Friday triage and all week proposed migration duty

Christian Ehrhardt christian.ehrhardt at canonical.com
Fri May 1 09:00:28 UTC 2020


## Triage

A whopping 41 bugs to look at of which about half where touched due
being already handled or due to Parides Master-Triage yesterday. Of
the rest as usual most just needed a little bump or help, the
following where more interesting.

Strongswan 1875504
An apparmor issue, I pinged Lucas about it who is working on a merge
for Groovy already.

snmp 1875926
This already came up in Parides triage - the bug now states what goes wrong.
I did some further analysis, it is a real issue but with an
interesting twist to "cause itself once when fixed". I marked it
server-next+bitesize as the patch exists and is in >=Disco. But at the
same time I think we should only prep it in bionic-propsosed and block
it there.

## Proposed migration

I was tracking that all week.
We mostly see Groovy building and testing a lot.
In rare cases there might be a flaky test that needed a restart and
people have to be aware that with the current test queue that will
need a few days. Under all these easy cases are a few we need to look
at.

PHP
php7.4 (7.4.3-4ubuntu1 to 7.4.3-4ubuntu2) in proposed for 5 days
Regressions
  php-nesbot-carbon/2.32.2-1: ppc64el
  php-parser/4.2.2-2: armhf
Bryce is already looking after those.

Libcap2
This triggers quite some fails in e.g. asterisk and chrony. But only
on armhf which is instant-suspicious.
I'm still waiting for the re-runs which shall then reveal if this is
an armhf issue with the new libcap2 or was just flaky after all.

six
This just has a zillion python dependencies in universe which not all
are of good quality.
The majority seems to be flaky, on other cases I found that others
restarted the tests already.
So I'm -not- restarting those and let things settle a bit. The queue
is full enough that these don't need to be double-retried.
Most of the fails (all but two) are also armhf only which again, is suspicious.

collectd.
Collectd got a bunch of new versions in Debian and we see other
packages depend on them.
So it would be time to merge the new version early in the cycle to
unblock these.
Gladly all our former Delta was accepted by Debian already, so we can
sync it in.
In that regard one puzzling thing is:
  Unsatisfiable depends: libcollectdclient1 (>= 4.6.0): amd64, arm64,
armhf, ppc64el, riscv64, s390x
Because we already have
  libcollectdclient1 | 5.9.2.g-1ubuntu5 | groovy/universe         |
amd64, arm64, armhf, ppc64el, riscv64, s390x
This might need a deeper look if it isn't auto-resolving over time
(might be an interim state).

-- 
Christian Ehrhardt
Staff Engineer, Ubuntu Server
Canonical Ltd



More information about the ubuntu-server mailing list