RE: [Xen-devel] Re: almost working

[James Harper <JamesH@xxxxxxxxxxxxxxxx>]

Just for feedback's sake - I've been running an unprivileged xen domain which has been happily processing spam for over 24 hours. It is configured as a backup mail server and because the primary server is up, it only really gets spam (spammers love to hit MX records other than the lowest as they often aren't configured for spam filtering). It get's a message every few minutes or so.

 

It's running on an SMP machine with two virtual interfaces mapped to two physical interfaces in dom0.

 

my xen tree is probably 36 hours old or so and therefore probably doesn't have any of the recent fixes in it which from what i've read fix quite a few issues - none of which are affecting me thus far except for a reboot problem I saw a few days ago.

 

just a suggestion, could you put those rm's which you mention in a make option, eg make installflush. Maybe exclude /etc/xen from that though or only remove specific files from there though, i've modified at least one config file in there.

 

James

---

From: Ian Pratt
Sent: Wed 14/07/2004 1:13 AM
To: Niraj Tolia
Cc: Keir Fraser; Ian Pratt; James Harper; xen-devel@xxxxxxxxxxxxxxxxxxxxx; Ian.Pratt@xxxxxxxxxxxx
Subject: Re: [Xen-devel] Re: almost working

> > > I have seen a number of ''heisenbugs" with xend. They especially occur
> > > when I stop xend and start it again. Every now and then a 'xend start'
> > > does nothing and sends no output to console. Commands such as 'xm
> > > list' will give a 'Connection refused' error. A 'xend stop' followed
> > > by a 'xend start' usually fixes this but there have been a few times
> > > where I have had to repeat the cycle a couple of times.
> > >
> > > Also, sometimes after a reboot, a domain creation just fails and the
> > > 'xm create' command does not return. There are no error messages from
> > > xend. Even though I suspect that things are hosed at this point in
> > > time, a 'xend stop' works. However xend will not startup again and
> > > gives a 'IOError: [Errno 16] Device or resource busy' error. The
> > > problem cannot be reproduced as the  same create command works after
> > > another reboot.

> Just FYI, I still see most of the above bugs. Running the latest and greatest.

I've actually found that xend has been very stable for me for the
last few days. How are other people finding it?

I think there's real merit in cleaning out old installations
before doing a new install e.g.:

 rm -rf "/etc/xen /usr/lib/python2.2/site-packages/xen* /usr/lib/libxc* /usr/lib/python2.2/site-packages/Xc*"

If you still get problems, please post any output.


I know of two outstanding Xen bugs, both of which Keir
understands and is about to vanquish:

 * instability after shutting a domain down on a CPU other than CPU 0.

 * dom0 oops when hitting local block devices hard from within
dom0 while doing block io from other domains.

Let's get these two fixed then see where we are.

Ian