This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
Home Products Support Community News


Re: [Xen-devel] [PATCH] xl: Check for dependencies in xl

On Mon, 2011-01-24 at 17:26 +0000, Kamala Narasimhan wrote:
> > 
> > NAK
> > 
> > Actually I think this type of check should go after the xenstore open
> > has failed, just as an informational/hint type of thing IF it fails.
> > Rather than applying what is really a heuristic and making it CAUSE a
> > failure.
> > 
> Except xs_daemon_open called within libxl_ctx_init does not fail when
> xenstored is not running!  We get this non-obvious error message -
> "failed to free memory for the domain" at a later point.  This patch
> was an attempt to make the "xenstored not running" issue more
> obvious.  

Oh yes, I didn't pick up on that.

> So, without this check we might have to end up chasing xs_* calls that
> are bound to fail because of using the non null xs_handle returned by
> xs_daemon_open and apply this heuristic upon its failure to make the
> issue more obvious.  Given this info, do you still think we shouldn't
> perform this check?  It would certainly make troubleshooting easier.

I think the idea of having such heuristic checks is a good one. Error
messages can be a bit baffling.

> Alternately, we could investigate why xs_daemon_open is returning a
> non null handle when xenstored is not running.

Sounds like a good idea since, if that is the case, how can a caller
know whether it really got a xenstore connection or not? Maybe we can
just try and ls / and that will fail?

I would like to hear Ian Campbells opinion on this but he's away until
Wednesday I think.


Xen-devel mailing list