Data retention
Each organization decides how long its data lives. Settings → Retention sets the window; when it is unset, the default of 90 days applies.
What the purge deletes
A nightly job (03:00) removes, for every organization past its window:
- events (errors and transactions)
- structured logs
- session replays
- issues left empty once their events are gone
Issue metadata for issues that still have events, monitors, dashboards, wallet history and settings are not touched.
Choosing a window
| Window | Fits |
|---|---|
| 7–14 days | High-volume services where only the present matters, or tight budgets |
| 30 days | Most teams: a month covers "did we ship this bug?" |
| 90 days | Regulated work, or slow release cycles |
Shorter retention lowers storage cost but not ingestion cost — the events were still metered when they arrived. To spend less, sample (SDK or project sample rate) rather than shortening retention.
Troubleshooting
| Symptom | Cause / fix |
|---|---|
| Data gone earlier than expected | The organization's retention window is shorter than you thought, or it is unset and the 90-day default applies |
| Old data still visible | The purge runs nightly at 03:00 — data disappears at the next run, not the moment you shorten the window |
| Need something before it expires | Export it over the REST API while it is still there |