An integration restarts in a loop
Weik.io 2026.2The agent keeps restarting the integration's Java process after each exit. The cause is most often a missing value or a process out of memory.
| Kind | Severity | Who sees it |
|---|---|---|
deployment- |
Error | Everyone who can read the deployment |
Its row in the problem list has a figure such as 6 restarts in 40 min, out of memory.
What it means#
Each time the integration’s Java process stops, its agent starts it again. While the loop goes on, the integration’s routes run only for moments at a time.
What raises it#
The agent reports a copy as crashed when its process exits after it was running, and starts it again after a wait that grows with each exit. The figure counts the copy’s crashes and failed starts in a row. The count starts over once the copy has run for 10 minutes without a break. When the process ran out of Java heap, the figure ends in out of memory. The problem’s detail has the agent and what it last reported.
What to check#
Read the last log lines before each exit:
$ weikio logs <project>/<integration> --env <environment> --since 1h --level warnSee each copy’s restarts and last exit:
$ weikio describe deployment <project>/<integration> --env <environment>List every deployment of the environment that is crashing:
$ weikio get fleet --env <environment> --state crashedHow to fix it#
The log usually shows the cause:
- A variable or secret that the route needs and doesn’t get
- An endpoint that fails at start
- A class missing from the dependencies
Fix the cause in the integration’s files or values and publish a new version, or roll back.
When the figure says out of memory, give the process more heap with -Xmx in the integration’s jvm.options in integration.yaml. You can also set -Xmx for every deployment of an environment:
$ weikio env set <environment> --launch-jvm-option -Xmx1g