If you are using lsyncd to keep a hot copy of some directories:
/proc/sys/fs/inotify/max_user_watches >= The number of files (and guesstimated future file count)
Log rotate (if you are not using init scripts):
modify your log file name and binary location and inotify max files tuning.
Kill seems harsh, but lsyncd only understands kill and hup currently. If you do a TERM, it logs an error to the log file.
If you do a HUP (with the version I am running, it just silently dies off).
Only Hup and Kill
/var/log/lsyncd {
daily
missingok
rotate 5
compress
delaycompress
notifempty
create 640 root adm
sharedscripts
postrotate
/usr/bin/killall -s KILL lsyncd
/bin/echo 1000000 > /proc/sys/fs/inotify/max_user_watches
/usr/local/bin/lsyncd
endscript
}
To force the logrotate in future:
logrotate -f /etc/logrotate.d/lsyncd.conf
Consider using:
su user group
if you are running and writing logs as non root user.
assuming that is where you put it for a daily run.