build(deps): bump pg-boss from 10.4.2 to 12.26.3 in /packages/server

[dependabot]

Bumps pg-boss from 10.4.2 to 12.26.3.

Release notes

Sourced from pg-boss's releases.

12.26.3

What's Changed

Fixes

• groupConcurrency fetch plan regression (#842, #845, thanks @​bcomnes)

  Queues using groupConcurrency could hit a severe query plan regression when a saturated group had a large backlog of pending jobs ahead of eligible jobs from other groups and table statistics underestimated the number of active groups. In one report, job fetch went from 4ms to 400ms and pinned database CPU at 100%.

• fromDrizzle crash on the postgres.js driver (#852)

  unwrapSQLResult now checks element shape and treats an array whose elements have no rows property as the row set itself.

Full Changelog: timgit/pg-boss@12.26.2...12.26.3

12.26.2

What's Changed

Fixes

Recover silent LISTEN/NOTIFY connection drops. The dedicated listener connection could go half-open — TCP dies with no FIN and no error event — leaving notify-enabled workers silently degraded to the 30s polling backstop with no recovery. The listener now runs two independent detectors:

• TCP keepalive on the listen connection for OS-level dead-peer detection.
• Same-session heartbeat that verifies the channel subscription is still live, catching both a hung socket and a lost subscription while the socket looks fine.

On any drop it tears down, reconnects with capped backoff, re-runs LISTEN, and fires the reconnect hook so workers drain jobs missed during the gap. Works identically over SSL/TLS connections.

New options

Heartbeat and keepalive timings are now tunable via the constructor (previously hardcoded) — useful for raising the heartbeat timeout on a loaded database:

• notifyHeartbeatIntervalMs — interval between listener heartbeat checks. Default 10000.
• notifyHeartbeatTimeoutMs — per-heartbeat query timeout; the listener reconnects if a heartbeat exceeds it. Default 5000.
• notifyKeepAliveInitialDelayMs — TCP keepalive initial delay on the listener connection. Default 10000.

Defaults preserve prior behavior, so no action is needed unless you want to tune them.

New Contributors

• @​unix made their first contribution in timgit/pg-boss#851

Full Changelog: timgit/pg-boss@12.26.1...12.26.2

12.26.1

What's Changed

• fetch() now uses start_after <= now() instead of < now() (#848, thanks @​ziemekobel). A job inserted with the default start_after = now() could be invisible to a fetch() issued immediately after the insert. now() is transaction-scoped, and on backends with coarse clock resolution — notably PGlite — consecutive autocommit statements often share the same timestamp, so the strict < skipped the job until the clock ticked. NOTIFY gating already used <=; fetch now matches.

New Contributors

• @​ziemekobel made their first contribution in timgit/pg-boss#848

Full Changelog: timgit/pg-boss@12.26.0...12.26.1

... (truncated)

Commits

• See full diff in compare view

Maintainer changes

This version was pushed to npm by GitHub Actions, a new releaser for pg-boss since your current version.

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

---

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

• @dependabot rebase will rebase this PR
• @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
• @dependabot show <dependency name> ignore conditions will show all of the ignore conditions of the specified dependency
• @dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)