Since we no longer log qemu monitor output to the /var/log/libvirt/qemu logfile, there isn't much chance of these log files becoming huge in short periods of time. Using aggressive logrotate settings then provides little benefit at the cost of debug ability (for example, bug 524508c#12 ).
I think weekly, biweekly, or even monthly rotation should be enough, maybe removing compression altogether.
I would really keep compression in any case, as this leads to no loss of
Now how much information we keep by default is a trade-off, honestly
keeping them for a week should be sufficient to diagnose problem, but
I would be okay to switch that to weekly and rotate 4 to keep a month
worth of data instead of a week.
this would also limit the number of entries in that directory, cut this
nearly by half for the same amount of VM, currently there is 8 files
per domain, usually of zero lenght if inactive, and we would lower this
to 5 files but bigger.
Yeah, this has been annoying me lately too
Let's switch to weekly
I posted the trivial patch upstream
Okay fixed upstream, it will reach rawhide in a couple of weeks I think
We should pull this into F-12 before GA
* Tue Oct 6 2009 Mark McLoughlin <firstname.lastname@example.org> - 0.7.1-9
- Change logrotate config to weekly (#526769)
tag request : https://fedorahosted.org/rel-eng/ticket/2354