Re: [Condor-users] Xen VM universe

["Jonathan D. Proulx" <jon@xxxxxxxxxxxxx>]

On Fri, Nov 20, 2009 at 10:15:37AM -0600, Timothy St. Clair wrote:
:Jon -
:
:	Rob and I had been looking in the logs, and there doesn't appear to be 
:an error in the vm_gahp.  The shadow & starter logs may shed some more 
:light on how it is failing.  The startd log may also be helpful..

I've changes something (not sure how or what) & I'm getting a
decernable error now in the vm_ghap about libvir tnot being able to
initialize xen:/// (though I can use virsh to start this domain by
hand)

What I did:

I made a copy of condor_vm_xen.sh as condor_vm_xen-debug.sh made a
small edit to make it copy the libvirt config file to /tmp so I could
examine the contents:


86a93,96
>       #<KLUDGE>
>       echo "configfile:  $XEN_CONFIG_FILE"
>       TMPFILE=`mktemp /tmp/xen-virt.XXXXXXXXX `;sleep 1
>       cp $XEN_CONFIG_FILE $TMPFILE
>       #</KLUDGE>

I'd done this on an earlier (condor 7.0?) attempt to get teh VM
universe running , this time I got tmp file and I got :

11/20 11:30:42 (fd:7) (pid:20251) ERROR "Failed to create libvirt connection: could not connect to xen:///" at line 1616 in file xen_type.cpp

in the VMGhapLog, surprisingly I still get this after pointing the
config back at the original condor_vm_xen.sh script...

I didn't think to look in the shaddow log before I started getting the
libvirt errors currently I'm seeing:

11/20 04:57:35 (306.0) (18170): Request to run on slot1@xxxxxxxxxxxxxxxxxxxx <128.30.112.149:37681> was ACCEPTED
11/20 04:57:45 (306.0) (18170): ERROR "Error from slot1@xxxxxxxxxxxxxxxxxxxx: Failed to create a new VM" at line 648 in file pseudo_ops.cpp

I guess finding out what I did or didn't do to libvirt is first order
of business, perhaps it's dropping root too soon?

Thanks,
-Jon