Xen users - Sep 2005 - Xen 2.0.7 network issues and Vncserver in Dom0

Has anyone else had problems keeping vncserver running in Dom0?

My set-up is as follows

Dom0
---------
Centos 4.1 on an Athlon XP 2000+ with 512Mb out of 1GB Ram allocated.
I have to run a custom build of the Xen kernel as I have an onboard VIA 
ide controller, plus two extra ide controllers (CMD649 and SIL680).

I have disabled /lib/tls and I''ve got a single static IP address

I run a vnc server instance as one of the users which is started 
automatically out of /etc/sysconfig/vncservers. I normally SSH into this 
environment and then tunnel vnc over SSH to access various gui based 
apps that I want to leave running in the background. On my local network 
I simply access the vnc instance directly on the network.

This instance is my main home server so runs imap, Apache, samba, nfs, 
MailScanner, postfix, clamav, spamassassin.

There are two physical Ethernet devices, but only one of them is used at 
present, and Intel eepro100 card mapped into eth0.

Dom1
---------
Debian Sarge with 128Mb Ram, 256Mb Swap, partitions allocated out of an 
LVM environment by Dom0. My aim is to move my externally facing web 
server into this environment. Whilst pro typing I''m running it with a 
fixed MAC address that gets an IP over DHCP which resides on my internal 
network.

The networking is set-up with a vanilla config using bridged networking.

# Define network interfaces.

# Number of network interfaces. Default is 1.
nics=1

# Optionally define mac and/or bridge for the network interfaces.
vif = [ ''mac=AA:00:00:3D:1A:12, bridge=xen-br0'' ]


The Problem
---------------------
Everything starts up fine at boot, although it appears the command 
"ide=reverse" is ignored by the Dom0 kernel so my onboard drives end
up
as hdi/j instead of hda/b.

The vncserver start up in Dom0 and the new Dom1 instance gets it IP 
address and starts it''s Apache instance.

I connect over the local network to the vnc environment and start up a 
couple of gui based applications such as Pan an Azureus. After a couple 
of minutes the vncserver quits, and I notice the following in 
/var/log/messages

Sep  9 21:31:42 centos vncserver: vncserver startup succeeded
Sep  9 21:31:42 centos anacron: anacron startup succeeded
Sep  9 21:31:42 centos atd: atd startup succeeded
Sep  9 21:31:44 centos messagebus: messagebus startup succeeded
Sep  9 21:31:44 centos cups-config-daemon: cups-config-daemon startup 
succeeded
Sep  9 21:31:44 centos haldaemon: haldaemon startup succeeded
Sep  9 21:31:49 centos fstab-sync[3983]: removed all generated mount points
Sep  9 21:31:54 centos fstab-sync[4061]: added mount point /media/zip1 
for /dev/hdk4
Sep  9 21:31:56 centos kernel: device eth0 entered promiscuous mode
Sep  9 21:31:56 centos kernel: xen-br0: port 1(eth0) entering learning state
Sep  9 21:31:56 centos kernel: xen-br0: topology change detected, 
propagating
Sep  9 21:31:56 centos kernel: xen-br0: port 1(eth0) entering forwarding 
state
Sep  9 21:31:57 centos xend: .
Sep  9 21:31:59 centos xend: .
Sep  9 21:32:00 centos fstab-sync[4310]: added mount point /media/floppy 
for /dev/fd0
Sep  9 21:32:01 centos xend: .
Sep  9 21:32:02 centos xend: .
Sep  9 21:32:04 centos kernel: ip_tables: (C) 2000-2002 Netfilter core team
Sep  9 21:32:05 centos xend: .
Sep  9 21:32:12 centos last message repeated 4 times
Sep  9 21:32:13 centos rc: Starting xend:  failed
Sep  9 21:33:56 centos kernel: device vif1.0 entered promiscuous mode
Sep  9 21:33:56 centos kernel: xen-br0: port 2(vif1.0) entering learning 
state
Sep  9 21:33:56 centos kernel: xen-br0: topology change detected, 
propagating
Sep  9 21:33:56 centos kernel: xen-br0: port 2(vif1.0) entering 
forwarding state


Now i''m not running any iptables rules within Dom0 as all traffic on my
LAN is already protected by my seperate firewall.

Other applications appear to work ok, and i''ve not had any network 
issues with SSH or NFS. I didn''t have try to try SMB connections from a
windows machine at this stage.

So any ideas how I can fix these odd network problems.

Steve.

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

Steven Ellis wrote:
>
> Has anyone else had problems keeping vncserver running in Dom0?
>
>
Further to my original query i''ve just reproduced the vnc server crash 
and foung the following in the vncserver logfile

X connection to centos:1.0 host broken (explicit kill or server shutdown)
X connection to centos:1.0 host broken (explicit kill or server shutdown)
X connection to centos:1.0 broken (explicit kill or server shutdown).
X connection to centos:1.0 broken (explicit kill or server shutdown).

checking the network on Dom0

# brctl show
bridge name     bridge id               STP enabled     interfaces
xen-br0         8000.00508bb9f601       no              eth0
                                                        vif1.0

The really odd part is this time I don''t see ant of the bridge state 
changes in the logs on Dom0.

I''ll try changing the Dom1 instance from dynamic to static IP and see
if
that makes a difference.

Steve


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

Steven Ellis wrote:
>
> The really odd part is this time I don''t see ant of the bridge
state
> changes in the logs on Dom0.
>
> I''ll try changing the Dom1 instance from dynamic to static IP and
see
> if that makes a difference.
>
> Steve
>
Still no luck. I''m going to see how stable the vncserver is with no
Dom1
active.

Steve

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

Steven Ellis wrote:
> Steven Ellis wrote:
>
>>
>> The really odd part is this time I don''t see ant of the bridge
state
>> changes in the logs on Dom0.
>>
>> I''ll try changing the Dom1 instance from dynamic to static IP
and see
>> if that makes a difference.
>>
>> Steve
>>
> Still no luck. I''m going to see how stable the vncserver is with
no
> Dom1 active.
>
That didn''t work so I tried a clean Debian Dom0, and moved my 
Centos/RHEL 4.1 server into a DomU environment. I still have problems 
with VNC crashing after a few minutes. Reboot the server back to a stand 
alone Centos 4.1 and no issues with VNC.

Very, very odd.

Any ideas?

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