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

Re: [Xen-devel] [PATCH 0 of 5] xl shutdown compatibility with xm

Ian Campbell writes ("Re: [PATCH 0 of 5] xl shutdown compatibility with xm"):
> My main concern is that triggering a hibernation or suspend is, AUIU, a
> reasonably common out of the box configuration for some versions of
> Windows, which combined with hibernation being (historically at least)
> notoriously flakey on real hardware (and I would expect more so on a
> virtualised platform) worries me. Also note that we don't actually
> implement suspend to RAM -- the domain will quiesce itself and come to a
> stop but AFAIK it won't actually die and the domain's RAM won't be saved
> anywhere. Presumably this is roughly equivalent to your battery running
> out while suspended and therefore not all that dangerous.

If it triggers a suspend, the domain won't end up shutting down so if
we wait for it there will be no data loss.  If we time out and do a
destroy then doing the suspend first was probably a big improvement.

If it triggers a hibernation, then if the hibernation is broken the
effect will be just as if the user pressed the power button and the
hibernation is broken: the machine will fail to come back and
effectively will have been crashed.  This is probably worse than doing
nothing (in the case where the caller/user isn't going to time out and
do a destroy instead) but it is definitely better than a destroy.

> But anyway I seem to be in the minority. We can always revert it if it
> starts eating peoples data.

OK, I think that's a go-ahead.  Thanks.


Xen-devel mailing list



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