Christian Gennerat wrote:
>
> Thierry SAURA a écrit :
>
> > I have a problem with lograte and two rules :
> > /var/log/mail/* {
> > rotate 5
> > weekly
> > postrotate
> > /usr/bin/killall -HUP syslogd
> > endscript
> > }
> >
> > /var/log/news/* {
> > rotate 5
> > weekly
> > postrotate
> > /usr/bin/killall -HUP syslogd
> > endscript
> > }
> >
> > For exemple, in mail directory, you have before the rotation :
> > mail.err mail.info mail.warn
> >
> > after the first rotation :
> > mail.err mail.info mail.warn
> > mail.err.1.gz mail.info.1.gz mail.warn.1.gz
> >
> > after the seconf rotation :
> > mail.err mail.info mail.warn
> > mail.err.1.gz mail.info.1.gz mail.warn.1.gz
> > mail.err.2.gz mail.info.2.gz mail.warn.2.gz
> > mail.err.1.gz.1.gz mail.info.1.gz.1.gz mail.warn.1.gz.1.gz
> >
> > I let you imagine how many files are created after several rotations.
> > I rewrite rules to not have the problem ...
> >
> > somebody has an idea about where the problem comes from ? bad rules ?
> > logrotate bug ?
> >
> > Before discovering the problem, slocate and logrotate need time to process
> > (more than 120 000 file in /var/log/news and /var/log/mail) and i loose
> > 400 Mo of disk space.
> >
The problem is not in logrotate (or at least not 100% to logrotate).
For 7.1 it can be fixed upgrading syslog to the version in updates/7.1, where
rules have been rewritten to avoid gobble '*' which can confuse logrotate with .gz
extension. 7.2 is not affected.
Bye.
Giuseppe.