Repository navigation
Change the issues type filter
Issues
- #49 · renovate opened
on Mar 19, 2026 2
is:issue state:open
is:issue state:open
Issues
replicatedConfirmation counts peers that do not hold a copy of the record (non-resident or replicateTo-excluded)
Filter by Priority: P1Joining peer's reverse full copy re-streams records the peer originated; generalize #737's origin filter beyond clone attempts
Filter by Priority: P3Blob sender error frames for non-fs throws carry no errorCode/errorStatus, so the receiver holds the resume cursor with no terminal bound
Filter by Priority: P2Continuous CPU profiler: a worker's V8 SamplingEventsProcessor thread can spin at 100% of a core indefinitely (5.2.14, @datadog/pprof 5.15.1)
Filter by Priority: P2Prune table drop markers instead of keeping one per dropped name forever
Filter by Priority: P3replicatedConfirmation timeout is not independently configurable and has no cluster-level test
Filter by Priority: P3largeCatchup stress test's exact_count poll competes with the replay it measures
Filter by Priority: P3Removed node's transaction log is replayed from position 0 on every resubscribe; idle peers' cursors never advance and force a base copy
Filter by Priority: P1Declarative residency policy so threads without application code shard correctly
Filter by Priority: P21Record locks: enabling replication.recordLocks moves all of a database's inbound apply onto one worker, even with no lock use
Filter by Priority: P21Replication: operations from a node principal bypass all permission checks, so a node identity is cluster-wide super_user
Filter by Priority: P3HarperFast/harper-pro#976 · openedon Oct 5, 2026 Run replication on a dedicated worker pool (replication.threads)
Filter by Priority: P1Replication places subscriptions and record-lock ownership on isolated application workers
Filter by Priority: P2Record locks: record_lock_apply_homes cannot complete a transition once a ring member is already gone
Filter by Priority: P2Forwarded replication operations leak requestId into the operation body, breaking replicated set_configuration
Filter by Priority: P2cloneNode: leaderRequest has no response deadline, so a leader that accepts and never answers hangs clone setup forever
Filter by Priority: P3Replication failover: a deferred subscription dispatch consumes its self-catchup claim, so the catch-up range can be lost entirely
Filter by Priority: P2Pace replication sender yields by elapsed time to reduce per-record CPU overhead
Filter by Priority: P2Replication should treat a peer whose database generation changed as a new incarnation, so a restored node does not silently diverge
Filter by Priority: P21Base copy tail replay can overwrite an already-copied record's blob, leaving it permanently 503 while the copy reports complete
Filter by Priority: P11Outbound replication stalls silently when a native transaction handle leaks
Filter by Priority: P21WAF shadow-rule would-block events are logged at info and dropped at the default warn log level
Filter by Priority: P2Contract: do base-copy-acquired rows propagate onward over a live subscription?
Filter by Priority: P3Legacy v4 config-validator ReDoS wedges worktree test setups
Filter by Priority: P3getReplicationSharedStatus allocates per call on the blob-retrieval request path
Filter by Priority: P3