# Troubleshoot applications

Begin with the layer that failed: creation, runtime, access, data, or saved configuration. Repeatedly starting an app cannot repair a connection, and changing a share link cannot fix a missing workflow.

| Symptom | Check and next action |
| --- | --- |
| App id is taken, but no card matches | Ids are tenant-wide, including Trash and other members' apps. Use the suggested id or choose another. |
| Creation needs more inputs | Complete every required template parameter and connection requirement. |
| Model choice is disabled | Read the capability reason and choose a supported workflow model. |
| Running app will not open yet | Wait for readiness; retry the readiness action before issuing another start. |
| App sleeps while its tab is open | On-demand idle activity is platform-observed activity, not raw browser traffic. |
| Timed app stopped | Start it again; extend the running lease earlier or change its policy in Settings. |
| You do not have access | Check the signed-in identity and tenant, then ask the owner to review the intended sharing mode. |
| Share address says not found | Use a freshly copied link and check revocation, app lifecycle, and sharing availability. |
| Empty table or result view | Check the data-source workflow, run scope, record types, step filter, component binding, and producer activity. |
| Missing workflow | Replace the broken source reference in the builder. |
| Resource unavailable | Retry the lookup; unavailable is not proof of deletion. |
| Cannot delete producer | Remove or rebind the named consumer data sources first. |
| Parameters changed elsewhere | Note needed edits, Reload, and reapply them to the current values. |
| Portable import cannot be retried | Correct the named non-retryable workflow or validation problem and review the changed file. |

## Keep recovery actions specific

**Reload app** refreshes the frame only. **Start** controls runtime availability. **Restore version** changes the definition. **Activate release** changes the fixed shared deployment. **Restore from Trash** returns retained resources and sharing. Select the action that corresponds to the actual problem.

When an operation's outcome is uncertain, refresh its state or use the offered same-request retry. Avoid creating a different operation before you know whether the first succeeded.

If a problem persists, collect the app id, affected section, timestamp, visible status, and exact error text. Include the workflow/run or source/output id when relevant; leave credentials and private document content out of support messages.