[Xen-users] Guest Domains drop from network

[Scott Forten <scott.forten@xxxxxxxxxx>]

Xen Host Server OS:     SLES10
Kernel:            2.6.16.21-0.8-xen x86_64
Hardware:        Sun 4200             
Memory:            12Gb
Guest Domains (2 total)
RHEL4-U3
2.6.16-xen x86_64
2Gb Swap
4Gb RAM
2 Nics bridged to seperate physical Nics (Public and Private)


Systems come up normal and we are able to mount filer space to the guest 
domains and do work.  However, at some point, the guest OS will drop 
from the network. From the general network, we are unable to ssh or ping 
the guest domains.
If we log into the host server however, we are able to ssh to the guest 
domain. Doing this, the guest OS becomes immediately available to the 
general network again. At no time does the guest OS every shutdown.
I have tried turning off the powersaved and acpid services on the host 
server and guest domains with no success.
I have also seen some errors in /var/log/xend.log. Here is a section 
from xend.log that I think pertains to the problem, but I don't know how 
to address it
[2006-08-23 15:25:16 xend.XendDomainInfo] DEBUG (XendDomainInfo:700) 
Storing dom
ain details: {'console/ring-ref': '3861310', 'console/port': '2', 
'name': 'xen1'
, 'console/limit': '1048576', 'vm': 
'/vm/e6e867f7-fc36-67eb-def1-b694ea019626',
'domid': '25', 'cpu/0/availability': 'online', 'memory/target': 
'4194304', 'stor
e/ring-ref': '3861186', 'cpu/1/availability': 'online', 'store/port': '1'}
[2006-08-23 15:25:16 xend.XendDomainInfo] DEBUG (XendDomainInfo:881) 
XendDomainI
nfo.handleShutdownWatch
[2006-08-23 15:25:16 xend] DEBUG (DevController:132) Waiting for devices 
vif.
[2006-08-23 15:25:16 xend] DEBUG (DevController:138) Waiting for 0.
[2006-08-23 15:25:16 xend] DEBUG (DevController:403) 
hotplugStatusCallback /loca
l/domain/0/backend/vif/25/0/hotplug-status.
[2006-08-23 15:25:16 xend] DEBUG (DevController:403) 
hotplugStatusCallback /loca
l/domain/0/backend/vif/25/0/hotplug-status.
[2006-08-23 15:25:16 xend] DEBUG (DevController:417) 
hotplugStatusCallback 1.
[2006-08-23 15:25:16 xend] DEBUG (DevController:138) Waiting for 1.
[2006-08-23 15:25:16 xend] DEBUG (DevController:403) 
hotplugStatusCallback /loca
l/domain/0/backend/vif/25/1/hotplug-status.
[2006-08-23 15:25:16 xend] DEBUG (DevController:417) 
hotplugStatusCallback 1.
[2006-08-23 15:25:16 xend] DEBUG (DevController:138) Waiting for 2.
[2006-08-23 15:25:16 xend] DEBUG (DevController:403) 
hotplugStatusCallback /loca
l/domain/0/backend/vif/25/2/hotplug-status.
[2006-08-23 15:25:16 xend] DEBUG (DevController:417) 
hotplugStatusCallback 1.
[2006-08-23 15:25:16 xend] DEBUG (DevController:138) Waiting for 3.
[2006-08-23 15:25:16 xend] DEBUG (DevController:403) 
hotplugStatusCallback /loca
l/domain/0/backend/vif/25/3/hotplug-status.
[2006-08-23 15:25:16 xend] DEBUG (DevController:417) 
hotplugStatusCallback 1.
[2006-08-23 15:25:16 xend] DEBUG (DevController:132) Waiting for devices 
usb.
[2006-08-23 15:25:16 xend] DEBUG (DevController:132) Waiting for devices 
vbd.
[2006-08-23 15:25:16 xend] DEBUG (DevController:138) Waiting for 768.
[2006-08-23 15:25:16 xend] DEBUG (DevController:403) 
hotplugStatusCallback /loca
l/domain/0/backend/vbd/25/768/hotplug-status.
[2006-08-23 15:25:17 xend] DEBUG (DevController:403) 
hotplugStatusCallback /loca
l/domain/0/backend/vbd/25/768/hotplug-status.
[2006-08-23 15:25:17 xend] DEBUG (DevController:417) 
hotplugStatusCallback 1.
[2006-08-23 15:25:17 xend] DEBUG (DevController:138) Waiting for 2070.
[2006-08-23 15:25:17 xend] DEBUG (DevController:403) 
hotplugStatusCallback /loca
l/domain/0/backend/vbd/25/2070/hotplug-status.
[2006-08-23 15:25:17 xend] DEBUG (DevController:417) 
hotplugStatusCallback 1.
[2006-08-23 15:25:17 xend] DEBUG (DevController:132) Waiting for devices 
irq.
[2006-08-23 15:25:17 xend] DEBUG (DevController:132) Waiting for devices 
pci.
[2006-08-23 15:25:17 xend] DEBUG (DevController:132) Waiting for devices 
ioports
.
[2006-08-23 15:25:17 xend] DEBUG (DevController:132) Waiting for devices 
vtpm.
[2006-08-23 15:25:17 xend] INFO (XendDomain:363) Domain xen1 (25) unpaused.

There are numerous references to a bug and workaround using "ethtool -K 
eth0 tx off". But the descriptions of these workarounds indicate this 
solution is more for throughput / nic performance and not for the 
situation I'm facing, a complete loss of the guest domain from the network.
guest-os config:

disk = [ 'file:/xen1/rhel4u3.img,hda,w', 'file:/xen/swap/xen1swap,sdb6,w' ]
memory = 4096
vcpus = 2
builder = 'linux'
name = 'xen1'
vif = [ 'mac=00:16:3e:4b:bb:4e,bridge=xenbr0', 
'mac=00:16:3e:f1:56:2d,bridge=xen
br1', 'mac=00:16:3e:3e:e8:75', 'mac=00:16:3e:85:b5:60' ]
root = '/dev/hda1'
localtime = 0
on_poweroff = 'destroy'
on_reboot = 'restart'
on_crash = 'restart'
extra = ' TERM=xterm'
bootloader = '/usr/lib/xen/boot/domUloader.py'
bootentry = 'hda1:/boot/vmlinuz-2.6.16-xen,/boot/initrd-2.6.16-xen.img'

my-network-script (to enable bridging to specific NICs)

#!/bin/sh
dir=$(dirname "$0")
"$dir/network-bridge" "$@" vifnum=0 netdev=eth0
"$dir/network-bridge" "$@" vifnum=1 netdev=eth1



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