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

Re: [libvirt] Silently ignored virDomainRestore failures



On Mon, Sep 28, 2009 at 09:03:37PM -0500, Charles Duffy wrote:
> Charles Duffy wrote:
> >What I'm tempted to do is add a command which sends a sigil to stderr to 
> >the end of the exec: migration lines specified by libvirt, and wait for 
> >either that sigil or an error to show up in the log for that domain 
> >before issuing the cont; if my memory is at all correct, libvirt should 
> >have some helper functions useful for that purpose already available.
> 
> Ugh, bad idea -- I was thinking only of the migrate-from-file case, and 
> not of migration from any alternate source; any non-exec source doesn't 
> allow for this hack.
> 
> Anthony suggested dropping -S from the command line on incoming 
> migrations, and using "info status" to poll for when the guest resumes 
> itself; however, this doesn't work for cases when the VM was saved or 
> migrated in a paused state and the end-user expectation is for it to 
> remain paused.

We can't drop -S, because we need to be able to run 'info cpus' and
then do  sched_setaffinity on each vCPU, before it ever starts
executing, otherwise memory will end u pinned to the wrong NUMA nodes

> I'm tempted to add additional output to "info migrate" indicating 
> whether an inbound migration is ongoing; however, this wouldn't help 
> versions of qemu without the patch applied.

Yeah, I've been thinking much the same this morning. I think we should
consider what the optimal setup is for our needs long term and try and
do whatever we can for that in QEMU now. I think it'd definitely be
worthwhile to have an 'info migrate' impl for incoming migration, and
even to make '-incoming' optional, and add a 'migrate-incoming' command
to the monitor, which like 'migrate' could be run in either blocking 
or non-blocking mode.

For existing QEMU, it might be sufficient to just put an arbitrary
'sleep(5)' before issuing 'cont', which would at least give it a 
reasonable chance of avoiding the race condition.

Regards,
Daniel
-- 
|: Red Hat, Engineering, London   -o-   http://people.redhat.com/berrange/ :|
|: http://libvirt.org  -o-  http://virt-manager.org  -o-  http://ovirt.org :|
|: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
|: GnuPG: 7D3B9505  -o-  F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|


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