternalState::GenerateSnapshot {
                my_heaviest_fork_slot,
                my_snapshot,
            } => {
                let snapshot_record = match my_snapshot {
                    Some(record) => record,
                    None => match &config.snapshot_controller {
                        Some(snapshot_controller) => generate_snapshot(
                            config.bank_forks.clone(),
                            snapshot_controller,
                            &config.abs_status,
                            config.genesis_config_hash,
                            my_heaviest_fork_slot,
                        ),
                        None => {
                            // Only tests don't have a snapshot controller
                            Err(WenRestartError::GenerateSnapshotWhenDisabled.into())
                        }
                    }?,
                };
                WenRestartProgressInternalState::GenerateSnapshot {
                    my_heaviest_fork_slot,
                    my_snapshot: Some(snapshot_record),
                }
            }
            // Proceed to restart if we are ready to wait for supermajority.
            WenRestartProgressInternalState::Done {
                slot,
                hash,
                shred_version,
            } => {
                error!(
                    "Wen start finished, please remove --wen_restart and restart with \
                     --wait-for-supermajority {slot} --expected-bank-hash {hash} \
                     --expected-shred-version {shred_version} --no-snapshot-fetch",
                );
                if config.cluster_info.id() == config.wen_restart_coordinator {
                    aggregate_restart_heaviest_fork(
                        &config.wen_restart_path,
                        config.cluster_info.clone(),
                        config.bank_forks.clone(),
                        config.exit.clone(),
                        &mut progress,
                    )?;
                }
                return Ok(());
            }
        };
        state = increment_and_write_wen_restart_records(
            &config.wen_restart_path,
            state,
            &mut progress,
        )?;
    }
}

pub(crate) fn increment_and_write_wen_restart_records(
    records_path: &PathBuf,
    current_state: WenRestartProgressInternalState,
    progress: &mut WenRestartProgress,
) -> Result<WenRestartProgressInternalState> {
    let new_state = match current_state {
        WenRestartProgressInternalState::Init {
            last_voted_fork_slots,
            last_vote_bankhash: _,
        } => {
            progress.set_state(RestartState::LastVotedForkSlots);
            WenRestartProgressInternalState::LastVotedForkSlots {
                last_voted_fork_slots,
                aggregate_final_result: None,
            }
        }
        WenRestartProgressInternalState::LastVotedForkSlots {
            last_voted_fork_slots: _,
            aggregate_final_result,
        } => {
            if let Some(aggregate_final_result) = aggregate_final_result {
                progress.set_state(RestartState::HeaviestFork);
                if let Some(aggregate_record) = progress.last_voted_fork_slots_aggregate.as_mut() {
                    aggregate_record.final_result = Some(LastVotedForkSlotsAggregateFinal {
                        slots_stake_map: aggregate_final_result.slots_stake_map.clone(),
                        epoch_infos: aggregate_final_result
                            .epoch_info_vec
                            .iter()
                            .map(|info| LastVotedForkSlotsEpochInfoRecord {
                                epoch: info.epoch,
                                total_stake: info.total_stake,
                                actively_voting_stake: info.actively_voting_stake,
                                actively_voting_for_this_epoch_stake: info
                                    .actively_voting_for_this_epoch_stake,
                            })
                            .collect(),
                    });
                }
                WenRestartProgressInternalState::FindHeaviestFork {
                    aggregate_final_result,
                    my_heaviest_fork: None,
                }
            } else {
                return Err(
                    WenRestartError::UnexpectedState(RestartState::LastVotedForkSlots).into(),
                );
            }
        }
        WenRestartProgressInternalState::FindHeaviestFork {
            aggregate_final_result: _,
            my_heaviest_fork,
        } => {
            if let Some(my_heaviest_fork) = my_heaviest_fork {
                progress.my_heaviest_fork = Some(my_heaviest_fork.clone());
                WenRestartProgressInternalState::HeaviestFork {
                    my_heaviest_fork_slot: my_heaviest_fork.slot,
                    my_heaviest_fork_hash: Hash::from_str(&my_heaviest_fork.bankhash).unwrap(),
                }
            } else {
                return Err(WenRestartError::UnexpectedState(RestartState::HeaviestFork).into());
            }
        }
        WenRestartProgressInternalState::HeaviestFork {
            my_heaviest_fork_slot,
            ..
        } => {
            progress.set_state(RestartState::GenerateSnapshot);
            WenRestartProgressInternalState::GenerateSnapshot {
                my_heaviest_fork_slot,
                my_snapshot: None,
            }
        }
        WenRestartProgressInternalState::GenerateSnapshot {
            my_heaviest_fork_slot: _,
            my_snapshot,
        } => {
            if let Some(my_snapshot) = my_snapshot {
                progress.set_state(RestartState::Done);
                progress.my_snapshot = Some(my_snapshot.clone());
                WenRestartProgressInternalState::Done {
                    slot: my_snapshot.slot,
                    hash: Hash::from_str(&my_snapshot.bankhash).unwrap(),
                    shred_version: my_snapshot.shred_version as u16,
                }
            } else {
                return Err(WenRestartError::MissingSnapshotInProtobuf.into());
            }
        }
        WenRestartProgressInternalState::Done { .. } => {
            return Err(WenRestartError::UnexpectedState(RestartState::Done).into())
        }
    };
    write_wen_restart_records(records_path, progress)?;
    Ok(new_state)
}

pub(crate) fn initialize(
    records_path: &PathBuf,
    last_vote: VoteTransaction,
    blockstore: Arc<Blockstore>,
) -> Result<(WenRestartProgressInternalState, WenRestartProgress)> {
    let progress = match read_wen_restart_records(records_path) {
        Ok(progress) => progress,
        Err(e) => {
            let stdio_err = e.downcast_ref::<std::io::Error>();
            if stdio_err.is_some_and(|e| e.kind() == std::io::ErrorKind::NotFound) {
                info!("wen restart proto file not found at {records_path:?}, write init state");
                let progress = WenRestartProgress {
                    state: RestartState::Init.into(),
                    ..Default::default()
                };
                write_wen_restart_records(records_path, &progress)?;
                progress
            } else {
                return Err(e);
            }
        }
    };
    match progress.state() {
        RestartState::Done => {
            if let Some(my_snapshot) = progress.my_snapshot.as_ref() {
                Ok((
                    WenRestartProgressInternalState::Done {
                        slot: my_snapshot.slot,
                        hash: Hash::from_str(&my_snapshot.bankhash).unwrap(),
                        shred_version: my_snapshot.shred_version as u16,
                    },
                    progress,
                ))
            } else {
                Err(WenRestartError::MissingSnapshotInProtobuf.into())
            }
        }
        RestartState::Init => {
            let last_voted_fork_slots;
            let last_vote_bankhash;
            match &progress.my_last_voted_fork_slots {
                Some(my_last_voted_fork_slots) => {
                    last_voted_fork_slots = my_last_voted_fork_slots.last_voted_fork_slots.clone();
                    last_vote_bankhash =
                        Hash::from_str(&my_last_voted_fork_slots.last_vote_bankhash).unwrap();
                }
                None => {
                    // repair and restart option does not work without last voted slot.
                    if let Some(last_vote_slot) = last_vote.last_voted_slot() {
                        last_vote_bankhash = last_vote.hash();
                        last_voted_fork_slots =
                            AncestorIterator::new_inclusive(last_vote_slot, &blockstore)
                                .take(RestartLastVotedForkSlots::MAX_SLOTS)
                                .collect();
                    } else {
                        error!(
                            "Cannot find last voted slot in the tower storage, it either means \
                             that this node has never voted or the tower storage is corrupted. \
                             Unfortunately, since WenRestart is a consensus protocol depending on \
                             each participant to send their last voted fork slots, your validator \
                             cannot participate.Please check discord for the conclusion of the \
                             WenRestart protocol, then generate a snapshot and use \
                             --wait-for-supermajority to restart the validator."
                        );
                        return Err(WenRestartError::MissingLastVotedForkSlots.into());
                    }
                }
            }
            Ok((
                WenRestartProgressInternalState::Init {
                    last_voted_fork_slots,
                    last_vote_bankhash,
                },
                progress,
            ))
        }
        RestartState::LastVotedForkSlots => {
            if let Some(record) = progress.my_last_voted_fork_slots.as_ref() {
                Ok((
                    WenRestartProgressInternalState::LastVotedForkSlots {
                        last_voted_fork_slots: record.last_voted_fork_slots.clone(),
                        aggregate_final_result: progress
                            .last_voted_fork_slots_aggregate
                            .as_ref()
                            .and_then(|r| {
                                r.final_result.as_ref().map(|result| {
                                    LastVotedForkSlotsFinalResult {
                                        slots_stake_map: result.slots_stake_map.clone(),
                                        epoch_info_vec: result
                                            .epoch_infos
                                            .iter()
                                            .map(|info| LastVotedForkSlotsEpochInfo {
                                                epoch: info.epoch,
                                                total_stake: info.total_stake,
                                                actively_voting_stake: info.actively_voting_stake,
                                                actively_voting_for_this_epoch_stake: info
                                                    .actively_voting_for_this_epoch_stake,
                                            })
                                            .collect(),
                                    }
                                })
                            }),
                    },
                    progress,
                ))
            } else {
                Err(WenRestartError::MalformedLastVotedForkSlotsProtobuf(None).into())
            }
        }
        RestartState::HeaviestFork => Ok((
            WenRestartProgressInternalState::FindHeaviestFork {
                aggregate_final_result: progress
                    .last_voted_fork_slots_aggregate
                    .as_ref()
                    .and_then(|r| {
                        r.final_result
                            .as_ref()
                            .map(|result| LastVotedForkSlotsFinalResult {
                                slots_stake_map: result.slots_stake_map.clone(),
                                epoch_info_vec: result
                                    .epoch_infos
                                    .iter()
                                    .map(|info| LastVotedForkSlotsEpochInfo {
                                        epoch: info.epoch,
                                        total_stake: info.total_stake,
                                        actively_voting_stake: info.actively_voting_stake,
                                        actively_voting_for_this_epoch_stake: info
                                            .actively_voting_for_this_epoch_stake,
                                    })
                                    .collect(),
                            })
                    })
                    .ok_or(WenRestartError::MalformedProgress(
                        RestartState::HeaviestFork,
                        "final_result in last_voted_fork_slots_aggregate".to_string(),
                    ))?,
                my_heaviest_fork: progress.my_heaviest_fork.clone(),
            },
            progress,
        )),
        RestartState::GenerateSnapshot => Ok((
            WenRestartProgressInternalState::GenerateSnapshot {
                my_heaviest_fork_slot: progress
                    .my_heaviest_fork
                    .as_ref()
                    .ok_or(WenRestartError::MalformedProgress(
                        RestartState::GenerateSnapshot,
                        "my_heaviest_fork".to_string(),
                    ))?
                    .slot,
                my_snapshot: progress.my_snapshot.clone(),
            },
            progress,
        )),
    }
}

fn read_wen_restart_records(records_path: &PathBuf) -> Result<WenRestartProgress> {
    let buffer = read(records_path)?;
    let progress = WenRestartProgress::decode(&mut Cursor::new(buffer))?;
    info!("read record {progress:?}");
    Ok(progress)
}

pub(crate) fn write_wen_restart_records(
    records_path: &PathBuf,
    new_progress: &WenRestartProgress,
) -> Result<()> {
    // overwrite anything if exists
    let mut file = File::create(records_path)?;
    info!("writing new record {new_progress:?}");
    let mut buf = Vec::with_capacity(new_progress.encoded_len());
    new_progress.encode(&mut buf)?;
    file.write_all(&buf)?;
    Ok(())
}

#[cfg(test)]
mod tests {
    use {
        crate::wen_restart::{tests::wen_restart_proto::LastVotedForkSlotsAggregateFinal, *},
        agave_snapshots::{
            paths::build_incremental_snapshot_archive_path,
            snapshot_config::{SnapshotConfig, SnapshotUsage},
            snapshot_hash::SnapshotHash,
        },
        crossbeam_channel::unbounded,
        solana_entry::entry::create_ticks,
        solana_genesis_utils::MAX_GENESIS_ARCHIVE_UNPACKED_SIZE,
        solana_gossip::{
            cluster_info::ClusterInfo,
            contact_info::ContactInfo,
            crds::GossipRoute,
            crds_data::CrdsData,
            crds_value::CrdsValue,
            restart_crds_values::{RestartHeaviestFork, RestartLastVotedForkSlots},
        },
        solana_hash::Hash,
        solana_keypair::Keypair,
        solana_ledger::{
            blockstore::{create_new_ledger, entries_to_test_shreds, Blockstore},
            blockstore_options::LedgerColumnOptions,
            blockstore_processor::{fill_blockstore_slot_with_ticks, test_process_blockstore},
            get_tmp_ledger_path_auto_delete,
        },
        solana_net_utils::SocketAddrSpace,
        solana_pubkey::Pubkey,
        solana_runtime::{
            epoch_stakes::VersionedEpochStakes,
            genesis_utils::{
                create_genesis_config_with_vote_accounts, GenesisConfigInfo, ValidatorVoteKeypairs,
            },
            snapshot_bank_utils::bank_to_full_snapshot_archive,
        },
        solana_signer::Signer,
        solana_time_utils::timestamp,
        solana_vote::vote_account::VoteAccount,
        solana_vote_interface::state::{TowerSync, Vote},
        solana_vote_program::vote_state::create_v4_account_with_authorized,
        std::{fs::remove_file, sync::Arc, thread::Builder},
        tempfile::TempDir,
    };

    const SHRED_VERSION: u16 = 2;
    const EXPECTED_SLOTS: Slot = 40;
    const TICKS_PER_SLOT: u64 = 2;
    const TOTAL_VALIDATOR_COUNT: u16 = 20;
    const MY_INDEX: usize = TOTAL_VALIDATOR_COUNT as usize - 1;
    const COORDINATOR_INDEX: usize = 0;
    const WAIT_FOR_THREAD_TIMEOUT: u64 = 10_000;
    const WAIT_FOR_SUPERMAJORITY_THRESHOLD_PERCENT: u64 = 80;
    const NON_CONFORMING_VALIDATOR_PERCENT: u64 = 5;

    fn push_restart_last_voted_fork_slots(
        cluster_info: Arc<ClusterInfo>,
        node: &ContactInfo,
        last_voted_fork_slots: &[Slot],
        last_vote_hash: &Hash,
        node_keypair: &Keypair,
        wallclock: u64,
    ) {
        let slots = RestartLastVotedForkSlots::new(
            *node.pubkey(),
            wallclock,
            last_voted_fork_slots,
            *last_vote_hash,
            SHRED_VERSION,
        )
        .unwrap();
        let entries = vec![
            CrdsValue::new(CrdsData::from(node), node_keypair),
            CrdsValue::new(CrdsData::RestartLastVotedForkSlots(slots), node_keypair),
        ];
        {
            let mut gossip_crds = cluster_info.gossip.crds.write().unwrap();
            for entry in entries {
                assert!(gossip_crds
                    .insert(entry, /*now=*/ 0, GossipRoute::LocalMessage)
                    .is_ok());
            }
        }
    }

    fn push_restart_heaviest_fork(
        cluster_info: Arc<ClusterInfo>,
        node: &ContactInfo,
        heaviest_fork_slot: Slot,
        heaviest_fork_hash: &Hash,
        observed_stake: u64,
        node_keypair: &Keypair,
        wallclock: u64,
    ) {
        let heaviest_fork = RestartHeaviestFork {
            from: *node.pubkey(),
            wallclock,
            last_slot: heaviest_fork_slot,
            last_slot_hash: *heaviest_fork_hash,
            observed_stake,
            shred_version: SHRED_VERSION,
        };
        assert!(cluster_info
            .gossip
            .crds
            .write()
            .unwrap()
            .insert(
                CrdsValue::new(CrdsData::RestartHeaviestFork(heaviest_fork), node_keypair),
                /*now=*/ 0,
                GossipRoute::LocalMessage
            )
            .is_ok());
    }

    struct WenRestartTestInitResult {
        pub validator_voting_keypairs: Vec<ValidatorVoteKeypairs>,
        pub blockstore: Arc<Blockstore>,
        pub cluster_info: Arc<ClusterInfo>,
        pub bank_forks: Arc<RwLock<BankForks>>,
        pub last_voted_fork_slots: Vec<Slot>,
        pub wen_restart_proto_path: PathBuf,
        pub wen_restart_coordinator: Pubkey,
        pub last_blockhash: Hash,
        pub genesis_config_hash: Hash,
    }

    fn insert_slots_into_blockstore(
        blockstore: Arc<Blockstore>,
        first_parent: Slot,
        slots_to_insert: &[Slot],
        entries_per_slot: u64,
        start_blockhash: Hash,
    ) -> Hash {
        let mut last_hash = start_blockhash;
        let mut last_parent = first_parent;
        for i in slots_to_insert {
            last_hash = fill_blockstore_slot_with_ticks(
                &blockstore,
                entries_per_slot,
                *i,
                last_parent,
                last_hash,
            );
            last_parent = *i;
        }
        last_hash
    }

    fn wen_restart_test_init(ledger_path: &TempDir) -> WenRestartTestInitResult {
        let validator_voting_keypairs: Vec<_> = (0..TOTAL_VALIDATOR_COUNT)
            .map(|_| ValidatorVoteKeypairs::new_rand())
            .collect();
        let node_keypair = Arc::new(
            validator_voting_keypairs[MY_INDEX]
                .node_keypair
                .insecure_clone(),
        );
        let wen_restart_coordinator = validator_voting_keypairs[COORDINATOR_INDEX]
            .node_keypair
            .pubkey();
        let cluster_info = Arc::new(ClusterInfo::new(
            {
                let mut contact_info =
                    ContactInfo::new_localhost(&node_keypair.pubkey(), timestamp());
                contact_info.set_shred_version(SHRED_VERSION);
                contact_info
            },
            node_keypair.clone(),
            SocketAddrSpace::Unspecified,
        ));
        let GenesisConfigInfo {
            mut genesis_config, ..
        } = create_genesis_config_with_vote_accounts(
            10_000,
            &validator_voting_keypairs,
            vec![100; validator_voting_keypairs.len()],
        );
        genesis_config.ticks_per_slot = TICKS_PER_SLOT;
        let start_blockhash = create_new_ledger(
            ledger_path.path(),
            &genesis_config,
            MAX_GENESIS_ARCHIVE_UNPACKED_SIZE,
            LedgerColumnOptions::default(),
        )
        .unwrap();
        let blockstore = Arc::new(Blockstore::open(ledger_path.path()).unwrap());
        let (bank_forks, ..) = test_process_blockstore(
            &genesis_config,
            &blockstore,
            &ProcessOptions {
                run_verification: true,
                ..ProcessOptions::default()
            },
            Arc::default(),
        );
        let mut last_blockhash = start_blockhash;
        // Skip block 1, 2 links directly to 0.
        let last_parent: Slot = 2;
        let mut last_voted_fork_slots: Vec<Slot> = Vec::new();
        last_voted_fork_slots
            .extend(last_parent..last_parent.saturating_add(EXPECTED_SLOTS).saturating_add(1));
        last_blockhash = insert_slots_into_blockstore(
            blockstore.clone(),
            0,
            &last_voted_fork_slots,
            genesis_config.ticks_per_slot,
            last_blockhash,
        );
        last_voted_fork_slots.insert(0, 0);
        last_voted_fork_slots.reverse();
        let mut wen_restart_proto_path = ledger_path.path().to_path_buf();
        wen_restart_proto_path.push("wen_restart_status.proto");
        let _ = remove_file(&wen_restart_proto_path);
        WenRestartTestInitResult {
            validator_voting_keypairs,
            blockstore,
            cluster_info,
            bank_forks,
            last_voted_fork_slots,
            wen_restart_proto_path,
            wen_restart_coordinator,
            last_blockhash,
            genesis_config_hash: genesis_config.hash(),
        }
    }

    fn wait_on_expected_progress_with_timeout(
        wen_restart_proto_path: PathBuf,
        expected_progress: WenRestartProgress,
    ) {
        let start = timestamp();
        let mut progress = WenRestartProgress {
            state: RestartState::Init.into(),
            ..Default::default()
        };
        loop {
            if let Ok(new_progress) = read_wen_restart_records(&wen_restart_proto_path) {
                progress = new_progress;
                if let Some(my_last_voted_fork_slots) = &expected_progress.my_last_voted_fork_slots
                {
                    if let Some(record) = progress.my_last_voted_fork_slots.as_mut() {
                        record.wallclock = my_last_voted_fork_slots.wallclock;
                    }
                }
                if progress == expected_progress {
                    return;
                }
            }
            if timestamp().saturating_sub(start) > WAIT_FOR_THREAD_TIMEOUT {
                assert_eq!(
                    progress.my_last_voted_fork_slots,
                    expected_progress.my_last_voted_fork_slots
                );
                assert_eq!(
                    progress.last_voted_fork_slots_aggregate,
                    expected_progress.last_voted_fork_slots_aggregate
                );
                panic!(
                    "wait_on_expected_progress_with_timeout failed to get expected progress {:?} \
                     expected {:?}",
                    &progress, expected_progress
                );
            }
            sleep(Duration::from_millis(10));
        }
    }

    fn wen_restart_test_succeed_after_failure(
        test_state: WenRestartTestInitResult,
        last_vote_bankhash: Hash,
        expected_progress: WenRestartProgress,
    ) {
        // continue normally after the error, we should be good.
        let exit = Arc::new(AtomicBool::new(false));
        let last_vote_slot: Slot = test_state.last_voted_fork_slots[0];
        let wen_restart_config = WenRestartConfig {
            wen_restart_path: test_state.wen_restart_proto_path.clone(),
            wen_restart_coordinator: test_state.wen_restart_coordinator,
            last_vote: VoteTransaction::from(Vote::new(vec![last_vote_slot], last_vote_bankhash)),
            blockstore: test_state.blockstore.clone(),
            cluster_info: test_state.cluster_info.clone(),
            bank_forks: test_state.bank_forks.clone(),
            wen_restart_repair_slots: Some(Arc::new(RwLock::new(Vec::new()))),
            wait_for_supermajority_threshold_percent: 80,
            snapshot_controller: None,
            abs_status: AbsStatus::new_for_tests(),
            genesis_config_hash: test_state.genesis_config_hash,
            exit: exit.clone(),
        };
        let wen_restart_thread_handle = Builder::new()
            .name("solana-wen-restart".to_string())
            .spawn(move || {
                let _ = wait_for_wen_restart(wen_restart_config).is_ok();
            })
            .unwrap();
        wait_on_expected_progress_with_timeout(
            test_state.wen_restart_proto_path.clone(),
            expected_progress,
        );
        exit.store(true, Ordering::Relaxed);
        assert!(wen_restart_thread_handle.join().is_ok());
        let _ = remove_file(&test_state.wen_restart_proto_path);
    }

    #[test]
    fn test_wen_restart_normal_flow() {
        let ledger_path = get_tmp_ledger_path_auto_delete!();
        let wen_restart_repair_slots = Some(Arc::new(RwLock::new(Vec::new())));
        let test_state = wen_restart_test_init(&ledger_path);
        let last_vote_slot = test_state.last_voted_fork_slots[0];
        let last_vote_bankhash = Hash::new_unique();
        let expected_slots_to_repair: Vec<Slot> =
            (last_vote_slot + 1..last_vote_slot + 3).collect();
        let my_pubkey = &test_state.validator_voting_keypairs[MY_INDEX]
            .node_keypair
            .pubkey();

        let bank_snapshots_dir = tempfile::TempDir::new().unwrap();
        let full_snapshot_archives_dir = tempfile::TempDir::new().unwrap();
        let incremental_snapshot_archives_dir = tempfile::TempDir::new().unwrap();
        let snapshot_config = SnapshotConfig {
            bank_snapshots_dir: bank_snapshots_dir.as_ref().to_path_buf(),
            full_snapshot_archives_dir: full_snapshot_archives_dir.as_ref().to_path_buf(),
            incremental_snapshot_archives_dir: incremental_snapshot_archives_dir
                .as_ref()
                .to_path_buf(),
            ..Default::default()
        };
        let old_root_bank = test_state.bank_forks.read().unwrap().root_bank();
        // Trigger full snapshot generation on the old root bank.
        assert!(bank_to_full_snapshot_archive(
            snapshot_config.bank_snapshots_dir.clone(),
            &old_root_bank,
            Some(snapshot_config.snapshot_version),
            snapshot_config.full_snapshot_archives_dir.clone(),
            snapshot_config.incremental_snapshot_archives_dir.clone(),
            snapshot_config.archive_format,
        )
        .is_ok());

        let exit = Arc::new(AtomicBool::new(false));
        let (abs_request_sender, _abs_request_receiver) = unbounded();
        let snapshot_controller =
            SnapshotController::new(abs_request_sender, snapshot_config, last_vote_slot);
        let wen_restart_config = WenRestartConfig {
            wen_restart_path: test_state.wen_restart_proto_path.clone(),
            wen_restart_coordinator: test_state.wen_restart_coordinator,
            last_vote: VoteTransaction::from(Vote::new(vec![last_vote_slot], last_vote_bankhash)),
            blockstore: test_state.blockstore.clone(),
            cluster_info: test_state.cluster_info.clone(),
            bank_forks: test_state.bank_forks.clone(),
            wen_restart_repair_slots: wen_restart_repair_slots.clone(),
            wait_for_supermajority_threshold_percent: 80,
            snapshot_controller: Some(Arc::new(snapshot_controller)),
            abs_status: AbsStatus::new_for_tests(),
            genesis_config_hash: test_state.genesis_config_hash,
            exit: exit.clone(),
        };
        let wen_restart_thread_handle = Builder::new()
            .name("solana-wen-restart".to_string())
            .spawn(move || {
                assert!(wait_for_wen_restart(wen_restart_config).is_ok());
            })
            .unwrap();
        let mut rng = rand::thread_rng();
        let mut expected_received_last_voted_fork_slots = HashMap::new();
        let validators_to_take: usize =
            (WAIT_FOR_SUPERMAJORITY_THRESHOLD_PERCENT * TOTAL_VALIDATOR_COUNT as u64 / 100 - 1)
                .try_into()
                .unwrap();
        let mut last_voted_fork_slots_from_others = test_state.last_voted_fork_slots.clone();
        last_voted_fork_slots_from_others.reverse();
        last_voted_fork_slots_from_others.append(&mut expected_slots_to_repair.clone());
        for keypairs in test_state
            .validator_voting_keypairs
            .iter()
            .take(validators_to_take)
        {
            let node_pubkey = keypairs.node_keypair.pubkey();
            let node = ContactInfo::new_rand(&mut rng, Some(node_pubkey));
            let last_vote_hash = Hash::new_unique();
            let now = timestamp();
            push_restart_last_voted_fork_slots(
                test_state.cluster_info.clone(),
                &node,
                &last_voted_fork_slots_from_others,
                &last_vote_hash,
                &keypairs.node_keypair,
                now,
            );
            expected_received_last_voted_fork_slots.insert(
                node_pubkey.to_string(),
                LastVotedForkSlotsRecord {
                    last_voted_fork_slots: last_voted_fork_slots_from_others.clone(),
                    last_vote_bankhash: last_vote_hash.to_string(),
                    shred_version: SHRED_VERSION as u32,
                    wallclock: now,
                },
            );
        }

        // Simulating successful repair of missing blocks.
        let _ = insert_slots_into_blockstore(
            test_state.blockstore.clone(),
            last_vote_slot,
            &expected_slots_to_repair,
            TICKS_PER_SLOT,
            test_state.last_blockhash,
        );

        let my_heaviest_fork_slot = last_vote_slot + 2;
        let my_heaviest_fork_bankhash;
        loop {
            if let Some(bank) = test_state
                .bank_forks
                .read()
                .unwrap()
                .get(my_heaviest_fork_slot)
            {
                // When deciding the local heaviest fork, we will freeze the bank.
                if bank.is_frozen() {
                    my_heaviest_fork_bankhash = bank.hash();
                    break;
                }
            }
            sleep(Duration::from_millis(100));
        }
        // Now simulate receiving HeaviestFork messages from coordinator.
        let coordinator_heaviest_fork_slot = my_heaviest_fork_slot - 1;
        let coordinator_heaviest_fork_bankhash = test_state
            .bank_forks
            .read()
            .unwrap()
            .get(coordinator_heaviest_fork_slot)
            .unwrap()
            .hash();
        let coordinator_keypair =
            &test_state.validator_voting_keypairs[COORDINATOR_INDEX].node_keypair;
        let node = ContactInfo::new_rand(&mut rng, Some(coordinator_keypair.pubkey()));
        let now = timestamp();
        push_restart_heaviest_fork(
            test_state.cluster_info.clone(),
            &node,
            coordinator_heaviest_fork_slot,
            &coordinator_heaviest_fork_bankhash,
            0,
            coordinator_keypair,
            now,
        );

        assert!(wen_restart_thread_handle.join().is_ok());
        exit.store(true, Ordering::Relaxed);
        let progress = read_wen_restart_records(&test_state.wen_restart_proto_path).unwrap();
        let progress_start_time = progress
            .my_last_voted_fork_slots
            .as_ref()
            .unwrap()
            .wallclock;
        let mut expected_slots_stake_map: HashMap<Slot, u64> = test_state
            .last_voted_fork_slots
            .iter()
            .map(|slot| {
                (
                    *slot,
                    WAIT_FOR_SUPERMAJORITY_THRESHOLD_PERCENT * TOTAL_VALIDATOR_COUNT as u64,
                )
            })
            .collect();
        let stake_for_new_slots = validators_to_take as u64 * 100;
        expected_slots_stake_map.extend(
            expected_slots_to_repair
                .iter()
                .map(|slot| (*slot, stake_for_new_slots)),
        );
        let voted_stake = (validators_to_take + 1) as u64 * 100;
        assert_eq!(
            progress,
            WenRestartProgress {
                state: RestartState::Done.into(),
                my_last_voted_fork_slots: Some(LastVotedForkSlotsRecord {
                    last_voted_fork_slots: test_state.last_voted_fork_slots,
                    last_vote_bankhash: last_vote_bankhash.to_string(),
                    shred_version: SHRED_VERSION as u32,
                    wallclock: progress_start_time,
                }),
                last_voted_fork_slots_aggregate: Some(LastVotedForkSlotsAggregateRecord {
                    received: expected_received_last_voted_fork_slots,
                    final_result: Some(LastVotedForkSlotsAggregateFinal {
                        slots_stake_map: expected_slots_stake_map,
                        epoch_infos: vec![
                            LastVotedForkSlotsEpochInfoRecord {
                                epoch: 0,
                                total_stake: 2000,
                                actively_voting_stake: voted_stake,
                                actively_voting_for_this_epoch_stake: voted_stake,
                            },
                            LastVotedForkSlotsEpochInfoRecord {
                                epoch: 1,
                                total_stake: 2000,
                                actively_voting_stake: voted_stake,
                                actively_voting_for_this_epoch_stake: voted_stake,
                            },
                        ],
                    }),
                }),
                my_heaviest_fork: Some(HeaviestForkRecord {
                    slot: my_heaviest_fork_slot,
                    bankhash: my_heaviest_fork_bankhash.to_string(),
                    total_active_stake: 0,
                    shred_version: SHRED_VERSION as u32,
                    wallclock: 0,
                    from: my_pubkey.to_string(),
                }),
                heaviest_fork_aggregate: None,
                my_snapshot: Some(GenerateSnapshotRecord {
                    slot: coordinator_heaviest_fork_slot,
                    bankhash: progress.my_snapshot.as_ref().unwrap().bankhash.clone(),
                    shred_version: progress.my_snapshot.as_ref().unwrap().shred_version,
                    path: progress.my_snapshot.as_ref().unwrap().path.clone(),
                }),
                coordinator_heaviest_fork: Some(HeaviestForkRecord {
                    slot: coordinator_heaviest_fork_slot,
                    bankhash: coordinator_heaviest_fork_bankhash.to_string(),
                    total_active_stake: 0,
                    shred_version: SHRED_VERSION as u32,
                    wallclock: progress
                        .coordinator_heaviest_fork
                        .as_ref()
                        .unwrap()
                        .wallclock,
                    from: coordinator_keypair.pubkey().to_string(),
                }),
                ..Default::default()
            }
        );
    }

    fn change_proto_file_readonly(wen_restart_proto_path: &PathBuf, readonly: bool) {
        let mut perms = std::fs::metadata(wen_restart_proto_path)
            .unwrap()
            .permissions();
        perms.set_readonly(readonly);
        std::fs::set_permissions(wen_restart_proto_path, perms).unwrap();
    }

    #[test]
    fn test_wen_restart_divergence_across_epoch_boundary() {
        agave_logger::setup();
        let ledger_path = get_tmp_ledger_path_auto_delete!();
        let test_state = wen_restart_test_init(&ledger_path);
        let last_vote_slot = test_state.last_voted_fork_slots[0];

        let old_root_bank = test_state.bank_forks.read().unwrap().root_bank();

        // Add bank last_vote + 1 linking directly to 0, tweak its epoch_stakes, and then add it to bank_forks.
        let my_heaviest_fork_slot = last_vote_slot + 1;
        let mut new_root_bank = Bank::new_from_parent(
            old_root_bank.clone(),
            &Pubkey::default(),
            my_heaviest_fork_slot,
        );
        assert_eq!(new_root_bank.epoch(), 1);

        // For epoch 2, make validator 0 have 90% of the stake.
        let vote_accounts_hash_map = test_state
            .validator_voting_keypairs
            .iter()
            .enumerate()
            .map(|(i, keypairs)| {
                let stake = if i == 0 {
                    900 * (TOTAL_VALIDATOR_COUNT - 1) as u64
                } else {
                    100
                };
                let authorized_voter = keypairs.vote_keypair.pubkey();
                let node_id = keypairs.node_keypair.pubkey();
                (
                    authorized_voter,
                    (
                        stake,
                        VoteAccount::try_from(create_v4_account_with_authorized(
                            &node_id,
                            &authorized_voter,
                            &node_id,
                            None,
                            0,
                            100,
                        ))
                        .unwrap(),
                    ),
                )
            })
            .collect();
        let epoch2_epoch_stakes = VersionedEpochStakes::new_for_tests(vote_accounts_hash_map, 2);
        new_root_bank.set_epoch_stakes_for_test(2, epoch2_epoch_stakes);
        let _ = insert_slots_into_blockstore(
            test_state.blockstore.clone(),
            0,
            &[my_heaviest_fork_slot],
            TICKS_PER_SLOT,
            old_root_bank.last_blockhash(),
        );
        let replay_tx_thread_pool = rayon::ThreadPoolBuilder::new()
            .thread_name(|i| format!("solReplayTx{i:02}"))
            .build()
            .expect("new rayon threadpool");
        let mut timing = ExecuteTimings::default();
        let opts = ProcessOptions::default();
        let mut progress = ConfirmationProgress::new(old_root_bank.last_blockhash());
        let last_vote_bankhash = new_root_bank.hash();
        let bank_with_scheduler = test_state
            .bank_forks
            .write()
            .unwrap()
            .insert_from_ledger(new_root_bank);
        if let Err(e) = process_single_slot(
            &test_state.blockstore,
            &bank_with_scheduler,
            &replay_tx_thread_pool,
            &opts,
            &mut progress,
            None,
            None,
            None,
            &mut timing,
        ) {
            panic!("process_single_slot failed: {e:?}");
        }

        {
            let mut bank_forks = test_state.bank_forks.write().unwrap();
            let _ = bank_forks.set_root(last_vote_slot + 1, None, Some(last_vote_slot + 1));
        }
        let new_root_bank = test_state
            .bank_forks
            .read()
            .unwrap()
            .get(last_vote_slot + 1)
            .unwrap();

        // Add two more banks: old_epoch_bank (slot = last_vote_slot + 2) and
        // new_epoch_bank (slot = first slot in epoch 2). They both link to last_vote_slot + 1.
        // old_epoch_bank has everyone's votes except 0, so it has > 66% stake in the old epoch.
        // new_epoch_bank has 0's vote, so it has > 66% stake in the new epoch.
        let old_epoch_slot = my_heaviest_fork_slot + 1;
        let _ = insert_slots_into_blockstore(
            test_state.blockstore.clone(),
            new_root_bank.slot(),
            &[old_epoch_slot],
            TICKS_PER_SLOT,
            new_root_bank.last_blockhash(),
        );
        let new_epoch_slot = new_root_bank.epoch_schedule().get_first_slot_in_epoch(2);
        let _ = insert_slots_into_blockstore(
            test_state.blockstore.clone(),
            my_heaviest_fork_slot,
            &[new_epoch_slot],
            TICKS_PER_SLOT,
            new_root_bank.last_blockhash(),
        );
        let mut rng = rand::thread_rng();
        // Everyone except 0 votes for old_epoch_bank.
        for (index, keypairs) in test_state
            .validator_voting_keypairs
            .iter()
            .take(TOTAL_VALIDATOR_COUNT as usize - 1)
            .enumerate()
        {
            let node_pubkey = keypairs.node_keypair.pubkey();
            let node = ContactInfo::new_rand(&mut rng, Some(node_pubkey));
            let last_vote_hash = Hash::new_unique();
            let now = timestamp();
            // Validator 0 votes for the new_epoch_bank while everyone elese vote for old_epoch_bank.
            let last_voted_fork_slots = if index == 0 {
                vec![new_epoch_slot, my_heaviest_fork_slot, 0]
            } else {
                vec![old_epoch_slot, my_heaviest_fork_slot, 0]
            };
            push_restart_last_voted_fork_slots(
                test_state.cluster_info.clone(),
                &node,
                &last_voted_fork_slots,
                &last_vote_hash,
                &keypairs.node_keypair,
                now,
            );
        }

        assert_eq!(
            wait_for_wen_restart(WenRestartConfig {
                wen_restart_path: test_state.wen_restart_proto_path,
                wen_restart_coordinator: test_state.wen_restart_coordinator,
                last_vote: VoteTransaction::from(Vote::new(
                    vec![my_heaviest_fork_slot],
                    last_vote_bankhash
                )),
                blockstore: test_state.blockstore,
                cluster_info: test_state.cluster_info,
                bank_forks: test_state.bank_forks,
                wen_restart_repair_slots: Some(Arc::new(RwLock::new(Vec::new()))),
                wait_for_supermajority_threshold_percent: 80,
                snapshot_controller: None,
                abs_status: AbsStatus::new_for_tests(),
                genesis_config_hash: test_state.genesis_config_hash,
                exit: Arc::new(AtomicBool::new(false)),
            })
            .unwrap_err()
            .downcast::<WenRestartError>()
            .unwrap(),
            WenRestartError::BlockNotLinkedToExpectedParent(
                new_epoch_slot,
                Some(my_heaviest_fork_slot),
                old_epoch_slot
            )
        );
    }

    #[test]
    fn test_wen_restart_initialize() {
        agave_logger::setup();
        let ledger_path = get_tmp_ledger_path_auto_delete!();
        let test_state = wen_restart_test_init(&ledger_path);
        let last_vote_bankhash = Hash::new_unique();
        let mut last_voted_fork_slots = test_state.last_voted_fork_slots.clone();
        last_voted_fork_slots.reverse();
        let mut file = File::create(&test_state.wen_restart_proto_path).unwrap();
        file.write_all(b"garbage").unwrap();
        assert_eq!(
            initialize(
                &test_state.wen_restart_proto_path,
                VoteTransaction::from(Vote::new(last_voted_fork_slots.clone(), last_vote_bankhash)),
                test_state.blockstore.clone()
            )
            .unwrap_err()
            .downcast::<prost::DecodeError>()
            .unwrap(),
            prost::DecodeError::new("invalid wire type value: 7")
        );
        assert!(remove_file(&test_state.wen_restart_proto_path).is_ok());
        let last_vote_bankhash = Hash::new_unique();
        let empty_last_vote = VoteTransaction::from(Vote::new(vec![], last_vote_bankhash));
        assert_eq!(
            initialize(
                &test_state.wen_restart_proto_path,
                empty_last_vote.clone(),
                test_state.blockstore.clone()
            )
            .unwrap_err()
            .downcast::<WenRestartError>()
            .unwrap(),
            WenRestartError::MissingLastVotedForkSlots,
        );
        assert!(write_wen_restart_records(
            &test_state.wen_restart_proto_path,
            &WenRestartProgress {
                state: RestartState::LastVotedForkSlots.into(),
                ..Default::default()
            },
        )
        .is_ok());
        assert_eq!(
            initialize(
                &test_state.wen_restart_proto_path,
                VoteTransaction::from(Vote::new(last_voted_fork_slots.clone(), last_vote_bankhash)),
                test_state.blockstore.clone()
            )
            .err()
            .unwrap()
            .to_string(),
            "Malformed last voted fork slots protobuf: None"
        );
        let progress_missing_heaviest_fork_aggregate = WenRestartProgress {
            state: RestartState::HeaviestFork.into(),
            my_heaviest_fork: Some(HeaviestForkRecord {
                slot: 0,
                bankhash: Hash::new_unique().to_string(),
                total_active_stake: 0,
                shred_version: SHRED_VERSION as u32,
                wallclock: 0,
                from: Pubkey::new_unique().to_string(),
            }),
            ..Default::default()
        };
        assert!(write_wen_restart_records(
            &test_state.wen_restart_proto_path,
            &progress_missing_heaviest_fork_aggregate,
        )
        .is_ok());
        assert_eq!(
            initialize(
                &test_state.wen_restart_proto_path,
                VoteTransaction::from(Vote::new(last_voted_fork_slots.clone(), last_vote_bankhash)),
                test_state.blockstore.clone()
            )
            .err()
            .unwrap()
            .to_string(),
            "Malformed progress: HeaviestFork missing final_result in \
             last_voted_fork_slots_aggregate",
        );
        let progress_missing_my_heaviestfork = WenRestartProgress {
            state: RestartState::GenerateSnapshot.into(),
            my_snapshot: Some(GenerateSnapshotRecord {
                slot: 0,
                bankhash: Hash::new_unique().to_string(),
                shred_version: SHRED_VERSION as u32,
                path: "/path/to/snapshot".to_string(),
            }),
            ..Default::default()
        };
        assert!(write_wen_restart_records(
            &test_state.wen_restart_proto_path,
            &progress_missing_my_heaviestfork,
        )
        .is_ok());
        assert_eq!(
            initialize(
                &test_state.wen_restart_proto_path,
                VoteTransaction::from(Vote::new(last_voted_fork_slots.clone(), last_vote_bankhash)),
                test_state.blockstore.clone()
            )
            .err()
            .unwrap()
            .to_string(),
            "Malformed progress: GenerateSnapshot missing my_heaviest_fork",
        );

        // Now test successful initialization.
        assert!(remove_file(&test_state.wen_restart_proto_path).is_ok());
        // Test the case where the file is not found.
        let mut vote = TowerSync::from(vec![(test_state.last_voted_fork_slots[0], 1)]);
        vote.hash = last_vote_bankhash;
        let last_vote = VoteTransaction::from(vote);
        assert_eq!(
            initialize(
                &test_state.wen_restart_proto_path,
                last_vote.clone(),
                test_state.blockstore.clone()
            )
            .unwrap(),
            (
                WenRestartProgressInternalState::Init {
                    last_voted_fork_slots: test_state.last_voted_fork_slots.clone(),
                    last_vote_bankhash
                },
                WenRestartProgress {
                    state: RestartState::Init.into(),
                    ..Default::default()
                }
            )
        );
        let progress = WenRestartProgress {
            state: RestartState::Init.into(),
            my_last_voted_fork_slots: Some(LastVotedForkSlotsRecord {
                last_voted_fork_slots: test_state.last_voted_fork_slots.clone(),
                last_vote_bankhash: last_vote_bankhash.to_string(),
                shred_version: SHRED_VERSION as u32,
                wallclock: 0,
            }),
            ..Default::default()
        };
        assert!(write_wen_restart_records(&test_state.wen_restart_proto_path, &progress,).is_ok());
        assert_eq!(
            initialize(
                &test_state.wen_restart_proto_path,
                last_vote.clone(),
                test_state.blockstore.clone()
            )
            .unwrap(),
            (
                WenRestartProgressInternalState::Init {
                    last_voted_fork_slots: test_state.last_voted_fork_slots.clone(),
                    last_vote_bankhash,
                },
                progress
            )
        );
        let progress = WenRestartProgress {
            state: RestartState::LastVotedForkSlots.into(),
            my_last_voted_fork_slots: Some(LastVotedForkSlotsRecord {
                last_voted_fork_slots: test_state.last_voted_fork_slots.clone(),
                last_vote_bankhash: last_vote_bankhash.to_string(),
                shred_version: SHRED_VERSION as u32,
                wallclock: 0,
            }),
            ..Default::default()
        };
        assert!(write_wen_restart_records(&test_state.wen_restart_proto_path, &progress,).is_ok());
        assert_eq!(
            initialize(
                &test_state.wen_restart_proto_path,
                last_vote.clone(),
                test_state.blockstore.clone()
            )
            .unwrap(),
            (
                WenRestartProgressInternalState::LastVotedForkSlots {
                    last_voted_fork_slots: test_state.last_voted_fork_slots.clone(),
                    aggregate_final_result: None,
                },
                progress
            )
        );
        let progress = WenRestartProgress {
            state: RestartState::HeaviestFork.into(),
            my_heaviest_fork: Some(HeaviestForkRecord {
                slot: 0,
                bankhash: Hash::new_unique().to_string(),
                total_active_stake: 0,
                shred_version: SHRED_VERSION as u32,
                wallclock: 0,
                from: Pubkey::new_unique().to_string(),
            }),
            last_voted_fork_slots_aggregate: Some(LastVotedForkSlotsAggregateRecord {
                received: HashMap::new(),
                final_result: Some(LastVotedForkSlotsAggregateFinal {
                    slots_stake_map: HashMap::new(),
                    epoch_infos: vec![
                        LastVotedForkSlotsEpochInfoRecord {
                            epoch: 1,
                            total_stake: 1000,
                            actively_voting_stake: 800,
                            actively_voting_for_this_epoch_stake: 800,
                        },
                        LastVotedForkSlotsEpochInfoRecord {
                            epoch: 2,
                            total_stake: 1000,
                            actively_voting_stake: 900,
                            actively_voting_for_this_epoch_stake: 900,
                        },
                    ],
                }),
            }),
            ..Default::default()
        };
        assert!(write_wen_restart_records(&test_state.wen_restart_proto_path, &progress,).is_ok());
        assert_eq!(
            initialize(
                &test_state.wen_restart_proto_path,
                last_vote.clone(),
                test_state.blockstore.clone()
            )
            .unwrap(),
            (
                WenRestartProgressInternalState::FindHeaviestFork {
                    aggregate_final_result: LastVotedForkSlotsFinalResult {
                        slots_stake_map: HashMap::new(),
                        epoch_info_vec: vec![
                            LastVotedForkSlotsEpochInfo {
                                epoch: 1,
                                total_stake: 1000,
                                actively_voting_stake: 800,
                                actively_voting_for_this_epoch_stake: 800,
                            },
                            LastVotedForkSlotsEpochInfo {
                                epoch: 2,
                                total_stake: 1000,
                                actively_voting_stake: 900,
                                actively_voting_for_this_epoch_stake: 900,
                            }
                        ],
                    },
                    my_heaviest_fork: progress.my_heaviest_fork.clone(),
                },
                progress
            )
        );
        let progress = WenRestartProgress {
            state: RestartState::GenerateSnapshot.into(),
            my_heaviest_fork: Some(HeaviestForkRecord {
                slot: 0,
                bankhash: Hash::new_unique().to_string(),
                total_active_stake: 0,
                shred_version: SHRED_VERSION as u32,
                wallclock: 0,
                from: Pubkey::new_unique().to_string(),
            }),
            my_snapshot: Some(GenerateSnapshotRecord {
                slot: 0,
                bankhash: Hash::new_unique().to_string(),
                shred_version: SHRED_VERSION as u32,
                path: "/path/to/snapshot".to_string(),
            }),
            ..Default::default()
        };
        assert!(write_wen_restart_records(&test_state.wen_restart_proto_path, &progress,).is_ok());
        assert_eq!(
            initialize(
                &test_state.wen_restart_proto_path,
                VoteTransaction::from(Vote::new(last_voted_fork_slots.clone(), last_vote_bankhash)),
                test_state.blockstore.clone()
            )
            .unwrap(),
            (
                WenRestartProgressInternalState::GenerateSnapshot {
                    my_heaviest_fork_slot: 0,
                    my_snapshot: progress.my_snapshot.clone(),
                },
                progress,
            )
        );
        let last_vote_slot = test_state.last_voted_fork_slots[0];
        let snapshot_slot_hash = Hash::new_unique();
        let progress = WenRestartProgress {
            state: RestartState::Done.into(),
            my_last_voted_fork_slots: Some(LastVotedForkSlotsRecord {
                last_voted_fork_slots: test_state.last_voted_fork_slots.clone(),
                last_vote_bankhash: last_vote_bankhash.to_string(),
                shred_version: SHRED_VERSION as u32,
                wallclock: 0,
            }),
            my_heaviest_fork: Some(HeaviestForkRecord {
                slot: last_vote_slot,
                bankhash: snapshot_slot_hash.to_string(),
                total_active_stake: 0,
                shred_version: SHRED_VERSION as u32,
                wallclock: 0,
                from: Pubkey::new_unique().to_string(),
            }),
            my_snapshot: Some(GenerateSnapshotRecord {
                slot: last_vote_slot,
                bankhash: snapshot_slot_hash.to_string(),
                shred_version: SHRED_VERSION as u32,
                path: "/path/to/snapshot".to_string(),
            }),
            ..Default::default()
        };
        assert!(write_wen_restart_records(&test_state.wen_restart_proto_path, &progress,).is_ok());
        assert_eq!(
            initialize(
                &test_state.wen_restart_proto_path,
                VoteTransaction::from(Vote::new(last_voted_fork_slots, last_vote_bankhash)),
                test_state.blockstore.clone()
            )
            .unwrap(),
            (
                WenRestartProgressInternalState::Done {
                    slot: last_vote_slot,
                    hash: snapshot_slot_hash,
                    shred_version: SHRED_VERSION,
                },
                progress
            )
        );
    }

    #[test]
    fn test_wen_restart_send_last_voted_fork_failures() {
        let ledger_path = get_tmp_ledger_path_auto_delete!();
        let test_state = wen_restart_test_init(&ledger_path);
        let progress = wen_restart_proto::WenRestartProgress {
            state: RestartState::Init.into(),
            ..Default::default()
        };
        let original_progress = progress.clone();
        assert_eq!(
            send_restart_last_voted_fork_slots(
                test_state.cluster_info.clone(),
                &[],
                Hash::new_unique(),
            )
            .err()
            .unwrap()
            .to_string(),
            "Last voted fork cannot be empty"
        );
        assert_eq!(progress, original_progress);
        let last_vote_bankhash = Hash::new_unique();
        let last_voted_fork_slots = test_state.last_voted_fork_slots.clone();
        wen_restart_test_succeed_after_failure(
            test_state,
            last_vote_bankhash,
            WenRestartProgress {
                state: RestartState::LastVotedForkSlots.into(),
                my_last_voted_fork_slots: Some(LastVotedForkSlotsRecord {
                    last_voted_fork_slots,
                    last_vote_bankhash: last_vote_bankhash.to_string(),
                    shred_version: SHRED_VERSION as u32,
                    wallclock: 0,
                }),
                last_voted_fork_slots_aggregate: Some(LastVotedForkSlotsAggregateRecord {
                    received: HashMap::new(),
                    final_result: None,
                }),
                ..Default::default()
            },
        );
    }

    #[test]
    fn test_write_wen_restart_records_failure() {
        let ledger_path = get_tmp_ledger_path_auto_delete!();
        let test_state = wen_restart_test_init(&ledger_path);
        let progress = wen_restart_proto::WenRestartProgress {
            state: RestartState::Init.into(),
            ..Default::default()
        };
        assert!(write_wen_restart_records(&test_state.wen_restart_proto_path, &progress).is_ok());
        change_proto_file_readonly(&test_state.wen_restart_proto_path, true);
        assert_eq!(
            write_wen_restart_records(&test_state.wen_restart_proto_path, &progress)
                .unwrap_err()
                .downcast::<std::io::Error>()
                .unwrap()
                .kind(),
            std::io::ErrorKind::PermissionDenied,
        );
        change_proto_file_readonly(&test_state.wen_restart_proto_path, false);
        assert!(write_wen_restart_records(&test_state.wen_restart_proto_path, &progress).is_ok());
        let last_voted_fork_slots = test_state.last_voted_fork_slots.clone();
        let last_vote_bankhash = Hash::new_unique();
        wen_restart_test_succeed_after_failure(
            test_state,
            last_vote_bankhash,
            WenRestartProgress {
                state: RestartState::LastVotedForkSlots.into(),
                my_last_voted_fork_slots: Some(LastVotedForkSlotsRecord {
                    last_voted_fork_slots,
                    last_vote_bankhash: last_vote_bankhash.to_string(),
                    shred_version: SHRED_VERSION as u32,
                    wallclock: 0,
                }),
                last_voted_fork_slots_aggregate: Some(LastVotedForkSlotsAggregateRecord {
                    received: HashMap::new(),
                    final_result: None,
                }),
                ..Default::default()
            },
        );
    }

    #[test]
    fn test_wen_restart_aggregate_last_voted_fork_stop_and_restart() {
        let ledger_path = get_tmp_ledger_path_auto_delete!();
        let test_state = wen_restart_test_init(&ledger_path);
        let last_vote_slot: Slot = test_state.last_voted_fork_slots[0];
        let last_vote_bankhash = Hash::new_unique();
        let start_time = timestamp();
        assert!(write_wen_restart_records(
            &test_state.wen_restart_proto_path,
            &WenRestartProgress {
                state: RestartState::LastVotedForkSlots.into(),
                my_last_voted_fork_slots: Some(LastVotedForkSlotsRecord {
                    last_voted_fork_slots: test_state.last_voted_fork_slots.clone(),
                    last_vote_bankhash: last_vote_bankhash.to_string(),
                    shred_version: SHRED_VERSION as u32,
                    wallclock: start_time,
                }),
                last_voted_fork_slots_aggregate: Some(LastVotedForkSlotsAggregateRecord {
                    received: HashMap::new(),
                    final_result: None,
                }),
                ..Default::default()
            }
        )
        .is_ok());
        let mut rng = rand::thread_rng();
        let mut expected_messages = HashMap::new();
        let expected_slots_to_repair: Vec<Slot> =
            (last_vote_slot + 1..last_vote_slot + 3).collect();
        let mut last_voted_fork_slots_from_others = test_state.last_voted_fork_slots.clone();
        last_voted_fork_slots_from_others.reverse();
        last_voted_fork_slots_from_others.append(&mut expected_slots_to_repair.clone());
        let progress = WenRestartProgress {
            state: RestartState::LastVotedForkSlots.into(),
            my_last_voted_fork_slots: Some(LastVotedForkSlotsRecord {
                last_voted_fork_slots: test_state.last_voted_fork_slots.clone(),
                last_vote_bankhash: last_vote_bankhash.to_string(),
                shred_version: SHRED_VERSION as u32,
                wallclock: start_time,
            }),
            ..Default::default()
        };
        let validators_to_take: usize =
            (WAIT_FOR_SUPERMAJORITY_THRESHOLD_PERCENT * TOTAL_VALIDATOR_COUNT as u64 / 100 - 1)
                .try_into()
                .unwrap();
        for keypairs in test_state
            .validator_voting_keypairs
            .iter()
            .take(validators_to_take)
        {
            let wen_restart_proto_path_clone = test_state.wen_restart_proto_path.clone();
            let cluster_info_clone = test_state.cluster_info.clone();
            let bank_forks_clone = test_state.bank_forks.clone();
            let blockstore_clone = test_state.blockstore.clone();
            let exit = Arc::new(AtomicBool::new(false));
            let exit_clone = exit.clone();
            let mut progress_clone = progress.clone();
            let last_voted_fork_slots = test_state.last_voted_fork_slots.clone();
            let wen_restart_thread_handle = Builder::new()
                .name("solana-wen-restart".to_string())
                .spawn(move || {
                    assert!(aggregate_restart_last_voted_fork_slots(
                        &wen_restart_proto_path_clone,
                        WAIT_FOR_SUPERMAJORITY_THRESHOLD_PERCENT,
                        cluster_info_clone,
                        &last_voted_fork_slots,
                        bank_forks_clone,
                        blockstore_clone,
                        Arc::new(RwLock::new(Vec::new())),
                        exit_clone,
                        &mut progress_clone,
                    )
                    .is_ok());
                })
                .unwrap();
            let node_pubkey = keypairs.node_keypair.pubkey();
            let node = ContactInfo::new_rand(&mut rng, Some(node_pubkey));
            let last_vote_hash = Hash::new_unique();
            let now = timestamp();
            push_restart_last_voted_fork_slots(
                test_state.cluster_info.clone(),
                &node,
                &last_voted_fork_slots_from_others,
                &last_vote_hash,
                &keypairs.node_keypair,
                now,
            );
            expected_messages.insert(
                node_pubkey.to_string(),
                LastVotedForkSlotsRecord {
                    last_voted_fork_slots: last_voted_fork_slots_from_others.clone(),
                    last_vote_bankhash: last_vote_hash.to_string(),
                    shred_version: SHRED_VERSION as u32,
                    wallclock: now,
                },
            );
            wait_on_expected_progress_with_timeout(
                test_state.wen_restart_proto_path.clone(),
                WenRestartProgress {
                    state: RestartState::LastVotedForkSlots.into(),
                    my_last_voted_fork_slots: Some(LastVotedForkSlotsRecord {
                        last_voted_fork_slots: test_state.last_voted_fork_slots.clone(),
                        last_vote_bankhash: last_vote_bankhash.to_string(),
                        shred_version: SHRED_VERSION as u32,
                        wallclock: start_time,
                    }),
                    last_voted_fork_slots_aggregate: Some(LastVotedForkSlotsAggregateRecord {
                        received: expected_messages.clone(),
                        final_result: None,
                    }),
                    ..Default::default()
                },
            );
            exit.store(true, Ordering::Relaxed);
            let _ = wen_restart_thread_handle.join();
        }

        // Simulating successful repair of missing blocks.
        let _ = insert_slots_into_blockstore(
            test_state.blockstore.clone(),
            last_vote_slot,
            &expected_slots_to_repair,
            TICKS_PER_SLOT,
            test_state.last_blockhash,
        );

        let last_voted_fork_slots = test_state.last_voted_fork_slots.clone();
        wen_restart_test_succeed_after_failure(
            test_state,
            last_vote_bankhash,
            WenRestartProgress {
                state: RestartState::LastVotedForkSlots.into(),
                my_last_voted_fork_slots: Some(LastVotedForkSlotsRecord {
                    last_voted_fork_slots,
                    last_vote_bankhash: last_vote_bankhash.to_string(),
                    shred_version: SHRED_VERSION as u32,
                    wallclock: start_time,
                }),
                last_voted_fork_slots_aggregate: Some(LastVotedForkSlotsAggregateRecord {
                    received: expected_messages,
                    final_result: None,
                }),
                ..Default::default()
            },
        );
    }

    #[test]
    fn test_increment_and_write_wen_restart_records() {
        agave_logger::setup();
        let my_dir = TempDir::new().unwrap();
        let mut wen_restart_proto_path = my_dir.path().to_path_buf();
        wen_restart_proto_path.push("wen_restart_status.proto");
        let last_vote_bankhash = Hash::new_unique();
        let my_last_voted_fork_slots = Some(LastVotedForkSlotsRecord {
            last_voted_fork_slots: vec![0, 1],
            last_vote_bankhash: last_vote_bankhash.to_string(),
            shred_version: 0,
            wallclock: 0,
        });
        let last_voted_fork_slots_aggregate = Some(LastVotedForkSlotsAggregateRecord {
            received: HashMap::new(),
            final_result: Some(LastVotedForkSlotsAggregateFinal {
                slots_stake_map: vec![(0, 900), (1, 800)].into_iter().collect(),
                epoch_infos: vec![LastVotedForkSlotsEpochInfoRecord {
                    epoch: 0,
                    total_stake: 2000,
                    actively_voting_stake: 900,
                    actively_voting_for_this_epoch_stake: 900,
                }],
            }),
        });
        let my_pubkey = Pubkey::new_unique();
        let my_heaviest_fork = Some(HeaviestForkRecord {
            slot: 1,
            bankhash: Hash::default().to_string(),
            total_active_stake: 900,
            shred_version: SHRED_VERSION as u32,
            wallclock: 0,
            from: my_pubkey.to_string(),
        });
        let coordinator_heaviest_fork = Some(HeaviestForkRecord {
            slot: 2,
            bankhash: Hash::new_unique().to_string(),
            total_active_stake: 800,
            shred_version: SHRED_VERSION as u32,
            wallclock: 0,
            from: Pubkey::new_unique().to_string(),
        });
        let my_bankhash = Hash::new_unique();
        let new_shred_version = SHRED_VERSION + 57;
        let my_snapshot = Some(GenerateSnapshotRecord {
            slot: 1,
            bankhash: my_bankhash.to_string(),
            path: "snapshot_1".to_string(),
            shred_version: new_shred_version as u32,
        });
        let expected_slots_stake_map: HashMap<Slot, u64> =
            vec![(0, 900), (1, 800)].into_iter().collect();
        for (entrance_state, exit_state, entrance_progress, exit_progress) in [
            (
                WenRestartProgressInternalState::Init {
                    last_voted_fork_slots: vec![0, 1],
                    last_vote_bankhash,
                },
                WenRestartProgressInternalState::LastVotedForkSlots {
                    last_voted_fork_slots: vec![0, 1],
                    aggregate_final_result: None,
                },
                WenRestartProgress {
                    state: RestartState::LastVotedForkSlots.into(),
                    my_last_voted_fork_slots: my_last_voted_fork_slots.clone(),
                    ..Default::default()
                },
                WenRestartProgress {
                    state: RestartState::LastVotedForkSlots.into(),
                    my_last_voted_fork_slots: my_last_voted_fork_slots.clone(),
                    ..Default::default()
                },
            ),
            (
                WenRestartProgressInternalState::LastVotedForkSlots {
                    last_voted_fork_slots: vec![0, 1],
                    aggregate_final_result: Some(LastVotedForkSlotsFinalResult {
                        slots_stake_map: expected_slots_stake_map.clone(),
                        epoch_info_vec: vec![LastVotedForkSlotsEpochInfo {
                            epoch: 0,
                            total_stake: 2000,
                            actively_voting_stake: 900,
                            actively_voting_for_this_epoch_stake: 900,
                        }],
                    }),
                },
                WenRestartProgressInternalState::FindHeaviestFork {
                    aggregate_final_result: LastVotedForkSlotsFinalResult {
                        slots_stake_map: expected_slots_stake_map.clone(),
                        epoch_info_vec: vec![LastVotedForkSlotsEpochInfo {
                            epoch: 0,
                            total_stake: 2000,
                            actively_voting_stake: 900,
                            actively_voting_for_this_epoch_stake: 900,
                        }],
                    },
                    my_heaviest_fork: None,
                },
                WenRestartProgress {
                    state: RestartState::LastVotedForkSlots.into(),
                    my_last_voted_fork_slots: my_last_voted_fork_slots.clone(),
                    last_voted_fork_slots_aggregate: last_voted_fork_slots_aggregate.clone(),
                    ..Default::default()
                },
                WenRestartProgress {
                    state: RestartState::HeaviestFork.into(),
                    my_last_voted_fork_slots: my_last_voted_fork_slots.clone(),
                    last_voted_fork_slots_aggregate: last_voted_fork_slots_aggregate.clone(),
                    ..Default::default()
                },
            ),
            (
                WenRestartProgressInternalState::FindHeaviestFork {
                    aggregate_final_result: LastVotedForkSlotsFinalResult {
                        slots_stake_map: expected_slots_stake_map,
                        epoch_info_vec: vec![LastVotedForkSlotsEpochInfo {
                            epoch: 0,
                            total_stake: 2000,
                            actively_voting_stake: 900,
                            actively_voting_for_this_epoch_stake: 900,
                        }],
                    },
                    my_heaviest_fork: Some(HeaviestForkRecord {
                        slot: 1,
                        bankhash: Hash::default().to_string(),
                        total_active_stake: 900,
                        shred_version: SHRED_VERSION as u32,
                        wallclock: 0,
                        from: my_pubkey.to_string(),
                    }),
                },
                WenRestartProgressInternalState::HeaviestFork {
                    my_heaviest_fork_slot: 1,
                    my_heaviest_fork_hash: Hash::default(),
                },
                WenRestartProgress {
                    state: RestartState::HeaviestFork.into(),
                    my_last_voted_fork_slots: my_last_voted_fork_slots.clone(),
                    last_voted_fork_slots_aggregate: last_voted_fork_slots_aggregate.clone(),
                    ..Default::default()
                },
                WenRestartProgress {
                    state: RestartState::HeaviestFork.into(),
                    my_last_voted_fork_slots: my_last_voted_fork_slots.clone(),
                    last_voted_fork_slots_aggregate: last_voted_fork_slots_aggregate.clone(),
                    my_heaviest_fork: my_heaviest_fork.clone(),
                    ..Default::default()
                },
            ),
            (
                WenRestartProgressInternalState::HeaviestFork {
                    my_heaviest_fork_slot: 1,
                    my_heaviest_fork_hash: Hash::default(),
                },
                WenRestartProgressInternalState::GenerateSnapshot {
                    my_heaviest_fork_slot: 1,
                    my_snapshot: None,
                },
                WenRestartProgress {
                    state: RestartState::HeaviestFork.into(),
                    my_last_voted_fork_slots: my_last_voted_fork_slots.clone(),
                    last_voted_fork_slots_aggregate: last_voted_fork_slots_aggregate.clone(),
                    my_heaviest_fork: my_heaviest_fork.clone(),
                    coordinator_heaviest_fork: coordinator_heaviest_fork.clone(),
                    ..Default::default()
                },
                WenRestartProgress {
                    state: RestartState::GenerateSnapshot.into(),
                    my_last_voted_fork_slots: my_last_voted_fork_slots.clone(),
                    last_voted_fork_slots_aggregate: last_voted_fork_slots_aggregate.clone(),
                    my_heaviest_fork: my_heaviest_fork.clone(),
                    coordinator_heaviest_fork: coordinator_heaviest_fork.clone(),
                    ..Default::default()
                },
            ),
            (
                WenRestartProgressInternalState::GenerateSnapshot {
                    my_heaviest_fork_slot: 1,
                    my_snapshot: my_snapshot.clone(),
                },
                WenRestartProgressInternalState::Done {
                    slot: 1,
                    hash: my_bankhash,
                    shred_version: new_shred_version,
                },
                WenRestartProgress {
                    state: RestartState::HeaviestFork.into(),
                    my_last_voted_fork_slots: my_last_voted_fork_slots.clone(),
                    last_voted_fork_slots_aggregate: last_voted_fork_slots_aggregate.clone(),
                    my_heaviest_fork: my_heaviest_fork.clone(),
                    coordinator_heaviest_fork: coordinator_heaviest_fork.clone(),
                    ..Default::default()
                },
                WenRestartProgress {
                    state: RestartState::Done.into(),
                    my_last_voted_fork_slots: my_last_voted_fork_slots.clone(),
                    last_voted_fork_slots_aggregate: last_voted_fork_slots_aggregate.clone(),
                    my_heaviest_fork: my_heaviest_fork.clone(),
                    coordinator_heaviest_fork,
                    my_snapshot: my_snapshot.clone(),
                    ..Default::default()
                },
            ),
        ] {
            let mut progress = entrance_progress;
            let state = increment_and_write_wen_restart_records(
                &wen_restart_proto_path,
                entrance_state,
                &mut progress,
            )
            .unwrap();
            assert_eq!(&state, &exit_state);
            assert_eq!(&progress, &exit_progress);
        }
        let mut progress = WenRestartProgress {
            state: RestartState::Done.into(),
            my_last_voted_fork_slots: my_last_voted_fork_slots.clone(),
            last_voted_fork_slots_aggregate: last_voted_fork_slots_aggregate.clone(),
            ..Default::default()
        };
        assert_eq!(
            increment_and_write_wen_restart_records(
                &wen_restart_proto_path,
                WenRestartProgressInternalState::Done {
                    slot: 1,
                    hash: my_bankhash,
                    shred_version: new_shred_version,
                },
                &mut progress
            )
            .unwrap_err()
            .downcast::<WenRestartError>()
            .unwrap(),
            WenRestartError::UnexpectedState(RestartState::Done),
        );
    }

    #[test]
    fn test_find_heaviest_fork_failures() {
        agave_logger::setup();
        let ledger_path = get_tmp_ledger_path_auto_delete!();
        let exit = Arc::new(AtomicBool::new(false));
        let test_state = wen_restart_test_init(&ledger_path);
        let last_vote_slot = test_state.last_voted_fork_slots[0];
        let slot_with_no_block = 1;
        // This fails because corresponding block is not found, which is wrong, we should have
        // repaired all eligible blocks when we exit LastVotedForkSlots state.
        assert_eq!(
            find_heaviest_fork(
                LastVotedForkSlotsFinalResult {
                    slots_stake_map: vec![(0, 900), (slot_with_no_block, 800)]
                        .into_iter()
                        .collect(),
                    epoch_info_vec: vec![LastVotedForkSlotsEpochInfo {
                        epoch: 0,
                        total_stake: 1000,
                        actively_voting_stake: 900,
                        actively_voting_for_this_epoch_stake: 900,
                    }],
                },
                test_state.bank_forks.clone(),
                test_state.blockstore.clone(),
                exit.clone(),
            )
            .unwrap_err()
            .downcast::<WenRestartError>()
            .unwrap(),
            WenRestartError::BlockNotFound(slot_with_no_block),
        );
        // The following fails because we expect to see the first slot in slots_stake_map doesn't chain to local root.
        assert_eq!(
            find_heaviest_fork(
                LastVotedForkSlotsFinalResult {
                    slots_stake_map: vec![(3, 900)].into_iter().collect(),
                    epoch_info_vec: vec![LastVotedForkSlotsEpochInfo {
                        epoch: 0,
                        total_stake: 1000,
                        actively_voting_stake: 900,
                        actively_voting_for_this_epoch_stake: 900,
                    }],
                },
                test_state.bank_forks.clone(),
                test_state.blockstore.clone(),
                exit.clone(),
            )
            .unwrap_err()
            .downcast::<WenRestartError>()
            .unwrap(),
            WenRestartError::BlockNotLinkedToExpectedParent(3, Some(2), 0),
        );
        // The following fails because we expect to see the some slot in slots_stake_map doesn't chain to the
        // one before it.
        assert_eq!(
            find_heaviest_fork(
                LastVotedForkSlotsFinalResult {
                    slots_stake_map: vec![(2, 900), (5, 900)].into_iter().collect(),
                    epoch_info_vec: vec![LastVotedForkSlotsEpochInfo {
                        epoch: 0,
                        total_stake: 1000,
                        actively_voting_stake: 900,
                        actively_voting_for_this_epoch_stake: 900,
                    }],
                },
                test_state.bank_forks.clone(),
                test_state.blockstore.clone(),
                exit.clone(),
            )
            .unwrap_err()
            .downcast::<WenRestartError>()
            .unwrap(),
            WenRestartError::BlockNotLinkedToExpectedParent(5, Some(4), 2),
        );
        // The following fails because the new slot is not full.
        let not_full_slot = last_vote_slot + 5;
        let parent_slot = last_vote_slot;
        let num_slots = (not_full_slot - parent_slot).max(1);
        let mut entries = create_ticks(num_slots * TICKS_PER_SLOT, 0, test_state.last_blockhash);
        assert!(entries.len() > 1);
        entries.pop();
        let shreds = entries_to_test_shreds(&entries, not_full_slot, parent_slot, false, 0);
        test_state
            .blockstore
            .insert_shreds(shreds, None, false)
            .unwrap();
        let mut slots_stake_map: HashMap<Slot, u64> = test_state
            .last_voted_fork_slots
            .iter()
            .map(|slot| (*slot, 900))
            .collect();
        slots_stake_map.insert(not_full_slot, 800);
        assert_eq!(
            find_heaviest_fork(
                LastVotedForkSlotsFinalResult {
                    slots_stake_map,
                    epoch_info_vec: vec![
                        LastVotedForkSlotsEpochInfo {
                            epoch: 0,
                            total_stake: 1000,
                            actively_voting_stake: 900,
                            actively_voting_for_this_epoch_stake: 900,
                        },
                        LastVotedForkSlotsEpochInfo {
                            epoch: 1,
                            total_stake: 1000,
                            actively_voting_stake: 900,
                            actively_voting_for_this_epoch_stake: 900,
                        },
                    ],
                },
                test_state.bank_forks.clone(),
                test_state.blockstore.clone(),
                exit.clone(),
            )
            .unwrap_err()
            .downcast::<WenRestartError>()
            .unwrap(),
            WenRestartError::BlockNotFull(not_full_slot)
        );
        // The following fails because we added two blocks at the end of the chain, they are full in blockstore
        // but the parent of the first one is missing.
        let missing_parent = last_vote_slot.saturating_add(1);
        let new_slot = last_vote_slot.saturating_add(2);
        let new_hash = insert_slots_into_blockstore(
            test_state.blockstore.clone(),
            last_vote_slot,
            &[missing_parent],
            1,
            test_state.last_blockhash,
        );
        let _ = insert_slots_into_blockstore(
            test_state.blockstore.clone(),
            missing_parent,
            &[new_slot],
            TICKS_PER_SLOT,
            new_hash,
        );
        let mut slots_stake_map: HashMap<Slot, u64> = test_state
            .last_voted_fork_slots
            .iter()
            .map(|slot| (*slot, 900))
            .collect();
        slots_stake_map.insert(missing_parent, 800);
        slots_stake_map.insert(new_slot, 800);
        assert_eq!(
            find_heaviest_fork(
                LastVotedForkSlotsFinalResult {
                    slots_stake_map,
                    epoch_info_vec: vec![
                        LastVotedForkSlotsEpochInfo {
                            epoch: 0,
                            total_stake: 1000,
                            actively_voting_stake: 900,
                            actively_voting_for_this_epoch_stake: 900,
                        },
                        LastVotedForkSlotsEpochInfo {
                            epoch: 1,
                            total_stake: 1000,
                            actively_voting_stake: 900,
                            actively_voting_for_this_epoch_stake: 900,
                        },
                    ],
                },
                test_state.bank_forks.clone(),
                test_state.blockstore.clone(),
                exit.clone(),
            )
            .unwrap_err()
            .downcast::<WenRestartError>()
            .unwrap(),
            WenRestartError::BlockNotFrozenAfterReplay(
                missing_parent,
                Some("invalid block error: incomplete block".to_string())
            ),
        );
    }

    fn start_aggregate_heaviest_fork_thread(
        test_state: &WenRestartTestInitResult,
        heaviest_fork_slot: Slot,
        heaviest_fork_bankhash: Hash,
        exit: Arc<AtomicBool>,
        expected_error: Option<WenRestartError>,
    ) -> std::thread::JoinHandle<()> {
        let progress = wen_restart_proto::WenRestartProgress {
            state: RestartState::HeaviestFork.into(),
            my_heaviest_fork: Some(HeaviestForkRecord {
                slot: heaviest_fork_slot,
                bankhash: heaviest_fork_bankhash.to_string(),
                total_active_stake: WAIT_FOR_SUPERMAJORITY_THRESHOLD_PERCENT
                    .saturating_mul(TOTAL_VALIDATOR_COUNT as u64),
                shred_version: SHRED_VERSION as u32,
                wallclock: 0,
                from: test_state.cluster_info.id().to_string(),
            }),
            ..Default::default()
        };
        let wen_restart_path = test_state.wen_restart_proto_path.clone();
        let cluster_info = test_state.cluster_info.clone();
        let bank_forks = test_state.bank_forks.clone();
        Builder::new()
            .name("solana-wen-restart-aggregate-heaviest-fork".to_string())
            .spawn(move || {
                let result = aggregate_restart_heaviest_fork(
                    &wen_restart_path,
                    cluster_info,
                    bank_forks,
                    exit,
                    &mut progress.clone(),
                );
                if let Some(expected_error) = expected_error {
                    assert_eq!(
                        result.unwrap_err().downcast::<WenRestartError>().unwrap(),
                        expected_error
                    );
                } else {
                    assert!(result.is_ok());
                }
            })
            .unwrap()
    }

    #[test]
    fn test_aggregate_heaviest_fork() {
        let ledger_path = get_tmp_ledger_path_auto_delete!();
        let test_state = wen_restart_test_init(&ledger_path);
        let heaviest_fork_slot = test_state.last_voted_fork_slots[0] + 3;
        let heaviest_fork_bankhash = Hash::new_unique();
        let expected_active_stake = (WAIT_FOR_SUPERMAJORITY_THRESHOLD_PERCENT
            - NON_CONFORMING_VALIDATOR_PERCENT)
            * TOTAL_VALIDATOR_COUNT as u64;
        let exit = Arc::new(AtomicBool::new(false));
        let thread = start_aggregate_heaviest_fork_thread(
            &test_state,
            heaviest_fork_slot,
            heaviest_fork_bankhash,
            exit.clone(),
            None,
        );
        let validators_to_take: usize =
            (WAIT_FOR_SUPERMAJORITY_THRESHOLD_PERCENT * TOTAL_VALIDATOR_COUNT as u64 / 100 - 1)
                .try_into()
                .unwrap();
        for keypair in test_state
            .validator_voting_keypairs
            .iter()
            .take(validators_to_take)
        {
            let node_pubkey = keypair.node_keypair.pubkey();
            let node = ContactInfo::new_rand(&mut rand::thread_rng(), Some(node_pubkey));
            let now = timestamp();
            push_restart_heaviest_fork(
                test_state.cluster_info.clone(),
                &node,
                heaviest_fork_slot,
                &heaviest_fork_bankhash,
                expected_active_stake,
                &keypair.node_keypair,
                now,
            );
        }
        exit.store(true, Ordering::Relaxed);
        assert!(thread.join().is_ok());
    }

    #[test]
    fn test_generate_snapshot() {
        agave_logger::setup();
        let ledger_path = get_tmp_ledger_path_auto_delete!();
        let test_state = wen_restart_test_init(&ledger_path);
        let bank_snapshots_dir = tempfile::TempDir::new().unwrap();
        let full_snapshot_archives_dir = tempfile::TempDir::new().unwrap();
        let incremental_snapshot_archives_dir = tempfile::TempDir::new().unwrap();
        let snapshot_config = SnapshotConfig {
            bank_snapshots_dir: bank_snapshots_dir.as_ref().to_path_buf(),
            full_snapshot_archives_dir: full_snapshot_archives_dir.as_ref().to_path_buf(),
            incremental_snapshot_archives_dir: incremental_snapshot_archives_dir
                .as_ref()
                .to_path_buf(),
            usage: SnapshotUsage::LoadAndGenerate,
            ..Default::default()
        };
        let old_root_bank = test_state.bank_forks.read().unwrap().root_bank();
        let old_root_slot = old_root_bank.slot();
        let new_root_slot = test_state.last_voted_fork_slots[1];
        let exit = Arc::new(AtomicBool::new(false));
        let mut slots = test_state.last_voted_fork_slots.clone();
        slots.reverse();
        let old_last_vote_bankhash = find_bankhash_of_heaviest_fork(
            new_root_slot,
            slots,
            test_state.blockstore.clone(),
            test_state.bank_forks.clone(),
            &exit,
        )
        .unwrap();
        // We don't have any full snapshot, so if we call generate_snapshot() on the old
        // root bank now, it should generate a full snapshot.
        let (abs_request_sender, _abs_request_receiver) = unbounded();
        let snapshot_controller =
            SnapshotController::new(abs_request_sender.clone(), snapshot_config, new_root_slot);
        let snapshot_config = snapshot_controller.snapshot_config();
        let generated_record = generate_snapshot(
            test_state.bank_forks.clone(),
            &snapshot_controller,
            &AbsStatus::new_for_tests(),
            test_state.genesis_config_hash,
            old_root_slot,
        )
        .unwrap();
        assert!(Path::new(&generated_record.path).exists());
        assert!(generated_record.path.starts_with(
            snapshot_config
                .full_snapshot_archives_dir
                .to_string_lossy()
                .as_ref()
        ));
        let generated_record = generate_snapshot(
            test_state.bank_forks.clone(),
            &snapshot_controller,
            &AbsStatus::new_for_tests(),
            test_state.genesis_config_hash,
            new_root_slot,
        )
        .unwrap();
        let new_root_bankhash = test_state
            .bank_forks
            .read()
            .unwrap()
            .get(new_root_slot)
            .unwrap()
            .hash();
        assert_ne!(old_last_vote_bankhash, new_root_bankhash);
        let new_shred_version = generated_record.shred_version;
        assert_ne!(new_shred_version, SHRED_VERSION as u32);
        let snapshot_hash = Hash::from_str(
            generated_record
                .path
                .split('-')
                .next_back()
                .unwrap()
                .split('.')
                .next()
                .unwrap(),
        )
        .unwrap();
        assert_eq!(
            generated_record,
            GenerateSnapshotRecord {
                slot: new_root_slot,
                bankhash: new_root_bankhash.to_string(),
                shred_version: new_shred_version,
                path: build_incremental_snapshot_archive_path(
                    &snapshot_config.incremental_snapshot_archives_dir,
                    old_root_slot,
                    new_root_slot,
                    &SnapshotHash(snapshot_hash),
                    snapshot_config.archive_format,
                )
                .display()
                .to_string(),
            },
        );
        // Now generate a snapshot for older slot, it should fail because we already
        // have a full snapshot.
        assert_eq!(
            generate_snapshot(
                test_state.bank_forks.clone(),
                &snapshot_controller,
                &AbsStatus::new_for_tests(),
                test_state.genesis_config_hash,
                old_root_slot,
            )
            .unwrap_err()
            .downcast::<WenRestartError>()
            .unwrap(),
            WenRestartError::GenerateSnapshotWhenOneExists(
                old_root_slot,
                snapshot_config
                    .full_snapshot_archives_dir
                    .to_string_lossy()
                    .to_string()
            ),
        );
        // fails if we already have an incremental snapshot (we just generated one at new_root_slot).
        let older_slot = new_root_slot - 1;
        assert_eq!(
            generate_snapshot(
                test_state.bank_forks.clone(),
                &snapshot_controller,
                &AbsStatus::new_for_tests(),
                test_state.genesis_config_hash,
                older_slot,
            )
            .unwrap_err()
            .downcast::<WenRestartError>()
            .unwrap(),
            WenRestartError::FutureSnapshotExists(
                older_slot,
                new_root_slot,
                snapshot_config
                    .incremental_snapshot_archives_dir
                    .to_string_lossy()
                    .to_string()
            ),
        );
        // Generate snapshot for a slot without any block, it should fail.
        let empty_slot = new_root_slot + 100;
        assert_eq!(
            generate_snapshot(
                test_state.bank_forks.clone(),
                &snapshot_controller,
                &AbsStatus::new_for_tests(),
                test_state.genesis_config_hash,
                empty_slot,
            )
            .unwrap_err()
            .downcast::<WenRestartError>()
            .unwrap(),
            WenRestartError::BlockNotFound(empty_slot),
        );
        // Now turn off snapshot generation, we should generate a full snapshot.
        let snapshot_config = SnapshotConfig {
            bank_snapshots_dir: bank_snapshots_dir.as_ref().to_path_buf(),
            full_snapshot_archives_dir: full_snapshot_archives_dir.as_ref().to_path_buf(),
            incremental_snapshot_archives_dir: incremental_snapshot_archives_dir
                .as_ref()
                .to_path_buf(),
            usage: SnapshotUsage::LoadOnly,
            ..Default::default()
        };
        let snapshot_controller =
            SnapshotController::new(abs_request_sender.clone(), snapshot_config, new_root_slot);
        let snapshot_config = snapshot_controller.snapshot_config();
        let generated_record = generate_snapshot(
            test_state.bank_forks.clone(),
            &snapshot_controller,
            &AbsStatus::new_for_tests(),
            test_state.genesis_config_hash,
            test_state.last_voted_fork_slots[0],
        )
        .unwrap();
        assert!(Path::new(&generated_record.path).exists());
        assert!(generated_record.path.starts_with(
            snapshot_config
                .full_snapshot_archives_dir
                .to_string_lossy()
                .as_ref()
        ));
    }

    #[test]
    fn test_return_ok_after_wait_is_done() {
        let ledger_path = get_tmp_ledger_path_auto_delete!();
        let test_state = wen_restart_test_init(&ledger_path);
        let last_vote_slot = test_state.last_voted_fork_slots[0];
        let last_vote_bankhash = Hash::new_unique();
        let config = WenRestartConfig {
            wen_restart_path: test_state.wen_restart_proto_path.clone(),
            wen_restart_coordinator: test_state.wen_restart_coordinator,
            last_vote: VoteTransaction::from(Vote::new(vec![last_vote_slot], last_vote_bankhash)),
            blockstore: test_state.blockstore.clone(),
            cluster_info: test_state.cluster_info.clone(),
            bank_forks: test_state.bank_forks.clone(),
            wen_restart_repair_slots: Some(Arc::new(RwLock::new(Vec::new()))),
            wait_for_supermajority_threshold_percent: 80,
            snapshot_controller: None,
            abs_status: AbsStatus::new_for_tests(),
            genesis_config_hash: test_state.genesis_config_hash,
            exit: Arc::new(AtomicBool::new(false)),
        };
        assert!(write_wen_restart_records(
            &test_state.wen_restart_proto_path,
            &WenRestartProgress {
                state: RestartState::Done.into(),
                ..Default::default()
            }
        )
        .is_ok());
        assert_eq!(
            wait_for_wen_restart(config.clone())
                .unwrap_err()
                .downcast::<WenRestartError>()
                .unwrap(),
            WenRestartError::MissingSnapshotInProtobuf
        );
        assert!(write_wen_restart_records(
            &test_state.wen_restart_proto_path,
            &WenRestartProgress {
                state: RestartState::Done.into(),
                my_snapshot: Some(GenerateSnapshotRecord {
                    slot: 0,
                    bankhash: Hash::new_unique().to_string(),
                    shred_version: SHRED_VERSION as u32,
                    path: "snapshot".to_string(),
                }),
                ..Default::default()
            }
        )
        .is_ok());
        assert!(wait_for_wen_restart(config).is_ok());
    }

    #[test]
    fn test_receive_restart_heaviest_fork() {
        let mut rng = rand::thread_rng();
        let coordinator_keypair = Keypair::new();
        let node_keypair = Arc::new(Keypair::new());
        let cluster_info = Arc::new(ClusterInfo::new(
            {
                let mut contact_info =
                    ContactInfo::new_localhost(&node_keypair.pubkey(), timestamp());
                contact_info.set_shred_version(SHRED_VERSION);
                contact_info
            },
            node_keypair.clone(),
            SocketAddrSpace::Unspecified,
        ));
        let exit = Arc::new(AtomicBool::new(false));
        let random_keypair = Keypair::new();
        let random_node = ContactInfo::new_rand(&mut rng, Some(random_keypair.pubkey()));
        let random_slot = 3;
        let random_hash = Hash::new_unique();
        push_restart_heaviest_fork(
            cluster_info.clone(),
            &random_node,
            random_slot,
            &random_hash,
            0,
            &random_keypair,
            timestamp(),
        );
        let coordinator_node = ContactInfo::new_rand(&mut rng, Some(coordinator_keypair.pubkey()));
        let coordinator_slot = 6;
        let coordinator_hash = Hash::new_unique();
        push_restart_heaviest_fork(
            cluster_info.clone(),
            &coordinator_node,
            coordinator_slot,
            &coordinator_hash,
            0,
            &coordinator_keypair,
            timestamp(),
        );
        let mut progress = WenRestartProgress {
            state: RestartState::HeaviestFork.into(),
            ..Default::default()
        };
        assert_eq!(
            receive_restart_heaviest_fork(
                coordinator_keypair.pubkey(),
                cluster_info,
                exit,
                &mut progress
            )
            .unwrap(),
            (coordinator_slot, coordinator_hash)
        );
    }

    #[test]
    fn test_repair_heaviest_fork() {
        let ledger_path = get_tmp_ledger_path_auto_delete!();
        let my_heaviest_fork_slot = 1;
        let coordinator_heaviest_slot_parent = 2;
        let coordinator_heaviest_slot = 3;
        let exit = Arc::new(AtomicBool::new(false));
        let blockstore = Arc::new(Blockstore::open(ledger_path.path()).unwrap());
        let wen_restart_repair_slots = Arc::new(RwLock::new(Vec::new()));
        let exit_clone = exit.clone();
        let blockstore_clone = blockstore.clone();
        let wen_restart_repair_slots_clone = wen_restart_repair_slots.clone();
        let repair_heaviest_fork_thread_handle = Builder::new()
            .name("solana-repair-heaviest-fork".to_string())
            .spawn(move || {
                assert!(repair_heaviest_fork(
                    my_heaviest_fork_slot,
                    coordinator_heaviest_slot,
                    exit_clone,
                    blockstore_clone,
                    wen_restart_repair_slots_clone
                )
                .is_ok());
            })
            .unwrap();
        sleep(Duration::from_millis(GOSSIP_SLEEP_MILLIS));
        // When there is nothing in blockstore, should repair the heaviest slot.
        assert_eq!(
            *wen_restart_repair_slots.read().unwrap(),
            vec![coordinator_heaviest_slot]
        );
        // Now add block 3, 3's parent is 2, should repair 2.
        let _ = insert_slots_into_blockstore(
            blockstore.clone(),
            coordinator_heaviest_slot_parent,
            &[coordinator_heaviest_slot],
            TICKS_PER_SLOT,
            Hash::default(),
        );
        sleep(Duration::from_millis(GOSSIP_SLEEP_MILLIS));
        assert_eq!(
            *wen_restart_repair_slots.read().unwrap(),
            vec![coordinator_heaviest_slot_parent]
        );
        // Insert 2 which links to 1, should exit now.
        let _ = insert_slots_into_blockstore(
            blockstore.clone(),
            my_heaviest_fork_slot,
            &[coordinator_heaviest_slot_parent],
            TICKS_PER_SLOT,
            Hash::default(),
        );
        repair_heaviest_fork_thread_handle.join().unwrap();
    }

    #[test]
    fn test_verify_coordinator_heaviest_fork() {
        let ledger_path = get_tmp_ledger_path_auto_delete!();
        let test_state = wen_restart_test_init(&ledger_path);
        let last_vote = test_state.last_voted_fork_slots[0];
        let exit = Arc::new(AtomicBool::new(false));
        // Create two forks: last_vote -> last_vote+1 and last_vote -> last_vote+2
        let root_bank;
        {
            root_bank = test_state.bank_forks.read().unwrap().root_bank().clone();
        }
        let coordinator_slot = last_vote + 1;
        let my_slot = last_vote + 2;
        let _ = insert_slots_into_blockstore(
            test_state.blockstore.clone(),
            last_vote,
            &[coordinator_slot],
            TICKS_PER_SLOT,
            test_state.last_blockhash,
        );
        let _ = insert_slots_into_blockstore(
            test_state.blockstore.clone(),
            last_vote,
            &[my_slot],
            TICKS_PER_SLOT,
            test_state.last_blockhash,
        );
        let wen_restart_repair_slots = Arc::new(RwLock::new(Vec::new()));
        assert_eq!(
            verify_coordinator_heaviest_fork(
                my_slot,
                coordinator_slot,
                &Hash::default(),
                test_state.bank_forks.clone(),
                test_state.blockstore.clone(),
                exit.clone(),
                wen_restart_repair_slots.clone()
            )
            .unwrap_err()
            .downcast::<WenRestartError>()
            .unwrap(),
            WenRestartError::HeaviestForkOnLeaderOnDifferentFork(coordinator_slot, my_slot)
        );
        let coordinator_hash = Hash::new_unique();
        let my_hash = root_bank.hash();
        let root_slot = root_bank.slot();
        assert_eq!(
            verify_coordinator_heaviest_fork(
                root_slot,
                root_slot,
                &coordinator_hash,
                test_state.bank_forks.clone(),
                test_state.blockstore.clone(),
                exit.clone(),
                wen_restart_repair_slots.clone()
            )
            .unwrap_err()
            .downcast::<WenRestartError>()
            .unwrap(),
            WenRestartError::BankHashMismatch(root_slot, my_hash, coordinator_hash)
        );
    }

    #[test]
    fn test_send_and_receive_heaviest_fork() {
        let ledger_path = get_tmp_ledger_path_auto_delete!();
        let test_state = wen_restart_test_init(&ledger_path);
        let last_vote = test_state.last_voted_fork_slots[0];
        let exit = Arc::new(AtomicBool::new(false));
        let mut pushed_slot = 0;
        let mut pushed_hash = Hash::default();
        // The coordinator always sends its own choice.
        let coordinator_slot = last_vote;
        let mut slots = test_state.last_voted_fork_slots.clone();
        slots.reverse();
        let coordinator_hash = find_bankhash_of_heaviest_fork(
            coordinator_slot,
            slots,
            test_state.blockstore.clone(),
            test_state.bank_forks.clone(),
            &exit,
        )
        .unwrap();
        let mut progress = WenRestartProgress {
            state: RestartState::HeaviestFork.into(),
            ..Default::default()
        };
        // Set coordinator to myself, should return my choice.
        let mut config = WenRestartConfig {
            wen_restart_path: test_state.wen_restart_proto_path.clone(),
            wen_restart_coordinator: test_state.cluster_info.id(),
            last_vote: VoteTransaction::from(Vote::new(vec![last_vote], Hash::default())),
            blockstore: test_state.blockstore.clone(),
            cluster_info: test_state.cluster_info.clone(),
            bank_forks: test_state.bank_forks.clone(),
            wen_restart_repair_slots: Some(Arc::new(RwLock::new(Vec::new()))),
            wait_for_supermajority_threshold_percent: 80,
            snapshot_controller: None,
            abs_status: AbsStatus::new_for_tests(),
            genesis_config_hash: test_state.genesis_config_hash,
            exit: exit.clone(),
        };
        assert_eq!(
            send_and_receive_heaviest_fork(
                coordinator_slot,
                coordinator_hash,
                &config,
                &mut progress,
                |slot, hash| {
                    pushed_slot = slot;
                    pushed_hash = hash;
                }
            )
            .unwrap(),
            (coordinator_slot, coordinator_hash)
        );
        assert_eq!(pushed_slot, coordinator_slot);
        assert_eq!(pushed_hash, coordinator_hash);
        // Now set the coordinator the someone else, need to return their choice.
        let coordinator_keypair =
            &test_state.validator_voting_keypairs[COORDINATOR_INDEX].node_keypair;
        config.wen_restart_coordinator = coordinator_keypair.pubkey();
        let mut rng = rand::thread_rng();
        let node = ContactInfo::new_rand(&mut rng, Some(coordinator_keypair.pubkey()));
        let now = timestamp();
        push_restart_heaviest_fork(
            test_state.cluster_info.clone(),
            &node,
            coordinator_slot,
            &coordinator_hash,
            0,
            coordinator_keypair,
            now,
        );
        let my_slot = test_state.last_voted_fork_slots[1];
        let my_hash = test_state
            .bank_forks
            .read()
            .unwrap()
            .get(my_slot)
            .unwrap()
            .hash();
        assert_eq!(
            send_and_receive_heaviest_fork(
                my_slot,
                my_hash,
                &config,
                &mut progress,
                |slot, hash| {
                    pushed_slot = slot;
                    pushed_hash = hash;
                }
            )
            .unwrap(),
            (coordinator_slot, coordinator_hash)
        );
        assert_eq!(pushed_slot, coordinator_slot);
        assert_eq!(pushed_hash, coordinator_hash);
        // my slot on a different fork, should exit with error but still push heaviest fork.
        let my_slot = coordinator_slot + 1;
        let _ = insert_slots_into_blockstore(
            test_state.blockstore.clone(),
            0,
            &[coordinator_slot],
            TICKS_PER_SLOT,
            test_state.last_blockhash,
        );
        let my_hash = Hash::new_unique();
        assert_eq!(
            send_and_receive_heaviest_fork(
                my_slot,
                my_hash,
                &config,
                &mut progress,
                |slot, hash| {
                    pushed_slot = slot;
                    pushed_hash = hash;
                }
            )
            .unwrap_err()
            .downcast::<WenRestartError>()
            .unwrap(),
            WenRestartError::HeaviestForkOnLeaderOnDifferentFork(coordinator_slot, my_slot)
        );
        assert_eq!(pushed_slot, my_slot);
        assert_eq!(pushed_hash, my_hash);
    }

    fn run_and_check_find_bankhash_of_heaviest_fork(
        test_state: &WenRestartTestInitResult,
        slots: &[Slot],
        slot: Slot,
    ) {
        let exit = Arc::new(AtomicBool::new(false));
        assert_eq!(
            find_bankhash_of_heaviest_fork(
                slot,
                slots.to_vec(),
                test_state.blockstore.clone(),
                test_state.bank_forks.clone(),
                &exit,
            )
            .unwrap(),
            test_state
                .bank_forks
                .read()
                .unwrap()
                .get(slot)
                .unwrap()
                .hash()
        );
    }

    #[test]
    fn test_find_bankhash_of_heaviest_fork() {
        let ledger_path = get_tmp_ledger_path_auto_delete!();
        let test_state = wen_restart_test_init(&ledger_path);
        let last_vote = test_state.last_voted_fork_slots[0];
        let mut slots = test_state.last_voted_fork_slots.clone();
        slots.reverse();
        run_and_check_find_bankhash_of_heaviest_fork(&test_state, &slots, last_vote);
        let new_slot = last_vote + 1;
        let _ = insert_slots_into_blockstore(
            test_state.blockstore.clone(),
            last_vote,
            &[new_slot],
            TICKS_PER_SLOT,
            test_state.last_blockhash,
        );
        slots.push(new_slot);
        run_and_check_find_bankhash_of_heaviest_fork(&test_state, &slots, new_slot);
        let slot_full_but_not_replayed = last_vote + 2;
        let _ = insert_slots_into_blockstore(
            test_state.blockstore.clone(),
            last_vote,
            &[slot_full_but_not_replayed],
            TICKS_PER_SLOT,
            test_state.last_blockhash,
        );
        let new_bank = Bank::new_from_parent(
            test_state.bank_forks.read().unwrap().get(new_slot).unwrap(),
            &Pubkey::default(),
            slot_full_but_not_replayed,
        );
        let _ = test_state
            .bank_forks
            .write()
            .unwrap()
            .insert_from_ledger(new_bank);
        run_and_check_find_bankhash_of_heaviest_fork(
            &test_state,
            &slots,
            slot_full_but_not_replayed,
        );
    }
}