v0.3.13-pre.009

This commit is contained in:
2026-09-10 19:56:16 +02:00
parent 1746d47412
commit 665d7d2bb6
15 changed files with 1051 additions and 45 deletions

View File

@@ -1,5 +1,5 @@
// file: crates/ksp-worker-raw-transaction-ingest-lib/src/persistence.rs
// version: 3
// version: 4
/// Canonical entity disposition produced by one Worker Store persistence attempt.
#[derive(Clone, Copy, Debug, Eq, PartialEq)]
@@ -92,11 +92,23 @@ struct RawTransactionIngestPersistenceKey {
signature: ksp_store_lib::RawTransactionSignature,
}
#[derive(Clone, Eq, PartialEq)]
struct RawTransactionIngestCanonicalState {
block_time: std::option::Option<ksp_store_lib::RawTimestamp>,
content_hash: ksp_store_lib::RawContentHash,
format_id: ksp_store_lib::RawFormatId,
format_version: u32,
slot: u64,
}
/// Private bounded run-local cache serializing repeated canonical identities before Store writes.
pub(crate) struct RawTransactionIngestPersistenceConvergence {
max_entries: usize,
entries: std::sync::Mutex<
std::collections::BTreeMap<RawTransactionIngestPersistenceKey, std::sync::Arc<tokio::sync::Mutex<std::option::Option<ksp_store_lib::RawContentHash>>>>,
std::collections::BTreeMap<
RawTransactionIngestPersistenceKey,
std::sync::Arc<tokio::sync::Mutex<std::option::Option<RawTransactionIngestCanonicalState>>>,
>,
>,
}
@@ -110,7 +122,7 @@ impl crate::RawTransactionIngestPersistenceConvergence {
fn entry(
&self,
key: RawTransactionIngestPersistenceKey,
) -> std::option::Option<std::sync::Arc<tokio::sync::Mutex<std::option::Option<ksp_store_lib::RawContentHash>>>> {
) -> std::option::Option<std::sync::Arc<tokio::sync::Mutex<std::option::Option<RawTransactionIngestCanonicalState>>>> {
let mut entries = match self.entries.lock() {
std::result::Result::Ok(value) => value,
std::result::Result::Err(poisoned) => poisoned.into_inner(),
@@ -177,16 +189,16 @@ where
if acquisition.observation().transaction() != &reference {
return std::result::Result::Err(crate::runtime_error("persistence.acquisition_reference_mismatch"));
}
let content_hash = acquisition.transaction().payload().content_hash();
let canonical_state = canonical_state(acquisition.transaction());
let key = RawTransactionIngestPersistenceKey { network: reference.network().clone(), signature: reference.signature() };
let entry = convergence.entry(key);
let entry = match entry {
std::option::Option::Some(value) => value,
std::option::Option::None => return crate::persist_raw_transaction_ingest_acquisition(port, acquisition).await,
};
let mut known_hash = entry.lock().await;
if let std::option::Option::Some(known_hash_value) = *known_hash {
if known_hash_value != content_hash {
let mut known_state = entry.lock().await;
if let std::option::Option::Some(known_state_value) = known_state.as_ref() {
if known_state_value != &canonical_state {
return std::result::Result::Err(crate::content_conflict_error());
}
let (_transaction, observation) = acquisition.into_parts();
@@ -202,11 +214,21 @@ where
std::result::Result::Err(error) => return std::result::Result::Err(error),
};
if outcome.entity() != crate::RawTransactionIngestEntityPersistence::SkippedPurged {
*known_hash = std::option::Option::Some(content_hash);
*known_state = std::option::Option::Some(canonical_state);
}
return std::result::Result::Ok(outcome);
}
fn canonical_state(transaction: &ksp_store_lib::RawTransaction) -> RawTransactionIngestCanonicalState {
return RawTransactionIngestCanonicalState {
block_time: transaction.block_time(),
content_hash: transaction.payload().content_hash(),
format_id: transaction.payload().format_id().clone(),
format_version: transaction.payload().format_version(),
slot: transaction.slot(),
};
}
fn map_store_error(error: ksp_core_lib::Error) -> ksp_core_lib::Result<crate::RawTransactionIngestPersistenceOutcome> {
if error.code() == ksp_store_lib::ERROR_CODE_RAW_CONFLICT {
return std::result::Result::Err(crate::content_conflict_error());

View File

@@ -1,5 +1,5 @@
// file: crates/ksp-worker-raw-transaction-ingest-lib/src/runtime_resources.rs
// version: 23
// version: 24
use sha2::Digest; // rust-rules: trait-import
@@ -85,12 +85,18 @@ impl RawTransactionIngestLiveSource {
};
}
fn uses_hydration(&self) -> bool {
return matches!(self, Self::HeliusTransaction(_) | Self::StandardLogs(_) | Self::Yellowstone(_));
}
async fn run(
self,
settings: crate::RawTransactionIngestSettings,
stop_receiver: tokio::sync::watch::Receiver<bool>,
admission_sender: tokio::sync::mpsc::Sender<crate::RawTransactionIngress>,
global_hydration_registry: std::sync::Arc<RawTransactionIngestGlobalHydrationRegistry>,
hydration_pending_limit: usize,
hydration_in_flight_limit: usize,
inventory_publisher: RawTransactionIngestSourceInventoryPublisher,
) -> ksp_core_lib::Result<()> {
let (source_frontier_sender, mut source_frontier_receiver) =
@@ -98,12 +104,46 @@ impl RawTransactionIngestLiveSource {
let mut source_future = std::boxed::Box::pin(async move {
return match self {
Self::HeliusTransaction(source) => {
source.run(settings, stop_receiver, admission_sender, source_frontier_sender, global_hydration_registry).await
source
.run(
settings,
stop_receiver,
admission_sender,
source_frontier_sender,
global_hydration_registry,
hydration_pending_limit,
hydration_in_flight_limit,
)
.await
},
Self::HttpBlockPolling(source) => source.run(settings, stop_receiver, admission_sender, source_frontier_sender).await,
Self::StandardBlock(source) => source.run(settings, stop_receiver, admission_sender, source_frontier_sender).await,
Self::StandardLogs(source) => source.run(settings, stop_receiver, admission_sender, source_frontier_sender, global_hydration_registry).await,
Self::Yellowstone(source) => source.run(settings, stop_receiver, admission_sender, source_frontier_sender, global_hydration_registry).await,
Self::StandardLogs(source) => {
source
.run(
settings,
stop_receiver,
admission_sender,
source_frontier_sender,
global_hydration_registry,
hydration_pending_limit,
hydration_in_flight_limit,
)
.await
},
Self::Yellowstone(source) => {
source
.run(
settings,
stop_receiver,
admission_sender,
source_frontier_sender,
global_hydration_registry,
hydration_pending_limit,
hydration_in_flight_limit,
)
.await
},
};
});
loop {
@@ -628,6 +668,8 @@ impl crate::RawTransactionIngestYellowstoneSource {
admission_sender: tokio::sync::mpsc::Sender<crate::RawTransactionIngress>,
processing_frontier_sender: tokio::sync::watch::Sender<crate::RawTransactionIngestProcessingFrontierProjection>,
global_hydration_registry: std::sync::Arc<RawTransactionIngestGlobalHydrationRegistry>,
hydration_pending_limit: usize,
hydration_in_flight_limit: usize,
) -> ksp_core_lib::Result<()> {
let opened = tokio::select! {
biased;
@@ -641,7 +683,8 @@ impl crate::RawTransactionIngestYellowstoneSource {
std::result::Result::Err(error) => return std::result::Result::Err(source_transport_error(error.code())),
};
let hydration = self.hydration_context();
let mut coordinator = RawTransactionIngestHydrationCoordinator::with_global_registry(&settings, global_hydration_registry);
let mut coordinator =
RawTransactionIngestHydrationCoordinator::with_global_registry(global_hydration_registry, hydration_pending_limit, hydration_in_flight_limit);
let mut processing_frontier = RawTransactionIngestProcessingFrontierReporter::new(processing_frontier_sender);
let snapshot_source = session.snapshot_source();
if let std::result::Result::Err(error) = processing_frontier.observe_session_snapshot(snapshot_source.current()) {
@@ -865,6 +908,8 @@ impl crate::RawTransactionIngestHeliusTransactionSource {
admission_sender: tokio::sync::mpsc::Sender<crate::RawTransactionIngress>,
processing_frontier_sender: tokio::sync::watch::Sender<crate::RawTransactionIngestProcessingFrontierProjection>,
global_hydration_registry: std::sync::Arc<RawTransactionIngestGlobalHydrationRegistry>,
hydration_pending_limit: usize,
hydration_in_flight_limit: usize,
) -> ksp_core_lib::Result<()> {
let connected = tokio::select! {
biased;
@@ -894,7 +939,8 @@ impl crate::RawTransactionIngestHeliusTransactionSource {
},
};
let hydration = self.hydration_context();
let mut coordinator = RawTransactionIngestHydrationCoordinator::with_global_registry(&settings, global_hydration_registry);
let mut coordinator =
RawTransactionIngestHydrationCoordinator::with_global_registry(global_hydration_registry, hydration_pending_limit, hydration_in_flight_limit);
let mut processing_frontier = RawTransactionIngestProcessingFrontierReporter::new(processing_frontier_sender);
processing_frontier.set_source_state(crate::RawTransactionIngestSourceState::Active);
let mut fault = std::option::Option::None;
@@ -1584,6 +1630,8 @@ impl crate::RawTransactionIngestStandardLogsSource {
admission_sender: tokio::sync::mpsc::Sender<crate::RawTransactionIngress>,
processing_frontier_sender: tokio::sync::watch::Sender<crate::RawTransactionIngestProcessingFrontierProjection>,
global_hydration_registry: std::sync::Arc<RawTransactionIngestGlobalHydrationRegistry>,
hydration_pending_limit: usize,
hydration_in_flight_limit: usize,
) -> ksp_core_lib::Result<()> {
let connected = tokio::select! {
biased;
@@ -1613,7 +1661,8 @@ impl crate::RawTransactionIngestStandardLogsSource {
},
};
let hydration = self.hydration_context();
let mut coordinator = RawTransactionIngestHydrationCoordinator::with_global_registry(&settings, global_hydration_registry);
let mut coordinator =
RawTransactionIngestHydrationCoordinator::with_global_registry(global_hydration_registry, hydration_pending_limit, hydration_in_flight_limit);
let mut processing_frontier = RawTransactionIngestProcessingFrontierReporter::new(processing_frontier_sender);
processing_frontier.set_source_state(crate::RawTransactionIngestSourceState::Active);
let mut fault = std::option::Option::None;
@@ -1939,11 +1988,29 @@ impl crate::RawTransactionIngestRuntimeResources {
return std::result::Result::Err(crate::runtime_error("runtime_resources.source_collection_invalid"));
}
let source_keys = self.sources.iter().map(RawTransactionIngestLiveSource::source_key).collect::<std::vec::Vec<_>>();
let hydration_source_count = self.sources.iter().filter(|source| return source.uses_hydration()).count();
if let std::result::Result::Err(error) =
validate_hydration_fairness_capacity(settings.admission_queue_capacity(), settings.persistence_concurrency(), hydration_source_count)
{
return std::result::Result::Err(error);
}
let hydration_pending_budget = settings.admission_queue_capacity();
let hydration_in_flight_budget = settings.persistence_concurrency();
let inventory = std::sync::Arc::new(std::sync::Mutex::new(RawTransactionIngestSourceInventory::new(source_keys)));
let global_hydration_registry = std::sync::Arc::new(RawTransactionIngestGlobalHydrationRegistry::new(&settings));
let global_hydration_registry =
std::sync::Arc::new(RawTransactionIngestGlobalHydrationRegistry::new(hydration_pending_budget, hydration_in_flight_budget));
let (source_stop_sender, source_stop_receiver) = tokio::sync::watch::channel(false);
let mut children = tokio::task::JoinSet::new();
let mut hydration_entry_index = 0_usize;
for (entry_index, source) in self.sources.into_iter().enumerate() {
let (hydration_pending_limit, hydration_in_flight_limit) = if source.uses_hydration() {
let pending_limit = hydration_pending_limit(hydration_pending_budget, hydration_source_count, hydration_entry_index);
let in_flight_limit = hydration_in_flight_limit(hydration_in_flight_budget, hydration_source_count, hydration_entry_index);
hydration_entry_index += 1;
(pending_limit, in_flight_limit)
} else {
(0, 0)
};
let publisher = RawTransactionIngestSourceInventoryPublisher {
aggregate_sender: processing_frontier_sender.clone(),
entry_index,
@@ -1955,7 +2022,17 @@ impl crate::RawTransactionIngestRuntimeResources {
let source_stop_receiver = source_stop_receiver.clone();
let source_global_hydration_registry = std::sync::Arc::clone(&global_hydration_registry);
let _abort_handle = children.spawn(async move {
return source.run(source_settings, source_stop_receiver, source_admission_sender, source_global_hydration_registry, publisher).await;
return source
.run(
source_settings,
source_stop_receiver,
source_admission_sender,
source_global_hydration_registry,
hydration_pending_limit,
hydration_in_flight_limit,
publisher,
)
.await;
});
}
std::mem::drop(admission_sender);
@@ -1963,6 +2040,33 @@ impl crate::RawTransactionIngestRuntimeResources {
}
}
fn validate_hydration_fairness_capacity(
admission_queue_capacity: usize,
persistence_concurrency: usize,
hydration_source_count: usize,
) -> ksp_core_lib::Result<()> {
if hydration_source_count > admission_queue_capacity {
return std::result::Result::Err(crate::runtime_error("runtime_resources.hydration_pending_capacity_insufficient"));
}
if hydration_source_count > persistence_concurrency {
return std::result::Result::Err(crate::runtime_error("runtime_resources.hydration_concurrency_insufficient"));
}
return std::result::Result::Ok(());
}
fn hydration_pending_limit(total_budget: usize, hydration_source_count: usize, hydration_entry_index: usize) -> usize {
if hydration_source_count == 0 || hydration_entry_index >= hydration_source_count {
return 0;
}
let base = total_budget / hydration_source_count;
let remainder = total_budget % hydration_source_count;
return base + usize::from(hydration_entry_index < remainder);
}
fn hydration_in_flight_limit(total_budget: usize, hydration_source_count: usize, hydration_entry_index: usize) -> usize {
return hydration_pending_limit(total_budget, hydration_source_count, hydration_entry_index);
}
impl std::fmt::Debug for crate::RawTransactionIngestRuntimeResources {
fn fmt(&self, formatter: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
return formatter.debug_struct("RawTransactionIngestRuntimeResources").field("source_count", &self.sources.len()).finish();
@@ -3351,6 +3455,7 @@ enum RawTransactionIngestSharedHydrationResult {
}
struct RawTransactionIngestGlobalHydrationRegistry {
hydration_permits: std::sync::Arc<tokio::sync::Semaphore>,
max_pending: usize,
pending: std::sync::Mutex<
std::collections::BTreeMap<
@@ -3361,9 +3466,10 @@ struct RawTransactionIngestGlobalHydrationRegistry {
}
impl RawTransactionIngestGlobalHydrationRegistry {
fn new(settings: &crate::RawTransactionIngestSettings) -> Self {
fn new(max_pending: usize, max_in_flight: usize) -> Self {
return Self {
max_pending: settings.admission_queue_capacity(),
hydration_permits: std::sync::Arc::new(tokio::sync::Semaphore::new(max_in_flight)),
max_pending,
pending: std::sync::Mutex::new(std::collections::BTreeMap::new()),
};
}
@@ -3416,17 +3522,21 @@ struct RawTransactionIngestHydrationCoordinator {
impl RawTransactionIngestHydrationCoordinator {
#[cfg(test)]
fn new(settings: &crate::RawTransactionIngestSettings) -> Self {
return Self::with_global_registry(settings, std::sync::Arc::new(RawTransactionIngestGlobalHydrationRegistry::new(settings)));
let pending_limit = settings.admission_queue_capacity();
let in_flight_limit = settings.persistence_concurrency().min(pending_limit);
let registry = std::sync::Arc::new(RawTransactionIngestGlobalHydrationRegistry::new(pending_limit, settings.persistence_concurrency()));
return Self::with_global_registry(registry, pending_limit, in_flight_limit);
}
fn with_global_registry(
settings: &crate::RawTransactionIngestSettings,
global_registry: std::sync::Arc<RawTransactionIngestGlobalHydrationRegistry>,
max_pending_signals: usize,
max_in_flight: usize,
) -> Self {
return Self {
global_registry,
max_in_flight: settings.persistence_concurrency(),
max_pending_signals: settings.admission_queue_capacity(),
max_in_flight,
max_pending_signals,
pending_signal_count: 0,
pending: std::collections::BTreeMap::new(),
tasks: RawTransactionIngestHydrationTasks::new(),
@@ -3615,6 +3725,33 @@ async fn fetch_hydration(
return std::result::Result::Ok(RawTransactionIngestHydrationFetch { key, observed });
}
struct RawTransactionIngestHydrationLeaderGuard {
armed: bool,
key: RawTransactionIngestHydrationKey,
registry: std::sync::Arc<RawTransactionIngestGlobalHydrationRegistry>,
}
impl RawTransactionIngestHydrationLeaderGuard {
fn new(registry: std::sync::Arc<RawTransactionIngestGlobalHydrationRegistry>, key: RawTransactionIngestHydrationKey) -> Self {
return Self { armed: true, key, registry };
}
fn disarm(&mut self) {
self.armed = false;
return;
}
}
impl std::ops::Drop for RawTransactionIngestHydrationLeaderGuard {
fn drop(&mut self) {
if self.armed {
self.registry
.publish_and_remove(&self.key, RawTransactionIngestSharedHydrationResult::Failed(crate::ERROR_CODE_RAW_TRANSACTION_INGEST_SOURCE_FAILED));
}
return;
}
}
async fn fetch_hydration_shared(
global_registry: std::sync::Arc<RawTransactionIngestGlobalHydrationRegistry>,
http_pool: ksp_onchain_transport_lib::HttpTransportPool,
@@ -3628,14 +3765,22 @@ async fn fetch_hydration_shared(
std::result::Result::Err(error) => return std::result::Result::Err(error),
};
if leader {
let mut leader_guard = RawTransactionIngestHydrationLeaderGuard::new(std::sync::Arc::clone(&global_registry), key.clone());
let permit = std::sync::Arc::clone(&global_registry.hydration_permits).acquire_owned().await;
let _permit = match permit {
std::result::Result::Ok(value) => value,
std::result::Result::Err(_) => return std::result::Result::Err(crate::runtime_error("source.global_hydration_permit_closed")),
};
let fetched = fetch_hydration(http_pool, hydration_role, expected_network, key.clone(), commitment).await;
return match fetched {
std::result::Result::Ok(value) => {
global_registry.publish_and_remove(&key, RawTransactionIngestSharedHydrationResult::Available(value.observed.clone()));
leader_guard.disarm();
std::result::Result::Ok(value)
},
std::result::Result::Err(error) => {
global_registry.publish_and_remove(&key, RawTransactionIngestSharedHydrationResult::Failed(error.code()));
leader_guard.disarm();
std::result::Result::Err(error)
},
};