Merge lp:~xnox/britney/hints-ubuntu into lp:~ubuntu-release/britney/hints-ubuntu

Proposed by Dimitri John Ledkov on 2017-11-16
Status: Rejected
Rejected by: Steve Langasek on 2017-11-20
Proposed branch: lp:~xnox/britney/hints-ubuntu
Merge into: lp:~ubuntu-release/britney/hints-ubuntu
Diff against target: 22 lines (+14/-0)
1 file modified
vorlon (+14/-0)
To merge this branch: bzr merge lp:~xnox/britney/hints-ubuntu
Reviewer Review Type Date Requested Status
Dimitri John Ledkov (community) Disapprove on 2017-11-20
Steve Langasek 2017-11-16 Needs Fixing on 2017-11-16
Ubuntu Release Team 2017-11-16 Pending
Review via email: mp+333813@code.launchpad.net
To post a comment you must log in.
lp:~xnox/britney/hints-ubuntu updated on 2017-11-16
2685. By Dimitri John Ledkov on 2017-11-16

Force skiptest systemd.

2686. By Dimitri John Ledkov on 2017-11-16

Force skiptest util-linux.

Steve Langasek (vorlon) wrote :

Repeating here my questions from IRC:

review: Needs Information
Steve Langasek (vorlon) wrote :

I have confirmed with the Server Team that nested KVM is not supported. So a fix is needed for this assumption in the systemd autopkgtests.

review: Needs Fixing
Dimitri John Ledkov (xnox) wrote :

> I have confirmed with the Server Team that nested KVM is not supported. So a
> fix is needed for this assumption in the systemd autopkgtests.

The tests do detection and use qemu, with kvm, if available. It appears that bionic cloud images are shipping qemu with nested kvm enabled out of the box. If you don't want systemd ADT to use that, we should not be shipping qemu with nested kvm enabled by default.

> <slangasek> xnox: hmm, I was unaware systemd was doing nested KVM in its autopkgtests;

So, systemd assembles initrd/disk-images that e.g. test various things that might not boot at all. For example, unified cgroups only boot, racy boots, dependency loops, etc. And then inspect the kvm logs to see that things did the expected thing.

If these cannot be run without nested kvm (will not be possible on s390x for example), I guess i can move to run these off a MAAS provisioned bare metal instance.

> <slangasek> xnox: your mp doesn't say why the kde4libs autopkgtest failure should be ignored
> <slangasek> xnox: and this test doesn't appear to have been particularly racy in the past, and
> the failure when triggered by dbus is a different test failure

it passes the first time around, running locally.

> <slangasek> nut/s390x should just be a badtest

ok

Dimitri John Ledkov (xnox) wrote :

rejecting this one.

review: Disapprove

Unmerged revisions

2686. By Dimitri John Ledkov on 2017-11-16

Force skiptest util-linux.

2685. By Dimitri John Ledkov on 2017-11-16

Force skiptest systemd.

Preview Diff

[H/L] Next/Prev Comment, [J/K] Next/Prev File, [N/P] Next/Prev Hunk
1=== modified file 'vorlon'
2--- vorlon 2017-11-15 08:26:51 +0000
3+++ vorlon 2017-11-16 14:21:13 +0000
4@@ -197,3 +197,17 @@
5
6 # bad test, but better to have this package in xenial and unblock things than leave it on the list
7 force-badtest guessit/0.11.0-2
8+
9+# Remaining test failures are not reproducible in a VM and/or are known infrastructure (kernel) bugs
10+# rerun locally:
11+# conntrack-tools/s390x racy LP: #1732168
12+# nut/s390x
13+# dpdk/i386 LP: #1730717
14+# kde4libs/i386
15+# systemd/amd64 LP: #1682077
16+# all of above passed, with the same proposed/triggers, from a first manual pass
17+force-skiptest systemd/235-2ubuntu3
18+
19+# zfs "regression" because it previously passed in a nova instance
20+# systemd as above / flakey boot-smoke test
21+force-skiptest util-linux/2.30.1-0ubuntu5
22\ No newline at end of file

Subscribers

People subscribed via source and target branches