Analysis doesn't finish

A cluster’s root cause analysis on the Fix tab can stall instead of landing a result. Match what you see against the causes below:

• Couldn't start the analysis. Please try again. or Couldn't connect to the server. Please try again.: The run never starts
• Couldn't reach the investigator — the connection dropped. Hit Re-run.: Connection dropped after the run started
• Same message, but the workspace has no credit left: Workspace out of credit
• No message, but the run’s been going for an hour: Run exceeded the one-hour cap

The credit is taken when the run starts and refunded if the run fails, so a failed run should net out to nothing.

The run never starts

Couldn't start the analysis. Please try again. or Couldn't connect to the server. Please try again. The request to start the run failed outright. Nothing started. Re-run to try again.

Connection dropped after the run started

Couldn't reach the investigator — the connection dropped. Hit Re-run. The run did start. The connection carrying its progress back died before anything came through. Re-running is usually safe, but the same message also shows up when the workspace has no credit left, and re-running there just spends another credit without landing a result. Check Workspace out of credit before you re-run again.

Run exceeded the one-hour cap

Every run has a one-hour limit. If it’s still going when that’s reached, the run is cut off and no result lands in the thread. Re-run to start a fresh attempt.

Workspace out of credit

This shows up as the same Couldn't reach the investigator — the connection dropped. Hit Re-run. message you’d see from a dropped connection, not as nothing happening. The workspace needs credit before a run can complete. If there’s none left, the run fails with that message. Add credit to the workspace, then re-run.

What to do

Before you re-run, confirm the cluster still has traces inside the time range you’ve got selected on the Feed. If the window has moved past everything in the cluster, widen or shift it so the cluster’s traces fall inside, since re-running against an empty range spends a credit for nothing.

Press Re-run in the cluster’s header. Its tooltip reads Re-run with current cluster state (1 credit), since each run, including a re-run, draws a fresh credit.

Dive deeper