[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [xen-4.8-testing test] 124100: regressions - FAIL

>>> On 13.06.18 at 13:18, <ian.jackson@xxxxxxxxxx> wrote:
> Jan asked me to investigate why we weren't getting a push on Xen 4.8.
> I investigated the failures in this osstest report.
> I think we have a combination of:
>  * Flaky armhf hardware
>  * A real Xen-related heisenbug bug (but which is not a regression)
>    (the EFAULT libxc/linux bug; CC to Juergen)
>  * Mystery failures to make progress during local computation
>    and I/O which look like Linux kernel bugs
>  * Incompatibility between Xen 4.8 and osstest's approach to UEFI
>    booting, now fixed.
>  * A mystery libvirt heisenbug.  (Hence the CC to Jim.)

Thanks a lot for the analysis!

> Jan: I would be inclined to force push this.  OTOH, if we wait,
> eventually osstest will collect a set of flights which osstest's
> archeaologist can see justifies a push.


Last test of basis   123091  2018-05-23 07:11:28 Z   20 days
Failing since        123345  2018-05-29 08:36:34 Z   14 days   13 attempts
Testing same since   123492  2018-05-31 20:14:51 Z   12 days   11 attempts

I'd favor a sufficiently justified (as it is now) force push, and then
an immediate release. As you say elsewhere, the problem with in
particular the albanas has been bad enough for it to be sufficiently
unpredictable when a normal push might happen (in fact I was
surprised by how quickly this happened on 4.7, as I had also
expressed in a reply to that flight's report).

For the release you'd need to tag qemu-trad and (as was asked by
Wei iirc on irc) mini-os, such that I could then push the version
update on the main tree.


Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.