← Back to team overview

nova team mailing list archive

Re: Intermittent test failures


I can't replicate a failed test locally either.  It only seems to fail when run by Tarmac as part of the branch merging (and then, only intermittently).


From: Vishvananda Ishaya [mailto:vishvananda@xxxxxxxxx]
Sent: 13 September 2010 10:51
To: Andrew Shafer
Cc: Ewan Mellor; Nova
Subject: Re: [Nova] Intermittent test failures

I haven't been able to replicate a failed test locally, although it is possible that the issues were inadvertently fixed as part of the orm refactor.  The concurrent volume test is a bit useless at the moment since we're no longer using shell commands to find a free blade and volume, so the test doesn't actually allocate blades concurrently.

The purpose of the test was to make sure that two different volume workers were never allocated the same shelf and blade.


On Sep 12, 2010, at 9:55 PM, Andrew Shafer wrote:

-1 for deleting tests because they are annoying when they fail

If no one knows why they are failing, can anyone know why they pass?

Granted tests can become a technical debt unto themselves, but in practice I prefer test are deleted because the person deleting them understands why they no longer need to exist.

I'm just learning how all this stuff works, but bzr blame says all those tests were written by 'vishvan'.

I'm not 100% positive, but I believe that is Vishvananda. (https://launchpad.net/~vishvananda)

Perhaps we can start a conversation about what those tests were intending to do and why they might fail before they get deleted.

On Sun, Sep 12, 2010 at 10:19 PM, Ewan Mellor <ewan.mellor@xxxxxxxxxxxxx<mailto:ewan.mellor@xxxxxxxxxxxxx>> wrote:
So if no-one knows why they're failing, shall we remove them from the test

Mailing list: https://launchpad.net/~nova
Post to     : nova@xxxxxxxxxxxxxxxxxxx<mailto:nova@xxxxxxxxxxxxxxxxxxx>
Unsubscribe : https://launchpad.net/~nova
More help   : https://help.launchpad.net/ListHelp