The JVM “froze”: HikariCP Thread starvation or clock leap detected
- Platforms
- Paper · Purpur · Velocity · single server
- Verified
- Verified
- Confidence
- Confidence 75%
What it looks like in the log
Any one of these lines is enough. The agent matches them locally, on your server.
logs/latest.log
[WARN] Thread starvation or clock leap detected
Symptoms
- HikariCP reports that its housekeeping thread did not run for longer than expected (housekeeper delta)
- Often precedes lag, watchdog warnings and mass timeouts
Cause
The HikariCP pool's housekeeping thread woke up well behind schedule: the JVM was stalled (long GC pause, CPU starvation, VM freeze/migration) or the system clock jumped forward.
Possible causes
- Long garbage collection pauses due to insufficient heap (see MEM-06)
- CPU starvation: an overloaded host, CPU steal on a VPS, container limits
- Virtual machine freeze/snapshot/migration or a system clock jump (NTP)
How to fixLow risk
- Match the event time against GC metrics (long pauses) and CPU metrics (steal, load) — Tickwise links this to MEM-06/PERF-08.
- With frequent long GC pauses: check
-Xmx(the heap must not sit right at the container limit), use Aikar's flags/G1. - With high CPU steal, contact your host or change plan/node.
- A one-off event after a VM snapshot or a time sync is not a problem.
Translated from the Russian original; log lines are quoted verbatim.