i have a question about kibana. so i have a VM that installed kibana and elastic there. then at some point, my elastic is experience OOM, so it produce hprof file. but bufore i knew that my elastic was OOM, i found that kibana service is trying to restart but got stuck. My guess for now is that Kibana will try to restart if within a certain time he can't connect to elastic, but, is it true? is there any parameter i can set to make kibana wait longer?
Here is the log from /var/log/message that shows elastic was running out of memory
after that, elastic service was restarted twice by itself
Kibana itself is not designed to restart if it can't connect to Elasticsearch. It will log errors due to background tasks not being able to run their searches, and the the "state" of the Kibana process will be set to "degraded". I'm not an expert on systemd, but perhaps the Kibana service has its own health check that tries to restart the service when the status of the process is unhealthy.
I would look into tuning the restart controls in the kibana.service file. The template for this file is in our Github repo here: https://github.com/elastic/kibana/blob/8.8/src/dev/build/tasks/os_packages/service_templates/systemd/usr/lib/systemd/system/kibana.service#L1
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.