[Libvir] PATCH: Don't leave a 'braindead' Xen VM when device hotplug fails
Because of the way Xen guest creation works, the virDomainCreateLinux implementation for Xen is really a three stage process - first we create the raw domain, then we have to wait for device hotplug to complete, finally unpausing the domain. Currently if the device hotplug fails we error out, leaving a 'braindead' VM lying around in the paused state. This is rather confusing for users leading to people thinking everything was OK, and thus unpausing the guest manually, then filing bug reports when they find they've no devices in the guest VM ! The xm tool by comparison will tear down the VM if the device hotplug fails so the user doesn't see the paused domain after the failure. The attached patch re-arranges the Xen driver for creating new domains to do the same kind of thing - upon any failure we explicitly destroy the halfdead domain Regards, Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=|
On Wed, Feb 21, 2007 at 04:31:16PM +0000, Daniel P. Berrange wrote:
Because of the way Xen guest creation works, the virDomainCreateLinux implementation for Xen is really a three stage process - first we create the raw domain, then we have to wait for device hotplug to complete, finally unpausing the domain.
Currently if the device hotplug fails we error out, leaving a 'braindead' VM lying around in the paused state. This is rather confusing for users leading to people thinking everything was OK, and thus unpausing the guest manually, then filing bug reports when they find they've no devices in the guest VM !
Makes sense. I'm just not clear why code need to be moved from one function to another, but I'm lacking context ATM. Daniel -- Red Hat Virtualization group http://redhat.com/virtualization/ Daniel Veillard | virtualization library http://libvirt.org/ veillard@redhat.com | libxml GNOME XML XSLT toolkit http://xmlsoft.org/ http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/
On Thu, Feb 22, 2007 at 01:51:11AM -0500, Daniel Veillard wrote:
On Wed, Feb 21, 2007 at 04:31:16PM +0000, Daniel P. Berrange wrote:
Because of the way Xen guest creation works, the virDomainCreateLinux implementation for Xen is really a three stage process - first we create the raw domain, then we have to wait for device hotplug to complete, finally unpausing the domain.
Currently if the device hotplug fails we error out, leaving a 'braindead' VM lying around in the paused state. This is rather confusing for users leading to people thinking everything was OK, and thus unpausing the guest manually, then filing bug reports when they find they've no devices in the guest VM !
Makes sense. I'm just not clear why code need to be moved from one function to another, but I'm lacking context ATM.
It isn't moved from one function to another - there are two fairly similar functions - one deals with creating fresh domains (in xend_internal.c) and the other deals with starting inactive domains (in xm_internal.c). Both needed a fairly similar fix. Regards, Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=|
participants (2)
-
Daniel P. Berrange -
Daniel Veillard