Re: [libvirt] [PATCH v2 2/2] confirm compression program availability
On Tue, Oct 26, 2010 at 08:45:27AM +0900, KAMEZAWA Hiroyuki wrote:
On Mon, 25 Oct 2010 16:20:50 +0100
"Daniel P. Berrange" <berrange(a)redhat.com> wrote:
> On Mon, Oct 25, 2010 at 09:05:29AM +0900, KAMEZAWA Hiroyuki wrote:
> >
> > At compression, external programs are used but it is not checked whether
> > the program is available or not.
> > Check it at parsing qemu.conf and find problems in early stage.
>
> The problem with doing the error check here is that it will
> cause the entire libvirtd to fail to startup, and just syslog
> the error. We've found admins often miss these types of problem
> in syslog and just file bugs complaining that libvirtd doesn't
> start.
>
Hmm, yes.
> I think we should put an explicit check for existance of the
> compression program in the API code instead. That way the
> error message will get fed to the end user who will easily
> notice it.
>
Ok, I'll try it. I wonder which is better to make the command fail or run it
in "raw" format when compression program cannot be found. Because virsh dump
cannot be stopped by Ctrl-C, starting the dump can be critical decision.
The API impl should check whether the compression program exists, before
trying to start the dump operation. If it doesn't exist, then send an
error back to the user.
I think I should just make it fail and add "--raw" option
to virsh..
Yes, we could add a VIR_DOMAIN_DUMP_RAW flag that forces a raw dump
regardless of default compression setting.
Daniel
--
|: Red Hat, Engineering, London -o- http://people.redhat.com/berrange/ :|
|: http://libvirt.org -o- http://virt-manager.org -o- http://deltacloud.org :|
|: http://autobuild.org -o- http://search.cpan.org/~danberr/ :|
|: GnuPG: 7D3B9505 -o- F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|