Log retention

Log retention

Delivery logs are pruned daily once they age past the Log retention window. Default 14 days. Tenant-wide setting; not per-webhook.


Where to set it

Webhooks Pro → Settings → Logs → Log retention.

Option

Use when

Option

Use when

7 days

Light auditing; receivers fail rarely.

14 days (default)

Most teams. Covers a typical "we noticed the issue last week" investigation.

30 days

Compliance-leaning; one full month of evidence.

60 days

Two months — useful for receivers with long-cycle reconciliation.

90 days

Maximum. Compliance / audit-heavy environments.


How prune works

A daily background job deletes log rows older than today − retention. The cutoff is based on the row's delivery timestamp.

The Settings page shows a live preview under this setting:

N logs will be deleted, M will remain.

The numbers update as you change the retention option, so you see the impact before saving.

There's also a Prune now button — runs the daily prune immediately instead of waiting for the next scheduled run. Useful right after shrinking the retention window when you want the size reduction to happen now.

The button is disabled while the preview is still calculating, or when there's nothing to delete.


What's pruned and what's kept

Pruned:

  • Log rows older than today − retention, regardless of status (success, failure, skipped, replayed).

Kept:

  • Log rows within the retention window.

  • The webhook configurations themselves — retention only affects the delivery log, not webhooks, payload templates, settings, or the queue.

  • Replay relationships within the kept window — sourceLogId still resolves as long as both rows are within retention. If the parent has been pruned but the replay child hasn't, the child still shows but the chain breaks at the parent.


Picking the right window

A few rules of thumb:

  • Compliance is the binding constraint. If you're required to keep evidence of every notification for 30 days, set 30 — even if 14 would be functionally enough.

  • "How long do I take to notice a broken webhook?" is the floor. If your monitoring would alert you within 2 days, 7 days of retention is plenty. If it might take a week to spot a silent failure, go higher.

  • More retention = more storage. Forge SQL is generous, but very chatty webhooks (thousands of deliveries per day) will accumulate quickly.

  • You can shrink retention any time — the next prune (or Prune now) will catch up.


What does NOT live forever

  • Log rows — pruned per this page.

  • The Run again chain — once a parent or child is pruned, the chain breaks at that point. The remaining rows still work; you just lose the visible link.

  • Compressed bodies — the body of a row is stored compressed (gzip) for size reasons; the viewer transparently decompresses. When the row is pruned, the body goes too.

What lives forever (until you delete it):

  • Webhook configurations.

  • Payload templates.

  • Settings.

  • Workflow post-function bindings.


What about exports

Webhooks Pro doesn't have a built-in "export logs to CSV" yet. If you need long-term archive beyond the retention window:

  • Have your receiver log every incoming request with its own retention policy.

  • Send a parallel webhook to an audit sink (e.g. S3, an ELK pipeline) — see Audit Jira events to S3 or SIEM.

  • For compliance with formal evidence requirements, the receiver-side log is the source of truth — Webhooks Pro's log is for diagnostics, not durable archive.


Storage impact (rough order of magnitude)

Deliveries / day

Retention

Approx rows kept

Deliveries / day

Retention

Approx rows kept

100

14 d

~1,400

1,000

14 d

~14,000

1,000

90 d

~90,000

10,000

14 d

~140,000

10,000

90 d

~900,000

Bodies are compressed; per-row storage depends heavily on payload size (a Slack-shaped delivery is much smaller than a full Jira issue).

If you're hitting noticeable performance issues on the Logs page, shrink retention or run Prune now to clear the backlog.


Related