Arc v26.09.3

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

What's Changed

• fix(filereplication): a full-size .part with no final file is not "present" by @pujitha24 in #965
• feat(metrics): expose per-peer replication lag gauges by @efegokdemir in #908
• fix(backup): skip oversized backup destination keys by @efegokdemir in #903
• fix(query): rewrite comma cross-join tables to storage paths (#978) by @xe-nvdk in #980
• feat(backup): name skipped files, carry the counts into the listing, add arc_backup_skipped_files (#977) by @xe-nvdk in #981
• fix(rbac): make permission extraction and query rewriting agree by @xe-nvdk in #986
• fix(query): arm the replacement-scan guard on every relation keyword by @xe-nvdk in #992
• fix(rbac): make RBAC enforcement real, and gate the listing endpoints by @xe-nvdk in #1000
• fix(delete): validate the WHERE fragment for table-position literals by @xe-nvdk in #1001
• fix(cq): re-validate the stored definition before each execution by @xe-nvdk in #1002
• fix(filereplication): try the next peer on a checksum mismatch, keep the local copy (#999) by @xe-nvdk in #1003
• Reject unsafe edge-sync names by @efegokdemir in #996
• Validate continuous query database names by @efegokdemir in #995
• fix(cq): re-validate the stored database name before each run (#1010) by @xe-nvdk in #1013
• fix(ingest): start flush timeouts at dequeue and stop Close abandoning flushes (#1006, #1007) by @xe-nvdk in #1017
• fix(sql): disable DuckDB Hive partition inference on every read (#1005) by @xe-nvdk in #1018
• fix(ingest): defer flush when queue is full by @efegokdemir in #997
• fix(compaction): count rows from parquet file metadata by @efegokdemir in #1019
• docs: credit #997 in contributors by @xe-nvdk in #1024
• fix(cq): validate continuous-query measurement names, at the boundary and before each run (#1011) by @xe-nvdk in #1022
• fix(ingest): re-enqueue a deferred flush when a worker frees a slot (#1008) by @xe-nvdk in #1027
• docs(contributing): explain the arcx label and the unbuildable arcx_engine tag by @xe-nvdk in #1029
• fix(build): run cargo from the arcx directory so its cargo config applies by @xe-nvdk in #1028
• fix(config): stop overriding DuckDB's own container-aware memory and thread limits (#1026) by @xe-nvdk in #1031
• fix(api): mark panicked arcx Arrow streams as truncated by @efegokdemir in #904
• fix(api): register arcx Arrow IPC queries in query management by @efegokdemir in #911
• test(api): decode msgpack with the Basekick fork, not upstream v5 by @xe-nvdk in #1033
• fix(config): derive compaction.threads from the container's CPU quota (#1030) by @xe-nvdk in #1035
• fix(wal): checkpoint flushed entries before recovery by @efegokdemir in #998
• docs: release notes and credit for WAL flush checkpoints (#948) by @xe-nvdk in #1044
• fix(ingest): checkpoint WAL entries flushed on the asynchronous path (#1045) by @xe-nvdk in #1046
• feat(arcxrouter): HAVING and the bucket alias join the grouped allow-list by @xe-nvdk in #1047
• fix(wal): make replay idempotent by inheriting the replayed entry's identity (#1009) by @xe-nvdk in #1048
• feat(wal): per-file sequence bounds, a flush-aware purge, and an ordered Rotate (#1009) by @xe-nvdk in #1049
• fix(arcxrouter): shadow compared only the FIRST batch, so it alarmed on results arcx got right by @xe-nvdk in #1050
• fix(tiering): keep a cluster node's tier metadata in step with its own disk by @xe-nvdk in #1062
• fix(wal): purge only data proven flushed by @lecodev-26 in #1056
• fix(audit): copy Fiber request data before asynchronous logging by @0utsights in #871
• fix(cluster): keep Raft alive through the final flush so shutdown registers its files (#1014) by @xe-nvdk in #1063
• fix(cluster): honor fetch cancellation during TCP/TLS connection setup by @pujitha24 in #902
• fix(storage): make ListStaged report only keys DeleteStaged accepts by @pujitha24 in #906
• fix(wal): line-protocol WAL entries carry their database envelope (#889) by @lecodev-26 in #1055
• fix(cluster): ignore identical duplicate token creates by @efegokdemir in #973
• fix(cluster): refresh file origin after failover rewrite by @efegokdemir in #984
• fix(cluster): make Raft manifest batches atomically visible by @efegokdemir in #910
• fix(cluster): notify deletes during snapshot restore by @efegokdemir in #983
• chore(release): prepare 26.09.3 by @xe-nvdk in #1072

New Contributors

• @0utsights made their first contribution in #871

Full Changelog: v26.09.2...v26.09.3

Quick Start

Docker

docker run -d \
  -p 8000:8000 \
  -v arc-data:/app/data \
  ghcr.io/basekick-labs/arc:26.09.3

Debian/Ubuntu (amd64, arm64)

# Download and install
wget https://github.com/basekick-labs/arc/releases/download/v26.09.3/arc_26.09.3_amd64.deb
sudo dpkg -i arc_26.09.3_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/v26.09.3/arc-26.09.3-1.x86_64.rpm
sudo rpm -i arc-26.09.3-1.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/v26.09.3/arc-26.09.3-1-x86_64.pkg.tar.zst
sudo pacman -U arc-26.09.3-1-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/v26.09.3/arc-26.09.3.tgz

Download Artifacts

Documentation

• Full Documentation
• Migration Guide

Community

• Discord: Join
• Issues: [GitHub](https://github.com/basekick-labs/arc/issues