ShirleyFei
commented
1 year ago After defining "Restart=on-failure", there exists a bad case: dmesg will frequently report logs like "atop invoked oom-killer" once atop exceeds the defined "MemoryLimit=xxM", which may output to syslogs and then occupy quite a lot of disk spaces.
Thus close this PR to avoid the bad case.
In case atop crashes early the day, automatically restart atop.service to continue to record the following hours' logs, instead of relying on the daily atop-rotate.timer to restart next morning.