checkbox-dev team mailing list archive

[Daniel Manrique <daniel.manrique@xxxxxxxxxxxxx>]

On Tue, Sep 30, 2014 at 2:46 PM, Zygmunt Krynicki
<zygmunt.krynicki@xxxxxxxxxxxxx> wrote:
> Hey
>
> TL;DR; Please upgrade to plainbox 0.13.1 or plainbox 0.14 snapshot
> ASAP as plainbox 0.13 is critically broken.

Thanks for sending out this notice.

I forwarded it to another mailing list that needed to know (ask me
later), and this is a heads-up that plainbox 0.13.1 (stable) and
0.14.dev (dev snapshot) with the fixes for these critical bugs are now
available in the corresponding PPAs.

We decided to cherry-pick the fix directly to stable (0.13.1) since it
was broken anyway; the cherry-picked patches don't touch anything else
so should be pretty safe.

Apologies for the inconvenience.

- Daniel

>
> Earlier today we have started investigating two issues:
>
>  * https://bugs.launchpad.net/plainbox/+bug/1375158
>  * https://bugs.launchpad.net/plainbox/+bug/1375724
>
> Both bugs are critical and crippling. Both are recent regressions
> introduced during the 0.13 release cycle. The first of the two is now
> fixed. This is the bug that both Tai and Sam experienced today. The
> good news is that this bug is now gone. The bad news is that this bug
> is destroying the serialized session state thus making recovery
> impossible. Affected machines need to be re-tested. One cannot just
> 'plainbox session export' the logs again.
>
> The second bug is less serious but equally crippling. It's a follow up
> to the recent XML control characters issue that we've fixed before
> (twice, this is the second bug and third iteration of this patch).
>
> I've included *all* of the discussion about those bugs so that anyone
> following can cross-check my logic and verify that the bug is
> correctly identified and the discovered issue is fixed.
>
> Daniel will release an emergency 0.13.1 release and publish it in the
> stable PPA. The release will be made from plainbox-v0.13 with the two
> fixes cherry-picked (they are both single patches so this is easy).
> The relevant branches are linked to the bug reports.
>
> As for what we've learned: I would like to amend the release candidate
> testing procedure to include a test where we resume a newly created,
> suspended session. We can even work on automating that via a
> fully-automatic test case (a shell script anyone can run). Performing
> this test would have caught one of the issues and would have prevented
> us from having to re-test machines as the second bug can be simply
> fixed in the field allowing existing test results to be recovered.
>
> If you have any questions about this feel free to contact us directly
> either in the internal #Cert channel, on #checkbox or via email.
>
> NOTE: to those that wish to reply on one of the mailing lists, please
> note that checkbox-dev@xxxxxxxxxxxxxxxxxxx is a public mailing list.
>
> Best regards
> Zygmunt Krynicki
>
> --
> Mailing list: https://launchpad.net/~checkbox-dev
> Post to     : checkbox-dev@xxxxxxxxxxxxxxxxxxx
> Unsubscribe : https://launchpad.net/~checkbox-dev
> More help   : https://help.launchpad.net/ListHelp