Troubleshooting
Use the Health tab in Jira admin to diagnose mirroring status, queue depth, and failed deliveries.
Health tab overview
| Signal | Meaning | | --- | --- | | Per-table status | Whether rows are landing in each mirrored table | | Watermarks | Last successful sync timestamp per stream | | Backfill progress | Pages remaining for initial or history crawl | | Queue depth | Pending deliveries waiting for Fabric | | Dead-letter queue | Failed deliveries available for replay |
Open Jira admin → Apps → Fabric Link → Health.
Normal readings
Two signals look alarming but are usually fine:
- Fabric tables report Initialized until the first change files land — wait for initial backfill or a live event.
- Reconciliation sweep: 0 issues means no changes occurred in that window, not a failure.
Common issues
Entra secret expired
Symptom: Connection errors on Health tab; no new files in OneLake.
Fix: Rotate the secret in Entra, update Connecting Fabric settings in the wizard, and replay dead-letter items if any accumulated.
Field added but column empty on old issues
Symptom: New field appears in selector but historical rows are null.
Fix: Confirm the field backfill job on Health tab. Large projects take time; the column fills as the crawl progresses.
Dead-letter entries
Symptom: Health shows failed deliveries with error detail.
Fix: Resolve the root cause (auth, capacity, network), then use Replay on the dead-letter queue. Replayed items are idempotent.
Health tab → Dead letter → Select row → View error → Replay
Project removed from scope but rows remain
Symptom: Issues from an excluded project still appear in Fabric queries.
Fix: Scope changes stop new exports only. Drop or filter rows in Fabric if you need them removed from analytics.
Still stuck?
TODO: Add support contact / status page URL before launch.
Check the Security page for data-flow questions, or review Permissions & consent if auth is failing repeatedly.