CodeshipKohsuke Kawaguchi3 min readintermediate
Health Check-up for Your Jenkins
Summary
Kohsuke Kawaguchi outlines a few low‑effort ways to keep a Jenkins instance healthy: use jconsole to watch old‑gen heap usage, the Monitoring plugin (or Nagios) to record HTTP latency and queue length, inspect thread dumps for slow pages, and review the built‑in load chart to spot agent under‑/over‑utilisation.
- Run jconsole against the Jenkins JVM and watch the “PS Old Gen” used vs max; stay below ~90 % to avoid UI lag and OOM.
- Install the Monitoring plugin (or external Nagios checks) to capture per‑page request latency and queue size over time.
- When a page is slow, pull a thread dump (Jenkins wiki link) and look for hot threads that point to the offending code.
- Use the “Manage Jenkins → Load Statistics” chart: blue = total agent capacity, red = busy agents, gray = queue length. Mismatches reveal idle agents or insufficient capacity.
Jenkins is often a long‑running service; without periodic metrics you can miss memory bloat, latency regressions, or mis‑sized agent pools, which later manifest as flaky builds or exhausted resources.
4/10