logrotate.d configuration on an amazon linux 2 for awslogs is the following
/var/log/awslogs.log {
missingok
notifempty
size 100M
create 0600 root root
delaycompress
compress
rotate 4
postrotate
systemctl restart awslogsd
endscript
}
From time to time, the log has a tendency to quickly go over a GB (well over the size limit 100M) long before the cron.daily logrotate kicks the process. When I run logrotate manually with -vf flag though, it identifies the need to rotate this oversize log and takes action.
I cannot find a clue in messages log on whether or not the daily job errors out while trying to rotate it. I understand that post 3.8.1, this scenario should have been supported but it does not appear so. https://serverfault.com/questions/391538/logrotate-daily-and-size#434645
Any suggestions on how to troubleshoot why the log is not rotating when size hits the limit put in here?
I actually was using the wrong parameter. I should have used maxsize instead of size. This good answer