[Xen-users] Xen migration fails with "leaking g.e. and page still in use!"

[Matthew Collins <mcollins@xxxxxxxxxxxxx>]

I have a pair of Dell 1950's (dual dual core Xeon) with 8 GB of RAM 
running 64 bit CentOS 5.2. They are clustered with RHCS and connected to 
an iSCSI server for storage. I have a CLVM volume on the iSCSI storage 
that houses my Xen VM LV's. I'm running the pygrub boot loader and 
running the same CentOS kernel inside the VMs as the dom0.
The problem is that live migration always fails. It sort of goes of fin 
limbo and doesn't return for a while and when it does, the domain is 
locked up on the original host. It used to work when I had 2 GB of RAM 
in the servers. There is some discussion here about the issue but no 
proposed resolution:
http://lists.xensource.com/archives/html/xen-users/2007-03/msg00361.html

Is this something that has been addressed in more recent versions of 
Xen? (CentOS 5.2 uses a highly patched 3.04 which is nearly 3.1.) Is 
anyone running a 4GB+ CentOS server who could confirm/deny this issue?
Anyone have any suggestions? Running the latest Xen is of course the 
obvious solution but I'm not keen on stepping outside my distro's 
bounds. I'm capable but I'd really only want to if someone can point out 
a changelog entry that applies to this problem. I've posted this on the 
CentOS lists as well.
This appears on the console (and messages log) of the VM after I try to 
migrate it:
testserver.flmnh.ufl.edu login: WARNING: g.e. still in use!
WARNING: leaking g.e. and page still in use!
WARNING: g.e. still in use!
WARNING: leaking g.e. and page still in use!
netif_release_rx_bufs: 0 xfer, 52 noxfer, 204 unused
WARNING: g.e. still in use!
WARNING: leaking g.e. and page still in use!
WARNING: g.e. still in use!
WARNING: leaking g.e. and page still in use!


This appears in the xend.log of the server where the VM is originally 
running:
[2008-08-08 11:07:11 xend.XendDomainInfo 6973] INFO 
(XendDomainInfo:1722) Dev 51728 still active, looping...
[2008-08-08 11:07:11 xend.XendDomainInfo 6973] INFO 
(XendDomainInfo:1722) Dev 51712 still active, looping...
[2008-08-08 11:07:11 xend.XendDomainInfo 6973] INFO 
(XendDomainInfo:1722) Dev 51728 still active, looping...
[2008-08-08 11:07:11 xend.XendDomainInfo 6973] INFO 
(XendDomainInfo:1728) Dev still active but hit max loop timeout
[2008-08-08 11:08:12 xend 6973] DEBUG (balloon:127) Balloon: 2232 KiB 
free; need 2048; done.
[2008-08-08 11:08:12 xend 6973] DEBUG (XendCheckpoint:89) [xc_save]: 
/usr/lib64/xen/bin/xc_save 4 4 0 0 1
[2008-08-08 11:08:12 xend 6973] INFO (XendCheckpoint:351) ERROR Internal 
error: Couldn't enable shadow mode
[2008-08-08 11:08:12 xend 6973] INFO (XendCheckpoint:351) Save exit rc=1
[2008-08-08 11:08:12 xend 6973] ERROR (XendCheckpoint:133) Save failed 
on domain testserver.flmnh.ufl.edu (4).
Traceback (most recent call last):
  File "/usr/lib64/python2.4/site-packages/xen/xend/XendCheckpoint.py", 
line 110, in save
    forkHelper(cmd, fd, saveInputHandler, False)
  File "/usr/lib64/python2.4/site-packages/xen/xend/XendCheckpoint.py", 
line 339, in forkHelper
    raise XendError("%s failed" % string.join(cmd))
XendError: /usr/lib64/xen/bin/xc_save 4 4 0 0 1 failed


--
Matt Collins
Systems Administrator
Florida Museum of Natural History

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