On 6/7/22 15:10, Yedidyah Bar David wrote:
On Tue, Jul 5, 2022 at 4:59 PM nroach44--- via Users
<users(a)ovirt.org>
wrote:
I'm just wondering if anyone else has seen this, and if it's worth
logging a case?
I've found that this seems to manifest itself as "Stop service"
will succeed in the log, and them some time later the upgrade will
fail. rm -rfv /var/cache/dnf* will resolve it.
Might be a result of
https://bugzilla.redhat.com/2055829 .
Yup, that would make
sense - I've had to do this once or twice for 4.4
over it's whole lifecycle, but suddenly had to do it sooner for 4.5.
Can you check/share the used space you had on /var, /var/tmp,
/var/cache, before and after the upgrade?
Used space was about 80% (of 5G) on /var (which has /var/cache) in all
of the cases that I looked at, pre-intervention. The other /vars were
10% or less utilised the whole time.
Once I cleared it, /var was 5% used, and then filled to 30% once the
upgrade was done.
https://imgur.com/a/kd8eWHd
Anyway, might be worth reporting this as a bug. If so, I think the
fix
would be to check for enough free space on /var/cache early in the
upgrade process, failing if not enough with a suitable message. Makes
sense to you? Do you have some other suggestion?
I think there's two improvements, one is to clear the dnf cache or at
least check it, like you say, and the other would be to log the activity
that fails. Currently the activity log shows the last successful task,
and then will just report a failure, making it seem like it was the last
task that failed.
https://blog.nroach44.id.au/images/onn-upgrade-fail-stop-service.png
I'll log both later today.
Thanks and best regards,
--
Didi
--
*Nathaniel Roach*