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

[Xen-devel] blkback shutdown problem ? (Re: [linux-4.9 test] 110513: regressions - FAIL)



osstest service owner writes ("[linux-4.9 test] 110513: regressions - FAIL"):
> flight 110513 linux-4.9 real [real]
> http://logs.test-lab.xenproject.org/osstest/logs/110513/
> 
> Regressions :-(
> 
> Tests which did not succeed and are blocking,
> including tests which could not be run:
>  test-amd64-amd64-xl-credit2 19 guest-start/debian.repeat fail REGR. vs. 
> 110456

In the spirit of trying to chase down heisenbugs: failures like this
seem to occur occasionally.  Here are the job-specific logs:

 
http://logs.test-lab.xenproject.org/osstest/logs/110513/test-amd64-amd64-xl-credit2/info.html

My observations looking at the logs:

 * The test was declared a failure because something, almost certainly
   blkback, still had the LV open when osstest wanted to restart the
   guest.  This occurred 14 seconds after `xl shutdown -w' had
   completed and `xl list' said the domain did not exist, so it is
   not a simple race between the old and new domains.

 * The xenstored access log shows the toolstack deleting the vbd
   backend during the shutdown, as expected.

 * `xl list' shows two zombie domains: 17 (the recently-shut-down
   guest) and 16 (the previous one).

 * The toolstack log shows that xl thinks it successfully shut down
   domain 17 and then destroyed it.

 * The hotplug script log is empty.

 * I see nothing relevant in the dom0 serial log, before the failure

I conjecture that blkback is not shutting down properly.

I haven't investigated which osstest branches this occurs on and what
the failure frequency is like.

Ian.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel

 


Rackspace

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