feat(starfish-core): implement misbehavior tracking pipeline in DagState by piotrm50 · Pull Request #11531 · iotaledger/iota

piotrm50 · 2026-05-14T09:03:07Z

Description of change

Port the runtime misbehavior tracking logic onto the MisbehaviorStore types introduced in #10088.

MisbehaviorStore tracks per-authority misbehavior in two buckets (in_memory + persisted).
On flush: recompute the in-memory window, accumulate evicted counts, write the persisted snapshot to storage.
On startup: restore persisted counts from RocksDB and recompute in-memory counts from cached block refs.
Wire faulty block header detection into MisbehaviorStore from every peer receive site (subscriber main + bundle, header synchronizer including the own-last-header fetch, and both commit syncers).
Source-aware classification: Subscriber keeps UnexpectedAuthority as Unprovable; commit-chain / fetch-shape errors stay Untracked and belong to a separate commit-sync metric.

Stacks on top of #10088. Supersedes #10127.

Links to any relevant issues

fixes https://github.com/iotaledger/iota-private/issues/278
fixes https://github.com/iotaledger/iota-private/issues/280
Part of https://github.com/iotaledger/iota-private/issues/277

How the change has been tested

Basic tests (linting, compilation, formatting, unit/integration tests)
Patch-specific tests (correctness, functionality coverage)
I have added tests that prove my fix is effective or that my feature works
I have checked that new and existing unit tests pass locally with my changes

…ten gauge updates Address pending review comments on PR #11531: - Drop `ErrorSource` enum and `classify_for_source`. With Provable-stays- Provable across all sources, the only source-specific case was the Subscriber's `UnexpectedAuthority` mapping; fold it into `classify_block_header_error` so every classifier path goes through one function. `record_faulty_block_header` no longer takes a source argument. - Pass `context: &Context` to `record_faulty_block_header` and bump the in_memory gauges (`faulty_blocks_provable_by_authority`, `faulty_blocks_unprovable_by_peer` with `source="in_memory"`) on every recorded fault. `flush_faulty_block_header_buffer` resets them to zero, matching the existing pattern for missing_proposals and equivocations. - Drop the verify_fetched_headers recording in commit_syncer fast/regular paths — those errors are fetch-shape and classify as Untracked today. Replace with TODO pointing at the recording entry point for when per-header faults become observable there. - Move `misbehavior_store.reset()` to step 1 of `reinitialize` alongside the other in-memory cache clears.

Port the runtime misbehavior tracking logic from PR #10127 onto MisbehaviorStore tracks per-authority misbehavior in two buckets: - in_memory: recomputed on each flush from blocks in the DAG cache - persisted: cumulative counts from evicted blocks, written to RocksDB On flush, update_misbehavior_counts_on_eviction recomputes the in-memory window, accumulates newly-evicted counts into persisted, and returns a snapshot for atomic storage writes. On startup, persisted counts are restored from storage and in-memory counts are recomputed from cached block refs. Type renames for consistency: - StarfishMisbehaviorCounts → CommitteeMisbehaviorCounts (Mutex-based) - StorageScoringMetrics → MisbehaviorCounts (versioned enum) - WriteBatch.scoring_metrics → WriteBatch.misbehavior_counts - scan_scoring_metrics → scan_misbehavior_counts

Add error classification for faulty block headers, ported from the old Mysticeti consensus crate. Errors are classified as: - Provable: valid signature + protocol violation (author is guilty) - Unprovable: can't prove authorship (bad sig, sync path) - Untracked: not misbehavior (e.g. BlockRejected) Classification is source-aware: subscriber errors keep provable status, while synchronizer/commit_syncer errors are downgraded to unprovable. Add record_faulty_block_header() which buffers events in the in_memory bucket. On flush, buffered counts are moved to persisted and written to storage, ensuring faulty block header counts are never lost.

Add new_with_misbehavior_store() constructor to DagState that accepts an Arc<MisbehaviorStore>, allowing the store to be shared with other components (authority_service, synchronizers). The existing new() constructor is kept for backward compatibility (creates its own store). Authority_node now creates the MisbehaviorStore and passes it to DagState. Add reset() method for fast sync reinitialization.

…rStore Call record_faulty_block_header from the three detection points: - authority_service: subscriber path (ErrorSource::Subscriber) - header_synchronizer: fetched headers path (ErrorSource::Synchronizer) - commit_syncer: fetch_loop error handler (ErrorSource::CommitSyncer) For the commit syncer, add misbehavior_store to the shared Inner struct so both FastCommitSyncer and RegularCommitSyncer can access it. The fetch_loop's Ok(Err(e)) branch now records the fault before the error is consumed by the into() conversion to a static label. Update all test call sites to pass the new misbehavior_store argument to HeaderSynchronizer::start and the two syncer constructors. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Add unit tests for the error classification and record_faulty_block_header pipeline in scoring_metrics_store: - test_subscriber_provable_errors: protocol-rule violations (TooManyAncestors, TooManyTransactions, InvalidTransaction) stay provable via Subscriber - test_subscriber_unprovable_errors: signature/context errors (WrongEpoch, UnexpectedGenesisHeader, UnexpectedAuthority) are unprovable - test_subscriber_untracked_errors: BlockRejected increments nothing - test_synchronizer_downgrades_provable_to_unprovable: provable errors become unprovable when coming from the synchronizer path - test_synchronizer_specific_unprovable_errors: UnexpectedNumberOfHeadersFetched is directly classified as unprovable - test_commit_syncer_downgrades_provable_to_unprovable: same downgrade via the commit syncer path - test_commit_syncer_specific_unprovable_errors: NoCommitReceived and FetchedTransactionsMismatch are classified as unprovable Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

…ibution record_faulty_block_header now takes separate `peer` and `author` arguments and chooses which to charge based on fault type: - Provable (valid signature, protocol violation) → charged to `author` because the block header itself is cryptographic proof of their fault. - Unprovable (bad signature, context-dependent) → charged to `peer` because we cannot verify the author field. This fixes a bug in the additional-block-headers subscriber path where verify() failures on headers authored by any validator were being charged to the claimed author even when the signature was invalid (unprovable), meaning we couldn't trust the author field at all. For synchronizer and commit_syncer paths, all errors are already downgraded to unprovable, so passing peer==author has no effect but makes the intent explicit. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

…it syncer The commit syncer's fetch_loop passes all ConsensusErrors to record_faulty_block_header, but not all errors from verify_commits are block header faults. Commit-level errors (MalformedCommit, UnexpectedStartCommit, UnexpectedCommitSequence, NoCommitReceived, NotEnoughCommitVotes, TooManyCommitsFromPeer, etc.) are faults about the commit chain, not block headers, so they should not increment faulty_blocks_unprovable. Change classify_commit_syncer_error to return Untracked for these commit-level errors. Only genuine block header errors that reach the commit syncer (e.g. invalid voting block header signature via block_verifier.verify) are classified as Unprovable and counted. Update test to verify commit-level errors produce no counts. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Audit all paths where peer-provided block headers are deserialized or verified and ensure misbehavior is recorded in every one: - authority_service: record MalformedHeader on bcs::from_bytes (main block + additional bundle headers) - header_synchronizer: record MalformedHeader in verify_block_headers and wire misbehavior_store through start_fetch_own_last_block_header_task - commit_syncer/mod: record MalformedHeader in verify_commits - commit_syncer/regular & fast: record errors from verify_fetched_headers on the serving peer Simplify classification: collapse classify_subscriber_error and classify_sync_error into a single classify_for_source. Commit-chain and fetch-shape errors (MalformedCommit, UnexpectedNumberOfHeadersFetched, UnexpectedBlockHeaderForCommit, FetchedTransactionsMismatch, etc.) now fall through to Untracked via classify_block_header_error's catch-all; they are protocol-level peer faults that belong to a separate commit-sync metric, not the per-header faulty_blocks counters. ErrorSource derives Clone only in test builds.

…ten gauge updates Address pending review comments on PR #11531: - Drop `ErrorSource` enum and `classify_for_source`. With Provable-stays- Provable across all sources, the only source-specific case was the Subscriber's `UnexpectedAuthority` mapping; fold it into `classify_block_header_error` so every classifier path goes through one function. `record_faulty_block_header` no longer takes a source argument. - Pass `context: &Context` to `record_faulty_block_header` and bump the in_memory gauges (`faulty_blocks_provable_by_authority`, `faulty_blocks_unprovable_by_peer` with `source="in_memory"`) on every recorded fault. `flush_faulty_block_header_buffer` resets them to zero, matching the existing pattern for missing_proposals and equivocations. - Drop the verify_fetched_headers recording in commit_syncer fast/regular paths — those errors are fetch-shape and classify as Untracked today. Replace with TODO pointing at the recording entry point for when per-header faults become observable there. - Move `misbehavior_store.reset()` to step 1 of `reinitialize` alongside the other in-memory cache clears.

- Gate `DagState::new` behind `#[cfg(test)]` (production uses `new_with_misbehavior_store` to share the misbehavior store). - Drop the unused test-only `MisbehaviorStore::dummy_for_test` and `DagState::evicted_rounds` helpers. - Remove redundant `use bcs;` in `scoring_metrics_store::tests`. - Use `BTreeMap::keys()` instead of `iter().map(|(k, _)| ..)` in DagState's flush trace log. - Drop redundant clones at `DagState::new` / `RegularCommitSyncer::new` call sites where the source values are dropped without further use.

…olve gauges Replace the generic `get` + `update` combinators on `CommitteeMisbehaviorCounts` with a fixed set of named methods — `record_block_fault_provable/unprovable`, `drain_block_faults`, `add_block_faults`, `set_block_faults`, `set_dag_faults`, `add_dag_faults`, `snapshot`. The names spell out which counter family the operation touches (block faults vs DAG-observed faults like missing proposals / equivocations) and which arithmetic discipline applies (overwrite vs additive). Three consequences: - The flush path no longer has a TOCTOU window. The previous `get → check → update(0)` sequence in `flush_faulty_block_header_buffer` could lose increments from concurrent `record_faulty_block_header` calls on network threads; `drain_block_faults` now reads-and-zeroes the counters under a single lock acquisition and the Prometheus gauges follow in the same critical section. Counter state and gauge readings can no longer disagree. - Restore-from-storage in `initialize_misbehavior_counts` is now idempotent for both fault families. The old impl used `=` for faulty counts but `+=` for missing/equivocation; the latter would silently double if a future caller forgot to `reset()` first. Both paths now use `set_*` semantics. - `(metrics, hostname)` no longer threads through every call site. `CommitteeMisbehaviorCounts::new` resolves the four Prometheus gauge handles per authority once, stored in a new `AuthorityGauges` struct. `MisbehaviorStore::new(&Context)` replaces `MisbehaviorStore::new(committee_size)` and gives internal methods the context they need at construction time. `record_faulty_block_header` and `update_misbehavior_counts_on_eviction` drop their per-call `&Context` / `hostname` args. 30+ call sites across `authority_service`, `header_synchronizer`, `commit_syncer`, `authority_node`, and `dag_state` simplify correspondingly. Also tightens the `FaultType::Unprovable` doc to cover the pre-signature `UnexpectedAuthority` case, and the `flush_faulty_block_header_buffer` test comment to spell out the second condition under which it returns `None`.

…ily to misbehavior_* The runtime type was renamed `ScoringMetricsStore` → `MisbehaviorStore` and the storage type `StorageScoringMetrics` → `MisbehaviorCounts` in the earlier commit; finish that rename for two name stragglers: - Move `scoring_metrics_store.rs` → `misbehavior_store.rs` via `git mv`, and update the module path everywhere (`crate::scoring_metrics_store` → `crate::misbehavior_store`). - Rename the RocksDB column family string `"scoring_metrics"` → `"misbehavior_counts"`. Safe to change because the column family was introduced in PR #10088, which has not merged.

…ent imports `mod misbehavior_store` and the cascading `misbehavior_store::*` use lines were declared out of alphabetical order, which `cargo +nightly fmt` (and CI's rustfmt check) flags. No code change.

…o::test] The 4 tests that invoke `Context::new_for_test` (`test_provable_errors`, `test_unprovable_errors`, `test_untracked_errors`, `test_provable_fault_charges_both_author_and_serving_peer`) were marked `#[test]`. `Context::new_for_test` transitively touches tokio runtime APIs; under plain `cargo test` that resolves benignly, but under `cargo simtest` msim intercepts tokio and `current_node()` panics at `msim/src/sim/runtime/context.rs:15` because plain `#[test]` runs outside any simulator node context. Matches the existing convention used by other tests in the file that call `Context::new_for_test` (`test_update_misbehavior_counts_on_eviction_edge_cases`, `test_metrics_flush_and_recovery`, `test_no_double_counting_on_restart`). Also declares the `iota-sdk-types` serde feature on starfish-core's own dependency so the crate builds standalone (workspace builds were masking this via feature unification through `iota-types`).

Keep the counter-and-gauge-always-agree invariant: reset() now zeros all four Prometheus gauges under the same per-authority lock that clears the counters.

…11534) # Description of change Stacks on top of #11531. Wires the producer-side misbehavior store (landed in #10088 + #11531) into the consumer in `iota-core` so that the `Scoreboard` / `ReportAggregator` (from #10779) finally receives per-authority misbehavior signal from Starfish. Each `CommittedSubDag` now carries a per-authority absolute snapshot of `persisted + in_memory` counts from `MisbehaviorStore` at commit time. Consumers diff against their own last-seen state if they want deltas — that responsibility doesn't belong in Starfish, and the sum is invariant across the eviction-time move between buckets, so the snapshot is race-free relative to flush. Producer (`starfish-core`): - `MisbehaviorStore::snapshot_totals()` — returns the per-authority absolute totals. - `CommittedSubDag` gets a new `misbehavior_counts: Vec<MisbehaviorCountsV1>` field, threaded at all three production `CommittedSubDag::new` call sites (`commit_solidifier`, `commit_observer`, `commit_syncer/fast`) via the `DagState`-owned `MisbehaviorStore`. - `CommittedSubDag` is **not** serialized over the wire (local `CommitConsumer` channel only), so the added `Vec` is an in-process cost only. Consumer (`iota-core`): - Overrides `ConsensusOutputAPI::misbehavior_counts()` on `starfish_core::CommittedSubDag` to transpose the per-authority struct-of-fields snapshot into the four per-field vecs expected by `ConsensusOutputMisbehaviorCounts`. Downstream wiring in `consensus_handler` and the report aggregator was already in place and waiting for non-empty data. Also folds in a standalone fix: `starfish-core` now declares its `iota-sdk-types` `serde` feature explicitly. Workspace builds were masking the missing feature via unification through `iota-types`; `cargo check -p starfish-core` alone fails without it. ## Links to any relevant issues fixes iotaledger/iota-private#406 Part of iotaledger/iota-private#173 ## How the change has been tested - [x] Basic tests (linting, compilation, formatting, unit/integration tests) - [x] Patch-specific tests (correctness, functionality coverage) - [x] I have added tests that prove my fix is effective or that my feature works - [x] I have checked that new and existing unit tests pass locally with my changes Local verification: - `cargo ci-clippy` — clean. - `cargo nextest run -p starfish-core --lib` — new `test_snapshot_totals_sums_persisted_and_in_memory` passes; existing misbehavior tests pass. - `cargo nextest run -p iota-core --lib consensus_output_api::tests` — new transpose + empty-snapshot tests pass.

…ate (#11531) Port the runtime misbehavior tracking logic onto the `MisbehaviorStore` types introduced in #10088. - `MisbehaviorStore` tracks per-authority misbehavior in two buckets (`in_memory` + `persisted`). - On flush: recompute the in-memory window, accumulate evicted counts, write the persisted snapshot to storage. - On startup: restore persisted counts from RocksDB and recompute in-memory counts from cached block refs. - Wire faulty block header detection into `MisbehaviorStore` from every peer receive site (subscriber main + bundle, header synchronizer including the own-last-header fetch, and both commit syncers). - Source-aware classification: `Subscriber` keeps `UnexpectedAuthority` as Unprovable; commit-chain / fetch-shape errors stay Untracked and belong to a separate commit-sync metric. Stacks on top of #10088. Supersedes #10127. fixes iotaledger/iota-private#278 fixes iotaledger/iota-private#280 Part of iotaledger/iota-private#277 - [x] Basic tests (linting, compilation, formatting, unit/integration tests) - [x] Patch-specific tests (correctness, functionality coverage) - [x] I have added tests that prove my fix is effective or that my feature works - [x] I have checked that new and existing unit tests pass locally with my changes --------- Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

…11534) # Description of change Stacks on top of #11531. Wires the producer-side misbehavior store (landed in #10088 + #11531) into the consumer in `iota-core` so that the `Scoreboard` / `ReportAggregator` (from #10779) finally receives per-authority misbehavior signal from Starfish. Each `CommittedSubDag` now carries a per-authority absolute snapshot of `persisted + in_memory` counts from `MisbehaviorStore` at commit time. Consumers diff against their own last-seen state if they want deltas — that responsibility doesn't belong in Starfish, and the sum is invariant across the eviction-time move between buckets, so the snapshot is race-free relative to flush. Producer (`starfish-core`): - `MisbehaviorStore::snapshot_totals()` — returns the per-authority absolute totals. - `CommittedSubDag` gets a new `misbehavior_counts: Vec<MisbehaviorCountsV1>` field, threaded at all three production `CommittedSubDag::new` call sites (`commit_solidifier`, `commit_observer`, `commit_syncer/fast`) via the `DagState`-owned `MisbehaviorStore`. - `CommittedSubDag` is **not** serialized over the wire (local `CommitConsumer` channel only), so the added `Vec` is an in-process cost only. Consumer (`iota-core`): - Overrides `ConsensusOutputAPI::misbehavior_counts()` on `starfish_core::CommittedSubDag` to transpose the per-authority struct-of-fields snapshot into the four per-field vecs expected by `ConsensusOutputMisbehaviorCounts`. Downstream wiring in `consensus_handler` and the report aggregator was already in place and waiting for non-empty data. Also folds in a standalone fix: `starfish-core` now declares its `iota-sdk-types` `serde` feature explicitly. Workspace builds were masking the missing feature via unification through `iota-types`; `cargo check -p starfish-core` alone fails without it. ## Links to any relevant issues fixes iotaledger/iota-private#406 Part of iotaledger/iota-private#173 ## How the change has been tested - [x] Basic tests (linting, compilation, formatting, unit/integration tests) - [x] Patch-specific tests (correctness, functionality coverage) - [x] I have added tests that prove my fix is effective or that my feature works - [x] I have checked that new and existing unit tests pass locally with my changes Local verification: - `cargo ci-clippy` — clean. - `cargo nextest run -p starfish-core --lib` — new `test_snapshot_totals_sums_persisted_and_in_memory` passes; existing misbehavior tests pass. - `cargo nextest run -p iota-core --lib consensus_output_api::tests` — new transpose + empty-snapshot tests pass.

…ate (#11531) Port the runtime misbehavior tracking logic onto the `MisbehaviorStore` types introduced in #10088. - `MisbehaviorStore` tracks per-authority misbehavior in two buckets (`in_memory` + `persisted`). - On flush: recompute the in-memory window, accumulate evicted counts, write the persisted snapshot to storage. - On startup: restore persisted counts from RocksDB and recompute in-memory counts from cached block refs. - Wire faulty block header detection into `MisbehaviorStore` from every peer receive site (subscriber main + bundle, header synchronizer including the own-last-header fetch, and both commit syncers). - Source-aware classification: `Subscriber` keeps `UnexpectedAuthority` as Unprovable; commit-chain / fetch-shape errors stay Untracked and belong to a separate commit-sync metric. Stacks on top of #10088. Supersedes #10127. fixes iotaledger/iota-private#278 fixes iotaledger/iota-private#280 Part of iotaledger/iota-private#277 - [x] Basic tests (linting, compilation, formatting, unit/integration tests) - [x] Patch-specific tests (correctness, functionality coverage) - [x] I have added tests that prove my fix is effective or that my feature works - [x] I have checked that new and existing unit tests pass locally with my changes --------- Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

…11534) # Description of change Stacks on top of #11531. Wires the producer-side misbehavior store (landed in #10088 + #11531) into the consumer in `iota-core` so that the `Scoreboard` / `ReportAggregator` (from #10779) finally receives per-authority misbehavior signal from Starfish. Each `CommittedSubDag` now carries a per-authority absolute snapshot of `persisted + in_memory` counts from `MisbehaviorStore` at commit time. Consumers diff against their own last-seen state if they want deltas — that responsibility doesn't belong in Starfish, and the sum is invariant across the eviction-time move between buckets, so the snapshot is race-free relative to flush. Producer (`starfish-core`): - `MisbehaviorStore::snapshot_totals()` — returns the per-authority absolute totals. - `CommittedSubDag` gets a new `misbehavior_counts: Vec<MisbehaviorCountsV1>` field, threaded at all three production `CommittedSubDag::new` call sites (`commit_solidifier`, `commit_observer`, `commit_syncer/fast`) via the `DagState`-owned `MisbehaviorStore`. - `CommittedSubDag` is **not** serialized over the wire (local `CommitConsumer` channel only), so the added `Vec` is an in-process cost only. Consumer (`iota-core`): - Overrides `ConsensusOutputAPI::misbehavior_counts()` on `starfish_core::CommittedSubDag` to transpose the per-authority struct-of-fields snapshot into the four per-field vecs expected by `ConsensusOutputMisbehaviorCounts`. Downstream wiring in `consensus_handler` and the report aggregator was already in place and waiting for non-empty data. Also folds in a standalone fix: `starfish-core` now declares its `iota-sdk-types` `serde` feature explicitly. Workspace builds were masking the missing feature via unification through `iota-types`; `cargo check -p starfish-core` alone fails without it. ## Links to any relevant issues fixes iotaledger/iota-private#406 Part of iotaledger/iota-private#173 ## How the change has been tested - [x] Basic tests (linting, compilation, formatting, unit/integration tests) - [x] Patch-specific tests (correctness, functionality coverage) - [x] I have added tests that prove my fix is effective or that my feature works - [x] I have checked that new and existing unit tests pass locally with my changes Local verification: - `cargo ci-clippy` — clean. - `cargo nextest run -p starfish-core --lib` — new `test_snapshot_totals_sums_persisted_and_in_memory` passes; existing misbehavior tests pass. - `cargo nextest run -p iota-core --lib consensus_output_api::tests` — new transpose + empty-snapshot tests pass.

iota-ci added consensus Issues related to the Core Consensus team core-protocol labels May 14, 2026

piotrm50 self-assigned this May 14, 2026

piotrm50 mentioned this pull request May 14, 2026

feat(starfish-core): score integration feature branch #11533

Closed

2 tasks

piotrm50 force-pushed the protocol-research/feat/score-integration-starfish-storage branch from d0d4a00 to b53b4fe Compare May 14, 2026 13:44

piotrm50 force-pushed the protocol-research/feat/score-integration-starfish-dag-state branch from 22ee8a7 to e4ad9e7 Compare May 14, 2026 13:45

piotrm50 mentioned this pull request May 14, 2026

feat(consensus): Score integration to Starfish: dag state updates #10127

Closed

piotrm50 force-pushed the protocol-research/feat/score-integration-starfish-dag-state branch 3 times, most recently from 898f11d to faa517a Compare May 15, 2026 09:11

piotrm50 mentioned this pull request May 15, 2026

feat(starfish-core,iota-core): emit misbehavior counts to iota-core #11534

Merged

4 tasks

piotrm50 force-pushed the protocol-research/feat/score-integration-starfish-dag-state branch 2 times, most recently from 32db939 to d867aca Compare May 15, 2026 13:01

piotrm50 marked this pull request as ready for review May 15, 2026 13:09

piotrm50 requested a review from a team as a code owner May 15, 2026 13:10

Base automatically changed from protocol-research/feat/score-integration-starfish-storage to consensus/feat/starfish-score-integration May 15, 2026 16:35

piotrm50 and others added 12 commits May 15, 2026 18:50

style(starfish-core): rustfmt-sort mod misbehavior_store and depend…

a641ba3

…ent imports `mod misbehavior_store` and the cascading `misbehavior_store::*` use lines were declared out of alphabetical order, which `cargo +nightly fmt` (and CI's rustfmt check) flags. No code change.

piotrm50 force-pushed the protocol-research/feat/score-integration-starfish-dag-state branch from 067b189 to a641ba3 Compare May 15, 2026 17:10

filipdulic approved these changes May 18, 2026

View reviewed changes

Comment thread crates/starfish/core/src/misbehavior_store.rs

piotrm50 requested a review from a team May 18, 2026 12:11

fix(starfish-core): reset gauges in CommitteeMisbehaviorCounts::reset

652115f

Keep the counter-and-gauge-always-agree invariant: reset() now zeros all four Prometheus gauges under the same per-authority lock that clears the counters.

piotrm50 mentioned this pull request May 18, 2026

feat(iota-core): Persist and restore scorer received reports state across restarts #10366

Closed

4 tasks

polinikita approved these changes May 19, 2026

View reviewed changes

alexsporn approved these changes May 19, 2026

View reviewed changes

piotrm50 merged commit 98bbb57 into consensus/feat/starfish-score-integration May 19, 2026
35 checks passed

piotrm50 deleted the protocol-research/feat/score-integration-starfish-dag-state branch May 19, 2026 11:33

piotrm50 mentioned this pull request May 19, 2026

feat(starfish-core,iota-core): Starfish score integration #11569

Open

12 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat(starfish-core): implement misbehavior tracking pipeline in DagState#11531

feat(starfish-core): implement misbehavior tracking pipeline in DagState#11531
piotrm50 merged 15 commits into
consensus/feat/starfish-score-integrationfrom
protocol-research/feat/score-integration-starfish-dag-state

piotrm50 commented May 14, 2026 •

edited

Loading

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

Conversation

piotrm50 commented May 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description of change

Links to any relevant issues

How the change has been tested

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

piotrm50 commented May 14, 2026 •

edited

Loading