Conditionally restarting systemd services

In a previous post I discussed how one of my systemd services was getting continuously restarted causing the CPU to spike. This isn’t ideal and after re-reading the systemd manual page I came across a couple of useful options to control when and how frequently a systemd service will restart. The first option is RestartSec which controls how long systemd will wait to restart a process after a failure occurs. Systemd also has the RestartForceExitStatus and RestartPreventExitStatus which allow you to define the signals that should or should not cause a restart. To ensure that logstash isn’t continuously restarted when a configuration issue is present I set RestartPreventExitStatus to 1. Abnormal signals like SIGSEGV, SIGABRT, etc. will trigger a restart but a return code of 1 will not.

This article was posted by Matty on 2017-09-03 08:55:00 -0400 EDT