When a remote agent loses the connectivity with the Bamboo server it tries to reconnect 5 times and then shuts down the wrapper.
The error messages are not easy to read in the logs and includes stack traces with 502 errors caused by the agent failed attempts to get the fingerprint.
Server outages are expected and should be shown in the logs with graceful messages, not recurrent stack traces with errors that might lead to time wasted investigations.
When restarting the JVM, the agent should check if the Bamboo server is online and if no response is received it should clearly states in the logs that the Bamboo server didn't answer and could be offline. The message should not include unnecessary stack traces.
Both linux and windows behaviors can be seen below in the comments
- Bamboo agents takes about 10min to be completely down. This KB article explains how to change this value to 2h: Remote agent does not restart after server outage - Atlassian Documentation
- This BAM is requesting this change to be permanent: [BAM-18608] Increase the default time taken for the agents to reconnect in case the server is temporarily down