We have recently run into a problem with too much log data for our CI logs. Not surprising since we still have logs from early 2019. So I guess we don't have a log retention policy. I was thinking 60 days? Too little? Too much?
The consensus of the system roles team is * 6 month retention - roughly corresponding to fedora release cycle * keep the last logs for a role if no logs are generated e.g. if there is no update to the kdump role for more than 6 months, still keep the last logs that were generated, even if older than 6 months
comments, concerns?
On Sun, Nov 7, 2021 at 2:57 PM Richard Megginson rmeggins@redhat.com wrote:
We have recently run into a problem with too much log data for our CI logs. Not surprising since we still have logs from early 2019. So I guess we don't have a log retention policy. I was thinking 60 days? Too little? Too much?
Hi,
Am Mo., 8. Nov. 2021 um 18:57 Uhr schrieb Richard Megginson < rmeggins@redhat.com>:
The consensus of the system roles team is
- 6 month retention - roughly corresponding to fedora release cycle
- keep the last logs for a role if no logs are generated e.g. if there is
no update to the kdump role for more than 6 months, still keep the last logs that were generated, even if older than 6 months
comments, concerns?
sounds good. If you have a way to have these complex criteria, it might also be good to restrict the amount of used space, so
1) more logs would be stored as long as there is enough storage 2) if there is not enough storage but above rules don't apply, still cleanup something
Cheers Till
systemroles@lists.fedorahosted.org