Symptom: The navigation bar at the top of the web UI is not visible.
Causes:cluster.json is incomplete or malformed; alternatively config.json contains invalid entries.
Check: Validate both JSON files (e.g., with jq) and compare them to a working example, then reload the web service.
No new jobs or metrics
Symptom: Neither jobs nor metrics appear in the backend.
Check: Inspect journalctl for the affected services and look for messages such as Can't decode jwt.
Cause/fix: The JWT (API key) expired. Generate a new token and update all components that use it.
Metric collector does not send metrics
Symptom: Expected metrics are missing from the output or do not reach cc-backend.
Configuration check: Make sure the collector is enabled in collectors.json. Verify in router.json that the metric is neither renamed away nor dropped. If it is a diff/derived metric, the collector must not be run with -once.
Debugging tip: Use a test configuration (config_stdout.json) that references router_stdout.json and sinks_stdout.json. The latter replaces the HTTP sink with stdout. Keep the router configuration small and run the collector via ./cc-metric-collector -config config_stdout.json to inspect the terminal output.
Downloads start while scrolling
Symptom: Scrolling in ClusterCockpit triggers a download of job/node views.
Fix: Disable script blockers (e.g., NoScript) or other content blockers so the web interface can load correctly.
Metrics show the wrong scale
Symptom: Values are converted multiple times and displayed incorrectly (e.g., 0.0031 MB/s instead of 3.1 MB/s).
Fix:change_unit_prefix in the router and unit+prefix in cluster.json must match.