Thomas Goirand <z...@debian.org> writes:
> Using puppet-master from both Stretch and Buster in production, I have found
> out that on each puppet run, a run report is saved under:
>
> /var/lib/puppet/reports/<HOSTNAME>/<DATE>.yaml
>
> unfortunately, with a moderatly sized cluster (about 30 nodes), this fills-up
> very fast. For me, the reports folder was 52 GBytes, leading to a disk full
> (as my puppet-master was running on a small-ish VM). Of course, once the
> disk is full, absolutely nothing works anymore (new nodes get their certs
> saved as zero bytes, etc.).
>
> What should be done, is clean old reports, let's say those that are at least
> one month old, using a cron.daily job.
Some people want these reports to continue to accumulate, some do not,
this is what I do:
class puppet::master::cleanup_reports {
# clean up reports older than $puppetmaster_cleanup_reports days
file { '/etc/cron.daily/puppet_reports_cleanup':
content => "#!/bin/bash\nfind ${puppet::master::reports_dir} -maxdepth 2
-type f -ctime +${puppet::master::cleanup_reports} -exec rm {} \\;\n",
owner => root,
group => 0,
mode => '0700';
}
--
micah
