Monitoring and Re-Running Analyses

Track queued analyses until they finish, then launch another pass when the first run fails, times out, or needs different AI settings. This lets you recover from transient problems and compare reruns without losing the original result.

Prerequisites

• A running JJI server that you can reach from the web app or the jji CLI.
• An analysis you want to follow, or permission to start one. See Running Your First Analysis.
• For the web app, a saved username/profile so you can open the Dashboard.

Note: This page covers queued Jenkins-backed analyses. For direct XML or raw failure analysis, see Analyzing JUnit XML and Raw Failures.

Quick Example

jji analyze --job-name "folder/job-name" --build-number 123 --provider claude --model "<model>"
jji status <job-id>
jji re-analyze <job-id>

Use the job_id printed by the first command. The rerun keeps the same Jenkins job and build, but creates a new JJI job ID so the original result stays intact.

Warning: jji re-analyze reruns with the original settings. If you need to change timeout, model, peer review, repository context, Jira search, or artifact collection, use the web UI's Re-Analyze button instead.

Step-by-Step

1. Queue the analysis and keep the job ID.

bash jji analyze --job-name "folder/job-name" --build-number 123 --provider claude --model "<model>"

The response is immediately queued and includes a poll link you can open in the browser.

Note: queued is the submit response. Once the run is stored, you will normally see pending or waiting, then running, and finally completed or failed.

1. Monitor the run from the CLI or the Dashboard.

bash jji status <job-id> jji results dashboard

Use jji status when you care about one run. Use jji results dashboard or the web Dashboard when you want a live overview of multiple jobs, including failure counts, review progress, comments, and child-job counts.

Tip: The Dashboard and status page refresh automatically, so they are the easiest way to follow a live run.

1. Open the finished result.

bash jji results show <job-id>

In the browser, click the row in the Dashboard. Active jobs open the status view; completed jobs open the full report.

See Reviewing, Commenting, and Reclassifying Failures for what to do after the report is ready.

1. Rerun the same analysis quickly.

bash jji re-analyze <job-id>

This is the fastest retry path when you want another pass with the same settings. JJI creates a brand-new analysis record and leaves the original one untouched.

1. Rerun with adjusted settings from the web UI.

2. Open the failed run's status page or the completed run's report page.

3. Click Re-Analyze.
4. Change only the fields you need, such as AI provider or model, AI CLI timeout, raw prompt, peer review, repository context, Jira search, or artifact collection.
5. Submit the rerun and monitor the new job from the Dashboard or the status page.

Tip: The Re-Analyze dialog starts from the original run's settings, so you usually only need to change the field that caused the rerun, such as a higher AI timeout or a different model.

See Improving Analysis with Repository Context for repository, prompt, and artifact guidance. See Adding Peer Review with Multiple AI Models for peer analysis settings.

Advanced Usage

The live status page can show more than a single state. On longer or more complex runs, it can step through waiting for Jenkins, failure analysis, child-job analysis, Jira search, peer review rounds, and the final save step.

Warning: Re-analysis does not change Jenkins wait behavior. If the error says it timed out waiting for Jenkins, submit a new analysis with different wait settings instead of relying on Re-Analyze.

Troubleshooting

• Re-Analyze is missing or unavailable: the job is still active, or the original result does not have reusable saved settings.
• The web UI shows Timed Out, but the CLI shows failed: that is expected. Timed Out is a web UI label for AI-analysis timeouts.
• The run says it timed out waiting for Jenkins: that is a Jenkins wait timeout, not an AI timeout. Start a fresh analysis with different wait settings.
• A rerun did not help: make sure you changed the setting that matches the failure. Raise AI timeout for AI stalls, switch model/provider for weak analysis, or update repository context when the model lacked enough code or artifact evidence.
• The server restarted while a job was live: jobs that were still waiting on Jenkins resume automatically, but jobs already in pending or running are marked failed and should be rerun.

Related Pages

• Analyzing Jenkins Jobs
• Reviewing, Commenting, and Reclassifying Failures
• Improving Analysis with Repository Context
• Adding Peer Review with Multiple AI Models
• Running Your First Analysis