+1 Maintenance Report

[Christian Ehrhardt]

This week Steve was also on +1 duty, which made me think the massive
AA-needing tasks most likely will happen on his side anyway. Therefore
I focused more on build/test fails. Unfortunately I must admit that
I clearly wasn't able to find as much time for +1 as usually :-/

# openvswitch

This was an armhf test fail [1] and while I wanted to look if it is
reproducible I got a ping from schopin hinting at the right combined
trigger as it was a known systemd issue. I Issues that and it TODO

# libgphobos rebuild making cheesecutter & dub FTBFS

I've seen rebuilds of those failing for essentially compiler errors (different
issues for both packages, but the same across architectures). It pretty much
looked the same way as it does when we build a new toolchain and since
libgphobos is from gcc-* that might very much be what happens.

I tried a local rebuild of those two and they reproduced the issue.
It seems that really are just new toolchain issues.

I quickly found fixes and rebuild to see if there are other issues.
Then I checked if upstreams had new releases/patches in that regard.

For dub that existed upstream [4], but also there is another issue in the
package [5] which would block an upload and soon cause it to be removed [6].
Since the fix is applied in an already released version by upstream I think
this does not need much more attention but time to sync the new builds - it
isn't critical to impish to get the latest version. On the other hand it will
be useful to document this in the open update-excuse bug so that others do
not have to re-investigate this, so I added some info there.

cheesecutter was rather similar, I found that this is a known and expected
change as it is part of a set of deprecations scheduled for removal in 2018 [3].
The page holds hints how to replace this, but actually upstream has already a
commit that fixes the problem [7].

I filed an update-excuse bug [8] and linked it to the related debian bug [9]
as well as sending an FYI about the fix to the Debian bug. Given that this isn't
blocking gcc we could just wait until Debian uploads a new git-snapshot with
the fix included. Also the forwarded upstream issue needed a bump to realize
bug and fix belong together.


# umockdev 0.16.3 breaking bolt autopkgtest

I analyzed this and eventually filed [10] for it. The TL;DR is that a
test wants to created a mocked $tmp/sys/bus but the new umockdev auto-creates
this on init, so the test fails as the dir already exists.

Martin is the upstream and Debian maintainer and has to choose if this shall be
fixed by partially reverting the offending commit or by adapting the test as I'm
unaware of the context that triggered adding the change.

The update-excuse bug will help to avoid re-analysis of the same case, but there
is no urgency to fix this in Impish, we can wait until it resolves in Debian
and sync it then. OTOH If a minimal fix is available early syncing it to
impish would be nice, a lot will depend on the reply on those bugs.

# mdbtools FTFBS on ppc64

This seemed to hang around so I fetched a ppc64 machine from maas and found it
to be reproducible. I found a fix and upstreamed it via [12] as it was just
a new gcc warning.

LP bug [11] marked update-excuse as long as it wasn't complete and an upload
of the upstream fix eventually resolved it.

# Firewalld autopkgtest fail

This looked interesting, but I failed to find time for it. Nevertheless
there was already decent context due to the answers I got on my ping for
potential prior activity on this. I Filed [13] to give anyone who will have
a deeper look at this a head start.

# Help

Gladly this week Paride was so kind to help me and pick a few more +1 tasks.
That way some more than what I managed to resolve got done. I assume he will
reply to this thread with his own content he worked on this week.
Thanks @paride!

[1]: https://autopkgtest.ubuntu.com/results/autopkgtest-impish/impish/armhf/n/netplan.io/20210927_104834_e5e17@/log.gz
[2]: https://people.canonical.com/~doko/ftbfs-report/test-rebuild-20210805-impish-impish.html
[3]: https://dlang.org/changelog/2.075.0.html#pattern-deprecate
[4]: https://github.com/dlang/dub/commit/aec94916cd856c662386bdea9038f19da27e457b
[5]: https://bugs.launchpad.net/ubuntu/+source/dub/+bug/1915312
[6]: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=981229
[7]: https://github.com/theyamo/CheeseCutter/commit/68d6518f0e6249a2a5d122fc80201578337c1277
[8]: https://bugs.launchpad.net/debian/+source/cheesecutter/+bug/1945212
[9]: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=984009
[10]: https://bugs.launchpad.net/ubuntu/+source/umockdev/+bug/1945321
[11]: https://bugs.launchpad.net/ubuntu/+source/mdbtools/+bug/1945478
[12]: https://github.com/mdbtools/mdbtools/issues/352
[13]: https://bugs.launchpad.net/ubuntu/+source/firewalld/+bug/1945596


-- 
Christian Ehrhardt
Staff Engineer, Ubuntu Server
Canonical Ltd