On 04/25/2013 01:35 PM, Jonas Israelsson wrote:
On 25/04/13 17:52, Mike Burns wrote:
Can you provide the output of the following:
du -sh /var/log
df -h /var/log
I sure can, I'll even give you two..
Bellow you find the output with about an hour between, during this the
the log should have been rotated four times, right ?
The /var/log/libvirtd.log grows about a MB/min
# First -- Thu Apr 25 16:33:36 UTC 2013
[root@fat-tony admin]# df -h /var/log/
Filesystem Size Used Avail Use% Mounted on
/dev/mapper/HostVG-Logging 2.0G 374M 1.6G 20% /var/log
[root@fat-tony admin]# du -sh /var/log/
168M /var/log/
[root@fat-tony admin]#
# Second -- Thu Apr 25 17:25:45 UTC 2013
[root@fat-tony admin]# df -h /var/log/
Filesystem Size Used Avail Use% Mounted on
/dev/mapper/HostVG-Logging 2.0G 443M 1.5G 24% /var/log
[root@fat-tony admin]# du -sh /var/log/
231M /var/log/
All rotated logs below are when running logrotate manually.
[root@fat-tony admin]# ll /var/log
-rw-------. 1 root root 144427492 Apr 25 17:30 libvirtd.log
-rw-------. 1 root root 1177724 Apr 25 15:30 libvirtd.log.1.xz
-rw-------. 1 root root 845740 Apr 25 15:02 libvirtd.log.2.xz
-rw-------. 1 root root 81856 Apr 25 14:42 libvirtd.log.3.xz
-rw-------. 1 root root 66488 Apr 25 14:40 libvirtd.log.4.xz
That's what I expected. There is an issue with vdsm that the logs
aren't being cleaned up after log rotate.
If you run lsof /var/log, you'll see something like:
python 20829 root 4r REG 253,10 25471139
280 /var/log/vdsm/vdsm.log.1 (deleted)
This can be resolved by doing the following:
Put the host in maintenance mode (recommended, but possible to skip)
edit /etc/logrotate.d/vdsm to add copytrucate to first entry[1].
restart vdsmd (systemctl restart vdsmd)
Now, the logs shouldn't fill up.
Mike
_______________________________________________
node-devel mailing list
[email protected]
http://lists.ovirt.org/mailman/listinfo/node-devel
