Arc 26.09.2

[xe-nvdk]

Arc v26.09.2

Arc is a high-performance time-series database built on DuckDB, optimized for IoT, observability, and analytics workloads.

What's Changed

• fix(iceberg): publish version hint after metadata copy by @bferanmi806-sketch in fix(iceberg): publish version hint after metadata copy #663
• fix(iceberg): validate reconcile interval at config load by @bferanmi806-sketch in fix(iceberg): validate reconcile interval at config load #664
• fix(iceberg): skip unreadable databases during reconciliation by @bferanmi806-sketch in fix(iceberg): skip unreadable databases during reconciliation #665
• fix(sqlite): harden dedicated WAL and SHM permissions by @bferanmi806-sketch in fix(sqlite): harden dedicated WAL and SHM permissions #666
• perf(iceberg): cache empty measurement negatives by @bferanmi806-sketch in perf(iceberg): cache empty measurement negatives #667
• fix(storage): use errors.As for azure response error inspection (medium(storage): Azure isResponseError reimplements errors.As #319) by @Thundercloud12 in fix(storage): use errors.As for azure response error inspection (#319) #670
• fix(backup): make live SQLite backup and restore safe by @atirna in fix(backup): make live SQLite backup and restore safe #678
• fix(compaction): avoid retaining evicted job history entries (medium(compaction): jobHistory trimming retains old entries #315) by @bferanmi806-sketch in fix(compaction): avoid retaining evicted job history entries (#315) #679
• fix: bound the local directory cache by @mah1104ahm in fix: bound the local directory cache #674
• test(cluster): deflake multi-writer election test on starved runners by @xe-nvdk in test(cluster): deflake multi-writer election test on starved runners #680
• fix(tiering): prune partitions per tier in multi-tier queries by @xe-nvdk in fix(tiering): prune partitions per tier in multi-tier queries #681
• fix(query): pin DuckDB session TimeZone to UTC by @xe-nvdk in fix(query): pin DuckDB session TimeZone to UTC #684
• test(cluster): wait for config replication before stopping the leader by @xe-nvdk in test(cluster): wait for config replication before stopping the leader #685
• fix(tiering): register day-level daily-compacted files for migration by @xe-nvdk in fix(tiering): register day-level daily-compacted files for migration #686
• fix(tiering): register spoke-namespace files; harden tier-drop rule by @xe-nvdk in fix(tiering): register spoke-namespace files; harden tier-drop rule #688
• feat(tiering): cold migration of spoke-namespace files, receipt-aware by @xe-nvdk in feat(tiering): cold migration of spoke-namespace files, receipt-aware #691
• fix(backup): report restart_required for config-only restores by @atirna in fix(backup): report restart_required for config-only restores #689
• feat(backup): stage metadata restores; apply at boot by @xe-nvdk in feat(backup): stage metadata restores; apply at boot #692
• fix(iceberg,backup): complete low(iceberg): audit sweep — interval validation, silent cluster no-run, dropped-DB residue, per-pass hygiene #639 items 2, 5, 7, 8 by @xe-nvdk in fix(iceberg,backup): complete #639 items 2, 5, 7, 8 #693
• feat(iceberg): clear catalog artifacts on database drop by @xe-nvdk in feat(iceberg): clear catalog artifacts on database drop #694
• fix(storage): use calendar days for S3 ranges by @copacabanaservice01 in fix(storage): use calendar days for S3 ranges #690
• fix(storage): anchor S3 range paths to UTC calendar days by @xe-nvdk in fix(storage): anchor S3 range paths to UTC calendar days #695
• docs: add pre-26.06.1 token upgrade runbook by @mah1104ahm in docs: add pre-26.06.1 token upgrade runbook #672
• fix: await MQTT unsubscribe tokens during shutdown by @mah1104ahm in fix: await MQTT unsubscribe tokens during shutdown #673
• fix: prune expired tier cache entries by @mah1104ahm in fix: prune expired tier cache entries #675
• fix(wal): chunk oversized payloads so wide-row ingest stays durable (WAL payload byte cap is irreconcilable with the row-bounded ingest buffer on wide rows: WAL silently records nothing #677) by @atirna in fix(wal): chunk oversized payloads so wide-row ingest stays durable (#677) #696
• feat(cla): require a signed CLA on every pull request by @xe-nvdk in feat(cla): require a signed CLA on every pull request #699
• fix(cluster): add periodic file replication reconciliation (Periodic reconciliation for file replication #393) by @bferanmi806-sketch in fix(cluster): add periodic file replication reconciliation (#393) #697
• fix(helm): protect WAL recovery with a startup probe by @atirna in fix(helm): protect WAL recovery with a startup probe #700
• test(query): cover Arrow IPC stream failures by @be-student in test(query): cover Arrow IPC stream failures #706
• feat(telemetry): report arcli installations seen per instance by @xe-nvdk in feat(telemetry): report arcli installations seen per instance #708
• docs(readme): add arcli to the ecosystem by @xe-nvdk in docs(readme): add arcli to the ecosystem #709
• chore(deps): bump google.golang.org/grpc from 1.83.1 to 1.83.2 in the go_modules group across 1 directory by @dependabot[bot] in chore(deps): bump google.golang.org/grpc from 1.83.1 to 1.83.2 in the go_modules group across 1 directory #710
• docs(release-notes): record the gRPC 1.83.2 bump in 26.09.2 by @xe-nvdk in docs(release-notes): record the gRPC 1.83.2 bump in 26.09.2 #711
• fix(auth): enforce token expiry on cache hits by @xe-nvdk in fix(auth): enforce token expiry on cache hits #712
• fix(query): apply queryTimeout to queryMeasurement endpoint by @MrBeldum in fix(query): apply queryTimeout to queryMeasurement endpoint #701
• fix(cluster): authenticate every coordinator handshake field and response by @xe-nvdk in fix(cluster): authenticate every coordinator handshake field and response #713
• fix(query): enforce query governance on every user-SQL endpoint by @xe-nvdk in fix(query): enforce query governance on every user-SQL endpoint #705
• fix(tiering): push measurement/time-range filters into SQL WHERE by @pujitha24 in fix(tiering): push measurement/time-range filters into SQL WHERE #707
• security(cluster): bind SupportsBinaryEntries into the replicate-sync HMAC by @pujitha24 in security(cluster): bind SupportsBinaryEntries into the replicate-sync HMAC #715
• fix(query): release Arrow IPC resources when the stream panics by @xe-nvdk in fix(query): release Arrow IPC resources when the stream panics #718
• docs(release-notes): disclose the arcli telemetry fields (feat(telemetry): report arcli installations seen per instance #708) by @xe-nvdk in docs(release-notes): disclose the arcli telemetry fields (#708) #719
• fix(api): recover from panics in every response stream writer by @xe-nvdk in fix(api): recover from panics in every response stream writer #720
• fix(query): mark truncated Arrow IPC streams so clients cannot read them as complete by @xe-nvdk in fix(query): mark truncated Arrow IPC streams so clients cannot read them as complete #722
• fix(query): mark truncated JSON responses instead of reporting success by @xe-nvdk in fix(query): mark truncated JSON responses instead of reporting success #725
• Mark results an Enterprise governance row cap truncated (A governance row cap is indistinguishable from a complete result #724) by @xe-nvdk in Mark results an Enterprise governance row cap truncated (#724) #730
• fix(query): record a governance row cap in the query registry by @xe-nvdk in fix(query): record a governance row cap in the query registry #732
• Publish Arrow IPC trailers from the connection goroutine (Data race: Arrow IPC trailers are Set from the stream-writer goroutine while fasthttp serialises the response header #729) by @xe-nvdk in Publish Arrow IPC trailers from the connection goroutine (#729) #734
• fix(query): register the measurement endpoint with the query registry by @xe-nvdk in fix(query): register the measurement endpoint with the query registry #735
• Free Arrow IPC resources from one defer, not two call sites (Arrow IPC cleanup runs twice on the panic path, relying on undocumented idempotency #733) by @xe-nvdk in Free Arrow IPC resources from one defer, not two call sites (#733) #736
• security(query): harden the read-SQL validator (GHSA-w6w2-x8xv-q8x2) by @xe-nvdk in security(query): harden the read-SQL validator (GHSA-w6w2-x8xv-q8x2) #738
• Reject edge-sync spoke IDs containing a parent-directory sequence (Edge-sync spoke IDs containing an embedded .. collide with another spoke's local namespace #737) by @xe-nvdk in Reject edge-sync spoke IDs containing a parent-directory sequence (#737) #742
• Reject malformed local storage paths instead of rewriting them (LocalBackend.sanitizePath rewrites paths instead of rejecting them, which is the root of the #737 collision class #741) by @xe-nvdk in Reject malformed local storage paths instead of rewriting them (#741) #745
• Enforce one storage key contract across local, S3 and Azure (Storage path validation is local-only, so the key contract differs per backend #743) by @xe-nvdk in Enforce one storage key contract across local, S3 and Azure (#743) #748
• Reserve the write-staging suffix and stop manifests overflowing (Remaining many-to-one path mappings outside LocalBackend.validatePath #744) by @xe-nvdk in Reserve the write-staging suffix and stop manifests overflowing (#744) #753
• Quarantine permanently unusable storage keys instead of retrying them (Cleanup and replication loops retry storage.ErrInvalidPath forever instead of quarantining #747) by @xe-nvdk in Quarantine permanently unusable storage keys instead of retrying them (#747) #755
• fix(storage): route the read path through the key contract (Arc's read path bypasses the storage Backend, so the key contract does not cover it #746) by @xe-nvdk in fix(storage): route the read path through the key contract (#746) #754
• docs(sandbox): record why per-tenant allowed_directories scoping is not buildable (Per-tenant scoping of the DuckDB sandbox allowed_directories #641) by @xe-nvdk in docs(sandbox): record why per-tenant allowed_directories scoping is not buildable (#641) #763
• fix(backup): report data files no listing returns (LocalBackend listings do not filter unusable keys, unlike S3 and Azure #756) by @xe-nvdk in fix(backup): report data files no listing returns (#756) #765
• security(query): add PREPARE and EXECUTE to dangerousSQLPattern (Add PREPARE/EXECUTE to dangerousSQLPattern as defense-in-depth #739) by @Thundercloud12 in security(query): add PREPARE and EXECUTE to dangerousSQLPattern (#739) #767
• fix(tiering): quarantine a permanently unusable storage key instead of retrying it every cycle (Tiering migration retries an unusable storage key every cycle with no backoff or attempt cap #758) by @xe-nvdk in fix(tiering): quarantine a permanently unusable storage key instead of retrying it every cycle (#758) #773
• fix(mqtt): prevent TOCTOU race in RestartSubscription with nil-placeholder (high(mqtt): TOCTOU race in RestartSubscription #301) by @Thundercloud12 in fix(mqtt): prevent TOCTOU race in RestartSubscription with nil-placeholder (#301) #766
• fix(backup): a restore no longer reports success while dropping files (Restore reports success while dropping files (no skip accounting) #762) by @xe-nvdk in fix(backup): a restore no longer reports success while dropping files (#762) #778
• fix(ingest): log numeric host coercion at debug level by @lecodev-26 in fix(ingest): log numeric host coercion at debug level #769
• refactor(sql): centralize DuckDB path quoting by @efegokdemir in refactor(sql): centralize DuckDB path quoting #780
• fix(retention): report unusable storage files by @efegokdemir in fix(retention): report unusable storage files #775
• fix(edgesync): reject colon in spoke IDs by @efegokdemir in fix(edgesync): reject colon in spoke IDs #776
• fix(tiering): remove redundant metadata mutex by @efegokdemir in fix(tiering): remove redundant metadata mutex #777
• fix(compaction): list each measurement once in the candidates preview by @alexeymoskalev-devops in fix(compaction): list each measurement once in the candidates preview #789
• fix(sql): centralize remaining DuckDB string escaping by @efegokdemir in fix(sql): centralize remaining DuckDB string escaping #782
• fix(backup): classify temp write failures correctly by @efegokdemir in fix(backup): classify temp write failures correctly #784
• fix(edgesync): bound source paths to storage budget by @efegokdemir in fix(edgesync): bound source paths to storage budget #783
• fix(mqtt): validate placeholder reservation in startSubscriber before… by @Thundercloud12 in fix(mqtt): validate placeholder reservation in startSubscriber before… #788
• fix(cluster): reopen the replication query gate when a manifest entry is deleted (A permanently unpullable manifest entry keeps the query gate red until the process restarts #759, cluster: the #759 self-heal must also remove the catch-up tag — a manifest delete during an in-flight pull re-records the failure and the gate stays red until restart #795) by @xe-nvdk in fix(cluster): reopen the replication query gate when a manifest entry is deleted (#759, #795) #800
• fix(shutdown): retain the WAL when the shutdown flush loses data (fix(shutdown): WAL is purged even when the shutdown buffer flush failed, silently losing data #803) by @xe-nvdk in fix(shutdown): retain the WAL when the shutdown flush loses data (#803) #806
• fix(cluster): sync a follower's manifest with the leader before the startup catch-up walk (cluster: the startup catch-up walk on a follower runs before its Raft log is applied (Barrier is leader-only), so the query gate can open on a stale manifest #799) by @xe-nvdk in fix(cluster): sync a follower's manifest with the leader before the startup catch-up walk (#799) #808
• fix(metrics): wire buffer, audit and MQTT metrics that were never populated (fix(metrics): exported metrics that are never incremented (umbrella) #802) by @xe-nvdk in fix(metrics): wire buffer, audit and MQTT metrics that were never populated (#802) #811
• fix(metrics): count requests and successes on the Arrow query endpoint (fix(metrics): /api/v1/query/arrow counts errors but not requests or successes #801) by @xe-nvdk in fix(metrics): count requests and successes on the Arrow query endpoint (#801) #812
• fix(deploy): correct storage env var, enable WAL, drop no-op autoscaling values (fix(deploy): wrong storage env var in k8s manifests, no-op autoscaling values, OSS chart ships without WAL #804) by @xe-nvdk in fix(deploy): correct storage env var, enable WAL, drop no-op autoscaling values (#804) #814
• fix(shutdown): honour server.shutdown_timeout instead of a hardcoded 30s (fix(shutdown): server.shutdown_timeout is never read; the shutdown budget is hardcoded to 30s #805) by @xe-nvdk in fix(shutdown): honour server.shutdown_timeout instead of a hardcoded 30s (#805) #815
• fix(cluster): FSM delete callback no longer takes the coordinator lock; Stop cannot deadlock on a delete applied during shutdown (cluster: Coordinator.Stop holds c.mu across raftNode.Stop while the FSM delete callback takes c.mu.RLock on the apply goroutine (shutdown deadlock) #797) by @xe-nvdk in fix(cluster): FSM delete callback no longer takes the coordinator lock; Stop cannot deadlock on a delete applied during shutdown (#797) #816
• fix(metrics): remove arc_replication_sequence_gaps_total (fix(cluster): arc_replication_sequence_gaps_total has no detection logic and can never fire #810) by @xe-nvdk in fix(metrics): remove arc_replication_sequence_gaps_total (#810) #820
• fix(metrics): populate the DuckDB connection-pool gauges (fix(metrics): DuckDB pool gauges are never sampled; /api/v1/metrics/query-pool reports zeros #809) by @xe-nvdk in fix(metrics): populate the DuckDB connection-pool gauges (#809) #821
• fix(metrics): remove arc_decomp_buffer_discards_total (fix(metrics): arc_decomp_buffer_discards_total counts a discard the pool never performs #817) by @xe-nvdk in fix(metrics): remove arc_decomp_buffer_discards_total (#817) #822
• fix(iceberg): snapshot-expiry data-file deletion (high(iceberg): ExpireSnapshots misreads successful expiry as failure; PostCommit hook holds delete authority over Arc's data files #632) and spoke-namespace export (high(iceberg): edge-sync hub spoke namespaces are exported as garbage tables — source discovery predates #619 #634) by @xe-nvdk in fix(iceberg): snapshot-expiry data-file deletion (#632) and spoke-namespace export (#634) #824
• fix(test): make the buffer flush-counter test independent of execution order by @xe-nvdk in fix(test): make the buffer flush-counter test independent of execution order #829
• fix(iceberg): refuse incomplete exports by @efegokdemir in fix(iceberg): refuse incomplete exports #785
• fix(compaction): clean exact job temp directory by @efegokdemir in fix(compaction): clean exact job temp directory #786
• fix(cluster): propagate manifest apply context by @efegokdemir in fix(cluster): propagate manifest apply context #787
• fix(rbac): preserve case in table reference deduplication (More lossy encodings that become paths, keys or authorization decisions #750) by @Thundercloud12 in fix(rbac): preserve case in table reference deduplication (#750) #832
• test(cluster): make the forward-apply deadline test independent of a two-voter election by @xe-nvdk in test(cluster): make the forward-apply deadline test independent of a two-voter election #833
• fix(query): keep DECIMAL JSON values numeric by @tayfuryldz in fix(query): keep DECIMAL JSON values numeric #831
• fix(iceberg): re-register files the delete API rewrote in place (high(iceberg): in-place parquet rewrite by row-level DELETE is invisible to the path-only fingerprint — external engines keep serving deleted rows #633) by @xe-nvdk in fix(iceberg): re-register files the delete API rewrote in place (#633) #836
• fix(backup): back up an Iceberg warehouse outside the storage root and stop the restore wedge (medium(backup): iceberg.warehouse outside the storage root is silently omitted from backups — DR restore wedges every measurement #637) by @xe-nvdk in fix(backup): back up an Iceberg warehouse outside the storage root and stop the restore wedge (#637) #839
• fix(mqtt): validate database, mapping targets and measurement names before they reach the buffer (high(mqtt): No database name validation on MQTT topic mapping #300) by @xe-nvdk in fix(mqtt): validate database, mapping targets and measurement names before they reach the buffer (#300) #840
• Clarify minimum files for hourly compaction by @jallegri in Clarify minimum files for hourly compaction #838
• test(cluster): let the forward-apply deadline test wait for the context timer by @xe-nvdk in test(cluster): let the forward-apply deadline test wait for the context timer #843
• fix(query): track Arrow IPC queries in the query registry and slow-query log (high(query): Arrow IPC endpoint bypasses enterprise features #309) by @xe-nvdk in fix(query): track Arrow IPC queries in the query registry and slow-query log (#309) #842
• fix(iceberg): keep a compacted file out of the table until its compaction has committed (medium(iceberg): compaction's upload-then-delete window commits snapshots with duplicate data (~2x rows transiently) #638) by @xe-nvdk in fix(iceberg): keep a compacted file out of the table until its compaction has committed (#638) #844
• fix(api): wrap the arcx stream writers in safeStream (Eight streaming response writers have no panic recovery, so a panic there crashes the process #717) by @xe-nvdk in fix(api): wrap the arcx stream writers in safeStream (#717) #845
• fix(cluster): authorise forwarded writes from the FSM node table and refill the registry on snapshot restore (cluster: the leader resolves a forwarding node only through its in-memory registry, so after a leader restart from snapshot forwarded writes are rejected as unknown node until a re-join #807) by @xe-nvdk in fix(cluster): authorise forwarded writes from the FSM node table and refill the registry on snapshot restore (#807) #852
• fix(cluster): join subsystems outside the coordinator and Raft locks on Stop (cluster: Coordinator.Stop and Node.Stop join subsystems while holding their own locks; hoist the joins out of c.mu / n.mu #813) by @xe-nvdk in fix(cluster): join subsystems outside the coordinator and Raft locks on Stop (#813) #855
• fix(cluster): elect a primary writer, and let the promotion reach the gate that reads it (cluster: the local node's writer state is never set, so IsPrimaryWriter() with a failover manager is always false on the primary itself #850) by @xe-nvdk in fix(cluster): elect a primary writer, and let the promotion reach the gate that reads it (#850) #859
• fix(sql): make the read-SQL gates' view of quoted constructs match DuckDB's by @xe-nvdk in fix(sql): make the read-SQL gates' view of quoted constructs match DuckDB's #860
• fix(cluster): stop trusting a forwarded-apply connection the leader has dropped (cluster: a forwarded write fails once when the cached leader connection has idled past the leader's 30 s read deadline #851) by @xe-nvdk in fix(cluster): stop trusting a forwarded-apply connection the leader has dropped (#851) #861
• fix(cluster): announce the nodes a snapshot restore dropped (cluster: registry keeps nodes that a restored snapshot no longer contains #847) by @xe-nvdk in fix(cluster): announce the nodes a snapshot restore dropped (#847) #863
• fix(cluster): announce a heartbeat from a node the cluster has forgotten (cluster: a heartbeat from a node absent from the registry is dropped silently but still acknowledged #849) by @xe-nvdk in fix(cluster): announce a heartbeat from a node the cluster has forgotten (#849) #864
• chore(helm): default to three writers, the minimum for HA in both patterns by @xe-nvdk in chore(helm): default to three writers, the minimum for HA in both patterns #865
• fix(shutdown): run equal-priority hooks in registration order, and stop schedulers before the coordinator (shutdown: the comment claiming a fixed stop order among priority-50 hooks is wrong; the hook sort is unstable #854) by @xe-nvdk in fix(shutdown): run equal-priority hooks in registration order, and stop schedulers before the coordinator (#854) #866
• fix(cluster): stop replication during shutdown, and stop it without holding the coordinator lock (cluster: StopReplication holds c.mu across the replication receiver's join, the same deadlock shape as #797/#813 #853) by @xe-nvdk in fix(cluster): stop replication during shutdown, and stop it without holding the coordinator lock (#853) #867
• feat(api): add GET /ready/write, a write-pool health check (cluster: a load balancer has no way to target the elected primary writer; /ready is role-agnostic #857) by @xe-nvdk in feat(api): add GET /ready/write, a write-pool health check (#857) #868
• feat(cluster): warn when a cluster has too few writer-role nodes (cluster: readers are not promotion candidates, so the documented 1-writer + N-reader Pattern 1 cannot fail over #856) by @xe-nvdk in feat(cluster): warn when a cluster has too few writer-role nodes (#856) #873
• fix(cluster): an unlimited license no longer rejects every cluster join (cluster: an unlimited license (max_cores = -1) rejects every join, so the cluster never forms #869) by @xe-nvdk in fix(cluster): an unlimited license no longer rejects every cluster join (#869) #874
• fix(helm): the compactor pod never joined the cluster (helm: the compactor pod has no Raft address, so it never joins the cluster #870) by @xe-nvdk in fix(helm): the compactor pod never joined the cluster (#870) #877
• fix(cluster): an unrecognised node role is no longer treated as standalone (cluster: handleJoinRequest stores the joiner's role unvalidated; an unknown role parses as standalone, which can ingest #848) by @xe-nvdk in fix(cluster): an unrecognised node role is no longer treated as standalone (#848) #878
• feat(cluster): only nodes that can ingest vote in Raft elections (cluster: a reader or compactor can win Raft leadership, which stops all singleton work in shared-storage mode #862) by @xe-nvdk in feat(cluster): only nodes that can ingest vote in Raft elections (#862) #879
• fix(cluster): a node that leaves gracefully and restarts re-joins again (cluster: a node that leaves gracefully and restarts does not re-join, and the rest of the cluster never sees it again #858) by @xe-nvdk in fix(cluster): a node that leaves gracefully and restarts re-joins again (#858) #881
• chore(ci): gofmt the eight dirty files and turn the gate on (ci: gofmt the eight dirty files on main and turn the gate on #875) by @xe-nvdk in chore(ci): gofmt the eight dirty files and turn the gate on (#875) #882
• feat(cluster): elect a primary writer even when automatic failover is off (cluster: with no failover manager, every writer-role node reports itself primary, so singleton tasks run N times #872) by @xe-nvdk in feat(cluster): elect a primary writer even when automatic failover is off (#872) #883
• fix(compaction): check the cluster gate on every tick, not only at Start (cluster: the compactor lease never moves to a dedicated compactor, and there is no way to move it by hand #876) by @xe-nvdk in fix(compaction): check the cluster gate on every tick, not only at Start (#876) #884
• fix(cluster): replicate from the primary writer, and survive its restart (cluster: WAL replication attaches to an arbitrary writer, not the primary, so replicas miss the live window #885, cluster: a writer restart wedges WAL replication permanently — sender sequence resets, receiver high-water mark does not #887) by @xe-nvdk in fix(cluster): replicate from the primary writer, and survive its restart (#885, #887) #891
• fix(compaction): append the implicit "time" sort key like ingest does by @pujitha24 in fix(compaction): append the implicit "time" sort key like ingest does #890
• feat(cluster): add DELETE /api/v1/cluster/files endpoint (cluster: DELETE /api/v1/cluster/files must be built as a destructive, cluster-wide operation (confirm, honest wording, audit target, validation, no test scaffolding) #794) by @Thundercloud12 in feat(cluster): add DELETE /api/v1/cluster/files endpoint (#794) #830
• fix(cluster): move the compactor lease to a dedicated compactor, and let an operator move it (cluster: the compactor lease never moves to a dedicated compactor, and there is no way to move it by hand #876) by @xe-nvdk in fix(cluster): move the compactor lease to a dedicated compactor, and let an operator move it (#876) #893
• feat(cluster): report and converge a Raft voter set that disagrees with node roles (cluster: converge an existing cluster's Raft voters onto the role-based rule #880) by @xe-nvdk in feat(cluster): report and converge a Raft voter set that disagrees with node roles (#880) #896
• test(replication): make sequence handshake tests deterministic by @xe-nvdk in test(replication): make sequence handshake tests deterministic #900
• fix(compaction): prevent input filename collisions by @efegokdemir in fix(compaction): prevent input filename collisions #898
• fix(cluster): enforce overall peer fetch timeout by @efegokdemir in fix(cluster): enforce overall peer fetch timeout #899
• fix(compaction): configure cycle deadline and account for cancellation by @efegokdemir in fix(compaction): configure cycle deadline and account for cancellation #922
• fix(query): bind measurement fields the same over every time range by @xe-nvdk in fix(query): bind measurement fields the same over every time range #929
• feat(query): answer a proven-empty time range from the schema anchor (experimental) by @xe-nvdk in feat(query): answer a proven-empty time range from the schema anchor (experimental) #931
• fix(backup): keep field schema anchors out of the database inventory by @xe-nvdk in fix(backup): keep field schema anchors out of the database inventory #932
• fix(backup): back up compaction recovery state and reconcile it on restore by @xe-nvdk in fix(backup): back up compaction recovery state and reconcile it on restore #933
• feat(metrics): count compaction manifests parked as unparseable by @xe-nvdk in feat(metrics): count compaction manifests parked as unparseable #934
• docs: reconcile the 26.09.2 notes for the compaction, schema anchor and backup entries by @xe-nvdk in docs: reconcile the 26.09.2 notes for the compaction, schema anchor and backup entries #935
• test: native acceptance run for range-independent field binding (fix(query): keep measurement field binding stable across time ranges #914) by @xe-nvdk in test: native acceptance run for range-independent field binding (#914) #938
• fix(ci): replace the retired MinIO image with SeaweedFS in the objectstore step by @xe-nvdk in fix(ci): replace the retired MinIO image with SeaweedFS in the objectstore step #943
• feat(compaction): per-database exclusion list (compaction.exclude_databases) by @xe-nvdk in feat(compaction): per-database exclusion list (compaction.exclude_databases) #942
• feat(deploy): replace bundled MinIO with SeaweedFS across chart, compose and docs by @xe-nvdk in feat(deploy): replace bundled MinIO with SeaweedFS across chart, compose and docs #944
• fix(deploy): restore bucket pre-creation and honest crash semantics in the SeaweedFS stacks by @xe-nvdk in fix(deploy): restore bucket pre-creation and honest crash semantics in the SeaweedFS stacks #947
• docs(release-notes): document at-least-once WAL crash-recovery replay (wal: replay after a crash re-ingests already-flushed records — permanent duplicates for tagless measurements #948) by @xe-nvdk in docs(release-notes): document at-least-once WAL crash-recovery replay (#948) #949
• fix(tiering): gate migration on the primary writer in shared-storage mode by @xe-nvdk in fix(tiering): gate migration on the primary writer in shared-storage mode #951
• fix(storage): buffer unrewindable bodies so S3 uploads work over plain HTTP by @xe-nvdk in fix(storage): buffer unrewindable bodies so S3 uploads work over plain HTTP #952
• fix(tiering): keep the cluster manifest in step with migration; gate replicating per-node clusters by @xe-nvdk in fix(tiering): keep the cluster manifest in step with migration; gate replicating per-node clusters #953
• fix(tiering): read an all-cold measurement without a time range; retire hot rows for vanished files by @xe-nvdk in fix(tiering): read an all-cold measurement without a time range; retire hot rows for vanished files #954
• fix(config): drop the cold tier's storage-class and retrieval keys, which were never read by @xe-nvdk in fix(config): drop the cold tier's storage-class and retrieval keys, which were never read #955
• refactor(pruning): move the tier-file walk into the pruning package by @xe-nvdk in refactor(pruning): move the tier-file walk into the pruning package #956
• fix(cluster): never drop a local delete, and drain the pending ones on stop by @xe-nvdk in fix(cluster): never drop a local delete, and drain the pending ones on stop #958
• fix(reconciliation): stop reporting replicated files as orphan storage on per-node clusters by @xe-nvdk in fix(reconciliation): stop reporting replicated files as orphan storage on per-node clusters #960
• fix(cluster): re-pull a node's own files after an empty-disk restore; hold the manifest sweep until then by @xe-nvdk in fix(cluster): re-pull a node's own files after an empty-disk restore; hold the manifest sweep until then #961

New Contributors

• @bferanmi806-sketch made their first contribution in fix(iceberg): publish version hint after metadata copy #663
• @Thundercloud12 made their first contribution in fix(storage): use errors.As for azure response error inspection (#319) #670
• @atirna made their first contribution in fix(backup): make live SQLite backup and restore safe #678
• @mah1104ahm made their first contribution in fix: bound the local directory cache #674
• @copacabanaservice01 made their first contribution in fix(storage): use calendar days for S3 ranges #690
• @be-student made their first contribution in test(query): cover Arrow IPC stream failures #706
• @MrBeldum made their first contribution in fix(query): apply queryTimeout to queryMeasurement endpoint #701
• @pujitha24 made their first contribution in fix(tiering): push measurement/time-range filters into SQL WHERE #707
• @lecodev-26 made their first contribution in fix(ingest): log numeric host coercion at debug level #769
• @efegokdemir made their first contribution in refactor(sql): centralize DuckDB path quoting #780
• @alexeymoskalev-devops made their first contribution in fix(compaction): list each measurement once in the candidates preview #789
• @tayfuryldz made their first contribution in fix(query): keep DECIMAL JSON values numeric #831
• @jallegri made their first contribution in Clarify minimum files for hourly compaction #838

Full Changelog: v26.09.1...v26.09.2

Quick Start

Docker

docker run -d \
  -p 8000:8000 \
  -v arc-data:/app/data \
  ghcr.io/basekick-labs/arc:26.09.2

Debian/Ubuntu (amd64, arm64)

# Download and install
wget https://github.com/basekick-labs/arc/releases/download/v26.09.2/arc_26.09.2_amd64.deb
sudo dpkg -i arc_26.09.2_amd64.deb

# Start and enable
sudo systemctl enable arc
sudo systemctl start arc

# Check status
curl http://localhost:8000/health

RHEL/Fedora/Rocky (x86_64, aarch64)

# Download and install
wget https://github.com/basekick-labs/arc/releases/download/v26.09.2/arc-26.09.2-1.x86_64.rpm
sudo rpm -i arc-26.09.2-1.x86_64.rpm

# Start and enable
sudo systemctl enable arc
sudo systemctl start arc

Arch Linux / Omarchy (x86_64, aarch64)

# Download and install
wget https://github.com/basekick-labs/arc/releases/download/v26.09.2/arc-26.09.2-1-x86_64.pkg.tar.zst
sudo pacman -U arc-26.09.2-1-x86_64.pkg.tar.zst

# Start and enable
sudo systemctl enable arc
sudo systemctl start arc

Kubernetes (Helm)

helm install arc https://github.com/basekick-labs/arc/releases/download/v26.09.2/arc-26.09.2.tgz

Download Artifacts

Platform	Architecture	Package
Docker	amd64, arm64	ghcr.io/basekick-labs/arc:26.09.2
Debian/Ubuntu	amd64	arc_26.09.2_amd64.deb
Debian/Ubuntu	arm64	arc_26.09.2_arm64.deb
RHEL/Fedora	x86_64	arc-26.09.2-1.x86_64.rpm
RHEL/Fedora	aarch64	arc-26.09.2-1.aarch64.rpm
Arch Linux / Omarchy	x86_64	arc-26.09.2-1-x86_64.pkg.tar.zst
Arch Linux / Omarchy	aarch64	arc-26.09.2-1-aarch64.pkg.tar.zst
Kubernetes	-	arc-26.09.2.tgz (Helm)

Documentation

• Full Documentation
• Migration Guide

Community

• Discord: Join
• Issues: GitHub

---

This discussion was created from the release Arc 26.09.2.