Race condition running PreciseHWETTestBonding integration test

Bug #1549904 reported by Diogo Matsubara
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
curtin
Confirmed
Medium
Unassigned

Bug Description

As seen in https://server-team-jenkins.canonical.com/job/curtin-vmtest-venonat/110/console, PreciseHWETTestBonding failed while booting up the just installed VM due to a race condition.

The error presented in the boot-log[1] was: udevd[534]: error changing net interface name rename5 to eth0: File exists

[1] https://server-team-jenkins.canonical.com/job/curtin-vmtest-venonat/110/artifact/output/PreciseHWETTestBonding/logs/boot-serial.log/*view*/

Revision history for this message
Scott Moser (smoser) wrote :
Revision history for this message
Scott Moser (smoser) wrote :
Changed in curtin:
status: New → Confirmed
importance: Undecided → Medium
Revision history for this message
Ryan Harper (raharper) wrote :

http://askubuntu.com/questions/382354/second-network-interface-fails-to-start-pre-start-killed-by-term

Looks like we get some fun races with ifupdown/ifenslave and nic renaming;

Not clear to me if part of the issue may be related to the "delay" involved with if-up.d/* scripts;
In some cases, I see module load errors when iscsi is restarted.

Temporarily, I'm going to see if I can fake out the open-iscsi restart by touching the initramfs file for open-iscsi that says we're root on iscsi which prevents the daemon from restarting.

Revision history for this message
Ryan Harper (raharper) wrote : Re: [Bug 1549904] Re: Race condition running PreciseHWETTestBonding integration test

A quick test of removing /etc/network/if-{up,down}.d/open-iscsi symlinks
post-install resolves the race.
That does mean if we're *not* using iscsi-root but the package is installed
and we use bonding; the race exists for users.

On Mon, Mar 7, 2016 at 12:45 PM, Ryan Harper <email address hidden>
wrote:

> http://askubuntu.com/questions/382354/second-network-interface-fails-to-
> start-pre-start-killed-by-term
>
> Looks like we get some fun races with ifupdown/ifenslave and nic
> renaming;
>
> Not clear to me if part of the issue may be related to the "delay"
> involved with if-up.d/* scripts;
> In some cases, I see module load errors when iscsi is restarted.
>
> Temporarily, I'm going to see if I can fake out the open-iscsi restart
> by touching the initramfs file for open-iscsi that says we're root on
> iscsi which prevents the daemon from restarting.
>
> --
> You received this bug notification because you are subscribed to curtin.
> Matching subscriptions: curtin-bugs-all
> https://bugs.launchpad.net/bugs/1549904
>
> Title:
> Race condition running PreciseHWETTestBonding integration test
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/curtin/+bug/1549904/+subscriptions
>

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.