Author: Christian Ehrhardt 
Revision Date: 2020-10-06 05:43:51 UTC

badtest openjdk-lts which breaks the test environment

Already quite a lot of openjdk-* packages are marked on i386
$ grep -Hrn openjdk *
ubuntu-release:969:force-badtest openjdk-13/all/i386
ubuntu-release:970:force-badtest openjdk-14/all/i386
ubuntu-release:971:force-badtest openjdk-8/all/i386
ubuntu-release:10489:force-badtest openjdk-lts/all/armhf
ubuntu-release:10542:force-badtest openjdk-lts/blacklisted \
   openjdk-13/blacklisted openjdk-14/blacklisted

But openjdk-lts has the same issues, none of the recent runs worked:
=> https://autopkgtest.ubuntu.com/packages/o/openjdk-lts/groovy/i386
=> https://autopkgtest.ubuntu.com/packages/o/openjdk-lts/focal/i386

Generally failing until timing out and then also breaking the test
infrastructure with errors like:
autopkgtest [21:00:43]: ERROR: "rm -rf /tmp/autopkgtest.QhyXlk/jdk-artifacts
  /tmp/autopkgtest.QhyXlk/autopkgtest_tmp" failed with stderr "rm: cannot
  remove '/tmp/autopkgtest.QhyXlk/autopkgtest_tmp/jdk/JTwork/classes/2':
  Directory not empty

Currently holding back nss (FTFBS fix before groovy release) and zlib
(important feat of s390x compression).

Even when the timeout and infrastructure problems would be solved these
tests don't work well on any architecture:
- armhf -> ignored
- amd64/ppc64el/s390x -> never worked
- i386 -> all runs have the same ~200 fails before crashing the infra

Author: Christian Ehrhardt 
Revision Date: 2020-09-16 06:11:50 UTC

force-badtest systemd (LP: #1892358)

Regression due to test env change made the systemd-fsckd way too flaky.
As discussed in (LP: #1892358) it has to be marked flaky. Mark it as badtest
until an SRU does so.

Non x86 arches seem not to hit this in Bionic, so they are not marked.
Last 20 test results on Bionic:
bionic
  amd64
    boot-and-services (F 5% f 0% S 0% B 5% => P 90%/) ..............B..F..
    boot-smoke (F 10% f 0% S 0% B 5% => P 85%/) .........F..F.B.....
    upstream (F 5% f 0% S 0% B 5% => P 90%/) ..............B..F..
    systemd-fsckd (F 70% f 0% S 0% B 5% => P 25%/) FFF..F.F.FFFFFBF.FFF
    root-unittests (F 5% f 0% S 0% B 5% => P 90%/) ..............B..F..
  armhf
    no failures
  ppc64el
    upstream (F 5% f 0% S 0% B 0% => P 95%/) ...........F........
    systemd-fsckd (F 10% f 0% S 0% B 0% => P 90%/) ..............F...F.
  s390x
    boot-and-services (F 5% f 0% S 0% B 0% => P 95%/) .F..................
  i386
    boot-smoke (F 5% f 0% S 0% B 5% => P 90%/) ...F..B.............
    upstream (F 15% f 0% S 0% B 5% => P 80%/) ..F...B.....F.....F.
    systemd-fsckd (F 60% f 0% S 0% B 5% => P 35%/) .F.FFFBFFFFFF.F..F..

Author: Christian Ehrhardt 
Revision Date: 2020-08-31 06:00:07 UTC

subtest systemd-fsckd broken (LP: #1892358) reset test to unblock SRUs

Author: Christian Ehrhardt 
Revision Date: 2020-09-02 07:47:34 UTC

reset the vagrant-libvirt version in -release

Author: Christian Ehrhardt 
Revision Date: 2020-08-31 06:00:01 UTC

subtest systemd-fsckd broken (LP: #1892358) reset test to unblock SRUs

Author: Christian Ehrhardt 
Revision Date: 2020-08-20 08:49:22 UTC

reset the never working mmdebstrap on armhf

Author: Christian Ehrhardt 
Revision Date: 2020-08-13 14:04:07 UTC

qflow - not aavialable on i386

qflow isn't important to the remaining i386 story but tests break
on install dependency issues. Mask the i386 test result.

Author: Christian Ehrhardt 
Revision Date: 2020-08-12 14:55:19 UTC

remove ntopng unable to build in Groovy due to versioned Build-Dependencies

Note: ntopng also need to be removed from groovy when this is accepted.

Author: Christian Ehrhardt 
Revision Date: 2020-08-12 14:55:19 UTC

remove ntopng unable to build in Groovy due to versioned Build-Dependencies

Note: ntopng also need to be removed from groovy when this is accepted.

Author: Christian Ehrhardt 
Revision Date: 2020-06-08 07:24:07 UTC

stop auto-syncing php-horde* (LP: #1868281)

Author: Christian Ehrhardt 
Revision Date: 2020-04-17 07:23:44 UTC

also reset mysql-connector-c++ on arm64/ppc64

This shall later allow to drop the versioned hints.
Since I had a few occasions where just "reset" wasn't enough lets add them
now and drop the versioned hint later on a cleanup run.

Author: Christian Ehrhardt 
Revision Date: 2020-03-18 09:39:46 UTC

Bump the libreoffice/i386 hint.

Author: Christian Ehrhardt 
Revision Date: 2020-03-16 06:24:43 UTC

fix typo in pglogical-ticker

It was correct to add the rule, but would have worked much
better without the typo :-/

Author: Christian Ehrhardt 
Revision Date: 2020-03-13 08:32:11 UTC

continue the pglogical journey in regard to PG-11/12 switch

This is further continuing to bring back thins removed by LP: #1862601
now that compatible code is available.

I yesterday added force-reset-test pglogical/2.3.0-1, but it didn't work.
All but ppc64 were fine and considered it "always failed" as intended.
But on ppc64 that didn't work out.
On http://autopkgtest.ubuntu.com/packages/p/pglogical/focal/ppc64el we have
the following even on retry.
2.3.0-1 pglogical/2.3.0-1 2020-03-13 07:40:28 UTC 0h 26m 17s paelzer faild
2.3.0-1 pglogical/2.3.0-1 2020-03-12 12:06:38 UTC 0h 27m 55s - fail

So lets switch this back to the force-badtest it always was.

Furthermore there is the dependent pglogical-ticker which also isn't working
since quite some time.
e.g. http://autopkgtest.ubuntu.com/packages/p/pglogical-ticker/focal/amd64
Mark the new version as a known badtest as well.

Note: I've added a team task to revisit the pglogical* tests once things
calm down a bit to make them better in the long run.

Author: Christian Ehrhardt 
Revision Date: 2020-03-12 06:38:35 UTC

bring back pglogical and re-add override

pglogical was removed by [1] for postgresql-11 dependency.
Later on it was dropped here by
  committer: Steve Langasek steve.langasek@canonical.com
  branch nick: hints-ubuntu
  timestamp: Wed 2020-03-04 05:59:36 -0800
  message: drop unused hints

That was correct at the time, but now needs to come back as the
new pglogical can work with postgresql-12 and therefore is synced in now.

Use the new force-reset-test [2] as it might come back working and we'd
want to see that to re-check why which we won't realize if we continuously
just badtest it.

[1]: https://bugs.launchpad.net/ubuntu/+source/pglogical/+bug/1862601
[2]: https://lists.ubuntu.com/archives/ubuntu-release/2020-February/004894.html

Author: Christian Ehrhardt 
Revision Date: 2020-02-26 09:48:33 UTC

make kronosnet badtest unversioned

Multiple i386 test issues present:

1. testname=configure
checking for nss... no
configure: error: Package requirements (nss) were not met:
No package 'nss' found

=> That is essentially an i386 test trying to build with mixed dependencies
   installed.

2. testname=libknet
+ pkg-config --libs libknet
Package libknet was not found in the pkg-config search path.
Perhaps you should add the directory containing `libknet.pc'
to the PKG_CONFIG_PATH environment variable

=> Again a mixed i386/amd64 config/build try failing.

2. testname=libnozzle

+ pkg-config --libs libnozzle
Package libnozzle was not found in the pkg-config search path.
Perhaps you should add the directory containing `libnozzle.pc'
to the PKG_CONFIG_PATH environment variable
No package 'libnozzle' found

=> Again a mixed i386/amd64 config/build try failing.

All reproducible in a local autopkgtest.

E.g. on nss due to being the i386 test we have in that test system:
ubuntu@autopkgtest:/$ dpkg -l '*libnss*' | grep ii
ii libnss3:i386 2:3.49.1-1ubuntu1 i386
ii libnss3-dev:i386 2:3.49.1-1ubuntu1 i386

But the arch is:
$ dpkg --print-architecture
amd64
$ dpkg --print-foreign-architectures
i386

Therefore the configure (and all others) fail.

The following makes the test pass (to then bock on the next dev/lib check)

$ sudo apt install libnss3-dev:amd64 libnss3:amd64 libnspr4-dev
Reading package lists... Done
Building dependency tree
Reading state information... Done
The following additional packages will be installed:
  libnspr4
The following packages will be REMOVED:
  libnspr4-dev:i386 libnss3-dev:i386
The following NEW packages will be installed:
  libnspr4 libnspr4-dev libnss3 libnss3-dev
0 upgraded, 4 newly installed, 2 to remove and 0 not upgraded.
Need to get 1715 kB of archives.
After this operation, 4334 kB of additional disk space will be used.
Do you want to continue? [Y/n] Y
...
$ sudo ./configure --disable-man

But then once we resolve all that way it is identical to the amd64 test.

Note: kronosnet exists for i386 as it is a dependency of corosync which
exists for i386 still.

Therefore overall I think it is ok to disable the i386 test of kronosnet as it
won't work that way in our new i386-reality.
To do so I'd suggest we change the versioned badtest that we already had to a
non versioned one (and moving it to the right section).

Author: Christian Ehrhardt 
Revision Date: 2020-02-07 11:30:57 UTC

Strongswan is not in the i386 whitelist but triggers tests

The issue reproduces in autopkgtest:
$ sudo ~/work/autopkgtest/autopkgtest/runner/autopkgtest --no-built-binaries
--apt-upgrade --apt-pocket=proposed=src:re2c --setup-commands="dpkg
--add-architecture i386; apt-get update" --shell-fail -a i386
strongswan_5.8.2-1ubuntu1.dsc -- qemu --qemu-options='-cpu host'
--ram-size=2048 --cpus 2 ~/work/autopkgtest-focal-amd64.img

But it makes no sense, if you check in there:

ubuntu@autopkgtest:~$ dpkg --print-foreign-architectures
i386
ubuntu@autopkgtest:~$ dpkg --print-architecture
amd64

And listed by
ubuntu@autopkgtest:~$ apt-cache policy strongswan
strongswan:
  Installed: (none)
  Candidate: 5.8.1-1ubuntu1
  Version table:
     5.8.2-1ubuntu1 500
        500 http://archive.ubuntu.com/ubuntu focal-proposed/main amd64 Packages
        500 http://archive.ubuntu.com/ubuntu focal-proposed/main i386 Packages
     5.8.1-1ubuntu1 990
        990 http://archive.ubuntu.com/ubuntu focal/main amd64 Packages
        990 http://archive.ubuntu.com/ubuntu focal/main i386 Packages

But still it can't find/install it:
ubuntu@autopkgtest:~$ sudo apt-get install strongswan:i386
Reading package lists... Done
Building dependency tree
Reading state information... Done
E: Unable to locate package strongswan:i386

So far so good, but why does it then trigger and fail on the i386 test?
http://autopkgtest.ubuntu.com/packages/s/strongswan/focal/i386

Because from d/control:
Package: strongswan
Architecture: all

But it is not just all being Arch:all it also happens to
Package: strongswan-pki
Architecture: any

And per https://lists.ubuntu.com/archives/ubuntu-devel/2019-December/040859.html
it isn't part of the whitelist.
So I'd expect the test to just "not run".

For now ignore it at least on the current version.

Author: Christian Ehrhardt 
Revision Date: 2020-02-05 08:43:10 UTC

re2c is working again @i386

Something changed in Focal or the test-env and now re2c is working fine again.
And I mean not due to the former masking, but really passing all tests.

See http://autopkgtest.ubuntu.com/packages/r/re2c/focal/i386

Therefore we can remove this hint for now to get back the i386 coverage it
could provide.

Author: Christian Ehrhardt 
Revision Date: 2020-02-04 14:57:48 UTC

only restrict 1.3-1 of re2c

Author: Christian Ehrhardt 
Revision Date: 2020-01-30 08:07:47 UTC

add postgresql-12 to the permanent i386 hints

The interdependencies of this can no more be fulfilled on i386. Lets override
the test on this arch.

Author: Christian Ehrhardt 
Revision Date: 2020-01-29 10:47:18 UTC

add universe postgres packages triggering i386 tests on postgresql-common

Author: Christian Ehrhardt 
Revision Date: 2020-01-07 09:55:09 UTC

libseccomp on i386 will pull in python modules breakign the tests

Case #2 of: https://lists.ubuntu.com/archives/ubuntu-devel/2019-December/040859.html

Author: Christian Ehrhardt 
Revision Date: 2019-11-21 08:31:36 UTC

mark python-boto failing (due to py2 deprecation removal of python-httpretty)

Author: Christian Ehrhardt 
Revision Date: 2019-11-15 20:05:20 UTC

spice still non functional on ppc64el, bump version

Author: Christian Ehrhardt 
Revision Date: 2019-09-26 14:48:22 UTC

mark systemd broken on armhf

In the discussions in LP: #1845337 we realized that since recent
changes [1] to the infrastructure new directories are exposed [2].
Due to that this test is broken on LXD which currently means on armhf.

A following systemd upload could fix the test by adding the
directories to the list of filtered paths, but for mask the
bug as known bad test.

[1]: https://git.launchpad.net/autopkgtest-cloud/commit/?id=b8c9165686c7598b3f1a68aa4684e7f382ad935c
[2]: https://bugs.launchpad.net/ubuntu/+source/qemu/+bug/1845337/comments/10

Author: Christian Ehrhardt 
Revision Date: 2019-05-09 06:28:13 UTC

resource-agents is broken on armhf (LP: #1828258)

Author: Christian Ehrhardt 
Revision Date: 2019-04-23 12:15:22 UTC

failing due to a LXD behavior change bleeding into stderr (LP: #1825968)

A fix is already in the xenial SRU queue

Author: Christian Ehrhardt 
Revision Date: 2019-04-23 12:12:10 UTC

failing due to a LXD behavior change bleeding into stderr (LP: #1825968)

FYI: A fix is already in the SRU queue as 0.12.8~16.04.3

Author: Christian Ehrhardt 
Revision Date: 2019-04-23 06:58:21 UTC

add linux-aws-edge that was missing in custom kernel overrides

Author: Christian Ehrhardt 
Revision Date: 2019-04-08 08:37:19 UTC

Use explcit versions for i386 hints

The kernel Team wants to resolve the issues or at least understand
what is going on. For now only mark the former few versions which
block things in proposed.

Author: Christian Ehrhardt 
Revision Date: 2019-04-01 07:32:48 UTC

bump libreoffice i386 to latest version

We need to bump it to libreoffice/1:5.1.6~rc2-0ubuntu1~xenial6/i386
as it seems other updates came in (without that bump which is odd) and the
old issue persists but the force-badtests needs to learn about the new
version as well.

Author: Christian Ehrhardt 
Revision Date: 2019-02-26 13:08:48 UTC

force-badtest systemd autopkgtest in bionic (LP: #1817721)

This is adding a force-badtest for seing 50-100 last tests failing
on all architetcures recently. Like that it is not usable for gating
anything.

Removing the root cause or at least masking the sub-test "upstream"
might help, but I leave that to xnox who is much deeper into the
systemd package.

This also cleans up former outdated systemd references.

Author: Christian Ehrhardt 
Revision Date: 2019-02-21 08:52:13 UTC

mark vagrant-mutate on armhf as force-badtest

Author: Christian Ehrhardt 
Revision Date: 2019-01-30 10:52:39 UTC

autopkgtests regressed and it is a false-positive for now (LP: #1813730)

Author: Christian Ehrhardt 
Revision Date: 2018-12-11 20:25:42 UTC

update for systemd/239-7ubuntu10.5 in cosmic-proposed

Author: Christian Ehrhardt 
Revision Date: 2018-12-03 09:55:23 UTC

systemd: tests are still too unreliable (LP: #1806347)

Author: Christian Ehrhardt 
Revision Date: 2018-09-21 08:21:07 UTC

ocfs2-tools is broken on s390x, upload implies a version bump

Author: Christian Ehrhardt 
Revision Date: 2018-06-19 07:09:17 UTC

New 2.x tests of pgbackrest fail on armhf

We recreated this on a raspbi 3 and we could bump a set of timeouts but that
breaks a lot of later assumptions (e.g. later expected output) and even then
sometimes times out. Since armhf isn't major DB backup place we should mask
it for now just on this architecture to get the features of the new version
available in Ubuntu.

Author: Christian Ehrhardt 
Revision Date: 2018-05-28 06:25:23 UTC

In the past we already covered other perma-fails related to ongoing postgresql MREs like pgfincore, orafce, postgresql-plproxy and pgpool2.

But the following are remaining
  Trusty - armhf on postgresql-9.3
  Xenial - armhf on postgresql-9.5
Those Are known issues that won't resolve and we agreed in the past that on the next non scurity pushed SRU we will add an override (They are fixed post Xenial).

This issue in particular is from switching testbeds from lxc -> lxd (documented by pitti and carried on since then).
This will not resolve, so make it a non versioned hint.

Author: Christian Ehrhardt 
Revision Date: 2018-05-28 10:45:03 UTC

mark current skytools as failing

Author: Christian Ehrhardt 
Revision Date: 2018-05-28 11:12:56 UTC

mark current pg-repack as failing

Author: Christian Ehrhardt 
Revision Date: 2018-05-01 07:18:55 UTC

disable cockpit for regular issues and not being a good test indicator in artful

Author: Christian Ehrhardt 
Revision Date: 2018-04-13 07:47:23 UTC

TEST EDU - superdiff

Author: Christian Ehrhardt 
Revision Date: 2018-04-12 05:51:06 UTC

make landscape-common a recommends to be removable for minimal

Author: Christian Ehrhardt 
Revision Date: 2018-04-10 07:01:10 UTC

cockpit on arm64 never worked in artful

There was only a single one-off good run which makes this block transitions.
See http://autopkgtest.ubuntu.com/packages/c/cockpit/artful/arm64
Also people are encouraged to use th much more recent version in -backports
164-1~ubuntu17.10.1 so let us badtest this.

Author: Christian Ehrhardt 
Revision Date: 2018-03-22 08:29:03 UTC

bump ifupdown for the latest upload, ppc64el in CI still an issue

Author: Christian Ehrhardt 
Revision Date: 2018-03-19 06:50:08 UTC

add landscape-common to server meta (LP: #1756261)

Author: Christian Ehrhardt 
Revision Date: 2018-03-07 17:27:24 UTC

18.04 cleanup: demote euca2tools

Author: Christian Ehrhardt 
Revision Date: 2018-03-07 17:35:25 UTC

18.04 cleanup: drop tomcat-server

Author: Christian Ehrhardt 
Revision Date: 2018-02-14 11:07:26 UTC

mark mimeo as failing on armhf since the infra change from LXC to LXD

Author: Christian Ehrhardt 
Revision Date: 2018-02-14 11:07:26 UTC

mark mimeo as failing on armhf since the infra change from LXC to LXD

Author: Christian Ehrhardt 
Revision Date: 2018-02-09 11:04:03 UTC

kernel still broken, still breaks new versions test

Author: Christian Ehrhardt 
Revision Date: 2018-02-08 10:57:25 UTC

postgres issues due to LXC to LXD should cover all versions (LP: #1748161)

In addition reorganize postgres rules to be in one file (more readable) and
add a comment to related rules (more understandable).

Author: Christian Ehrhardt 
Revision Date: 2018-02-08 10:57:57 UTC

postgres issues due to LXC to LXD should cover all versions (LP: #1748161)

Author: Christian Ehrhardt 
Revision Date: 2018-02-08 08:18:26 UTC

due to oslo libraries, test works but has deprecation notice on stderr (LP: #1748123)

Author: Christian Ehrhardt 
Revision Date: 2018-02-07 10:07:33 UTC

ntpd replaced by chrony >=18.04 (LP: #1744072)

Author: Christian Ehrhardt 
Revision Date: 2018-02-07 10:07:48 UTC

ntpd replaced by chrony >=18.04 (LP: #1744072)

Author: Christian Ehrhardt 
Revision Date: 2018-02-06 08:15:28 UTC

htop is used quite often and was requested to be supported.
MIR is fully ack'ed at LP: #1644364

Author: Christian Ehrhardt 
Revision Date: 2018-02-01 09:03:32 UTC

vagrant-libvirt fails always

It is considered working in the past due to being skipped when s390x was
container only (same for armhf where it still is skipped).

Logs indicate it never really worked, so mask it as bad-test.

Author: Christian Ehrhardt 
Revision Date: 2018-01-29 14:06:17 UTC

389-ds-base make Debian-only network assumptions

This breaks tests which made it block in proposed for quite a while.
Example:
https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-bionic/bionic/amd64/3/389-ds-base/20171221_085018_38cf2@/log.gz

Debian Bug: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=888745

Author: Christian Ehrhardt 
Revision Date: 2018-01-17 12:41:05 UTC

supported-misc-servers: bump to bionic postgres versions

Author: Christian Ehrhardt 
Revision Date: 2018-01-15 05:53:21 UTC

demote xinetd per (LP: #1738062)

Author: Christian Ehrhardt 
Revision Date: 2018-01-15 05:54:37 UTC

no more ship xinetd directly per (LP: #1738062)

Author: Christian Ehrhardt 
Revision Date: 2018-01-03 17:08:29 UTC

duped bugs for openvswitch issue, update comment

Author: Christian Ehrhardt 
Revision Date: 2018-01-03 15:50:57 UTC

bump version of openvswitch as infra is still broken

Author: Christian Ehrhardt 
Revision Date: 2017-07-10 09:14:12 UTC

Improve wording on KVM section

One of the arguments to drop the task was that there is no "the" virtualization
stack anymore. Therefore name the section KVM explicitly.

Author: Christian Ehrhardt 
Revision Date: 2017-07-10 09:07:12 UTC

Fully drop virt-host

Remove all references from STRUCTURE and the remaining file content.
The holding in main is replaced by an entry in
platform.artful/supported-misc-servers

Author: Christian Ehrhardt 
Revision Date: 2017-03-15 08:27:35 UTC

Enable Tri-State lights in qemu tests

Author: Christian Ehrhardt 
Revision Date: 2016-06-01 10:40:09 UTC

fix issue with recent partx requiring --nr to specify the partition

Author: Christian Ehrhardt 
Revision Date: 2016-05-25 11:59:41 UTC

initial vmtest for apt_source feature

Author: Christian Ehrhardt 
Revision Date: 2016-05-10 09:09:25 UTC

mention auto-install of test dependencies in HACKING.rst

Author: Christian Ehrhardt 
Revision Date: 2016-02-23 08:21:05 UTC

Fixup PsuedoImageStore regarding HWE Tests

Author: Christian Ehrhardt 
Revision Date: 2015-11-16 12:19:19 UTC

Avoid false positive errors as logger might not be defined when
calling remove_tmpdir

Author: Christian Ehrhardt 
Revision Date: 2015-11-10 07:34:36 UTC

fix fallback for non hwe cases now that we are specifying subarch instead of
krel

Author: Christian Ehrhardt 
Revision Date: 2015-11-10 07:34:36 UTC

fix fallback for non hwe cases now that we are specifying subarch instead of
krel

Author: Christian Ehrhardt 
Revision Date: 2015-10-13 19:11:37 UTC

Install certain dependencies to support more filesystem types

Author: Christian Ehrhardt 
Revision Date: 2015-09-30 13:51:11 UTC

* debian/patches/99-bash-make-kill-more-procops-compatible.patch:
  - Add -L to kill builtin to be more compatible to procps version of kill