Re: [libvirt test] 36071: tolerable all pass - PUSHED

[Jim Fehlig <jfehlig@suse.com>]

Ian Campbell wrote:
> On Tue, 2015-03-10 at 17:59 +0000, xen.org wrote:
>   
>> flight 36071 libvirt real [real]
>> http://www.chiark.greenend.org.uk/~xensrcts/logs/36071/
>>
>> Failures :-/ but no regressions.
>>
>> Tests which did not succeed, but are not blocking:
>>  test-armhf-armhf-libvirt     10 migrate-support-check        fail   never pass
>>  test-amd64-amd64-libvirt     10 migrate-support-check        fail   never pass
>>  test-amd64-i386-libvirt      10 migrate-support-check        fail   never pass
>>
>> version targeted for testing:
>>  libvirt              b39b1397ea6e6155b5e363d456196504093edd07
>> baseline version:
>>  libvirt              719cd2182bf06196b26204a8cf88d28001b1d79b
>>     
>
> This range contains a trio of libxl fixes:
> $ git log --oneline 719cd2182bf06196b26204a8cf88d28001b1d79b..b39b1397ea6e6155b5e363d456196504093edd07 | grep libxl
> 3b7f589 libxl: remove unneeded cleanup_unlock label
> b1d159d libxl: use libxl_ctx passed to libxlConsoleCallback
> fc3ef44 libxl: remove redundant calls to libxl_evdisable_domain_death
>
> Were any of them expected to resolve the libvirtd disappears issue which
> we've been suffering lately or is this just a lucky pass?
>   

I think it was just a lucky pass.  Fixes for the assertion are two part:
libxl patches from Ian J. to prevent looping in
domain_death_xswatch_callback and a series of libvirt patches that
switch to using a single libxl_ctx.  I'm waiting for Ian to post the
libxl fixes before posting another version of the libvirt fixes.  I'd
like to reference the libxl patches in the libvirt series.

> I suppose we should monitor things for a few flights and then consider
> removing the all.allow entry which makes libvirt failures non-blocking
> in all other branches.
>   

I'd expect random failures until this work finds its way to the
respective git repos.

Regards,
Jim