Hi,
thanks for your script. The current plan is to switch to the script
shipped by upstream. Hopefully we'll ship this with the next release.
Am 05.12.2010 17:57, schrieb Покотиленко Костик:
Hi.
I'm joining this subject, because this is one of several stoppers of
migration from xen.
So, this is another init.d script to autostart libvirt domains. It also
relays on libvirt autosave feature turned off. It's able to save running
domains and resume them at startup. It has a feature not present in
other scripts, it has autostart list of domain which should be started
at boot if they have not been saved or could not be restored.
There is a problem (probably a bug in libvirt in Debian Lenny) during
boot that some domains could not be resumed at several first tries.
Author of this script knows of this problem and there is support to
retry of resume in the script. There were 3 retries initially, I set it
to 10.
In my case I have 3 domains of 1Gb of RAM whose are restoring Ok all the
times, and 2 domain of 2 Gb RAM which have problems restoring. Sometime
they restore Ok, some times they are restoring during first 10 retries,
sometimes they are unable to restore even at 10 retries. I think this is
somehow relate to amount of memory of domain.
Also, I checked other two scripts posted here, there is still problem
with restore, it it doesn't relates to scripts being used. I think it's
libvirt related. I've tryed later version of libvirt/kvm from backport -
there were no problem with resume, but there is a problem with save
which takes forever to save domain, so I rolled back.
This last part is an issue with kvm that´s fixed in squeeze.
Cheers,
-- Guido
----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.
_______________________________________________
Pkg-libvirt-maintainers mailing list
[email protected]
http://lists.alioth.debian.org/mailman/listinfo/pkg-libvirt-maintainers
--
To UNSUBSCRIBE, email to [email protected]
with a subject of "unsubscribe". Trouble? Contact [email protected]
