[Xen-users] Live migration of HVM guest is hanging (Xen 3.1)

[Andrea Brugiolo <andrea@xxxxxxxxxxxx>]

Hi all

I am fighting with live migration and the process is hanging with this
errors in xend.log of the sender dom0:

------------------------------- SNIP ---------------------------------------
    ...
    [2008-01-17 00:30:20 663] INFO (XendCheckpoint:362) All memory is saved
    [2008-01-17 00:30:20 663] INFO (XendCheckpoint:362) Save exit rc=0
    [2008-01-17 00:30:20 663] ERROR (XendCheckpoint:140) Save failed on domain 
grifone (16).
    Traceback (most recent call last):
      File "/usr/lib/xen-3.1-1/lib/python/xen/xend/XendCheckpoint.py", line 
114, in save
        qemu_fd = os.open("/tmp/xen.qemu-dm.%d" % dominfo.getDomid(), 
os.O_RDONLY)
    OSError: [Errno 2] No such file or directory: '/tmp/xen.qemu-dm.16'
    [2008-01-17 00:30:20 663] DEBUG (XendDomainInfo:1793) 
XendDomainInfo.resumeDomain(16)
    [2008-01-17 00:30:20 663] DEBUG (XendCheckpoint:143) XendCheckpoint.save: 
resumeDomain
    [2008-01-17 00:30:20 663] ERROR (xmlrpclib2:178) Internal error handling 
xend.domain.migrate
    Traceback (most recent call last):
      File "/usr/lib/xen-3.1-1/lib/python/xen/util/xmlrpclib2.py", line 131, in 
_marshaled_dispatch
        response = self._dispatch(method, params)
      File "/usr/lib/python2.4/SimpleXMLRPCServer.py", line 415, in _dispatch
        """
      File "/usr/lib/xen-3.1-1/lib/python/xen/xend/XendDomain.py", line 1240, 
in domain_migrate
        XendCheckpoint.save(sock.fileno(), dominfo, True, live, dst)
      File "/usr/lib/xen-3.1-1/lib/python/xen/xend/XendCheckpoint.py", line 
150, in save
        raise exn
    OSError: [Errno 2] No such file or directory: '/tmp/xen.qemu-dm.16'
    ...
-------------------------------- SNAP ----------------------------------

What is strange is that, after I stop watching the logs, i can see
that the /tmp/xen.qemu-dm.16 file *does exist*

Maybe it was created too late to be read by a restore process...?

Of course, after the migration hangs, I can:

 - destroy both the partially?-migrated domU instances (on sender and
   on receiver);
 - create the domU again on the sender and it will get the next id,
   let's say 17;
 - rename /tmp/xen.qemu-dm.16 to /tmp/xen.qemu-dm.17, therefore the
   file is in place in advance (!);
 - restart the live migration

and everything will work, just after a longer delay.

But that's not the right way, I would say... :-]

I have read the Python functions in those files but they don't tell me
much more than the process cannot find a file and an exception is
handled...

More, I feel that my xend-config.sxp is OK for migration because I
made a migration of a GNU/Linux guest between the same dom0s just
before this one and it was perfect and immediate.

Now I am stracing the `xm migrate --live ...' command but I cannot see
anything significant... or I am too tired.

Any idea?

Thank you very much in advance!

Andrea

-- 
Andrea Brugiolo                          andrea.brugiolo@xxxxxxxx
Universita` degli Studi di Padova        http://www.unipd.it
Centro di Ateneo per le Biblioteche      http://www.cab.unipd.it
tel +39-049-827-3688                     fax +39-049-827-3651

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users