xenbsd /netbsd: xennetback: can't get new mcl pages (0)

Hi,

In the weekend I had trouble with netbsd / xen. Console
showed following
text and xen guestdomains where not accessible from network.
Also Dom0
not allowed ssh login, only ping worked. Finally I had to
reboot machine
from powerswitch.
I think that problem started somewhere where automated
rootfilesystem
snapshot was being made or copied to NFS.

This was in the /var/log/messages (I copied only the
relevant parts):

Mar  4 00:34:01 xenbsd /netbsd: nfs server
nfsHOST:/LinuxBackup: not
responding
Mar  4 00:34:01 xenbsd /netbsd: nfs server
nfsHOST:/LinuxBackup: is
alive again
Mar  4 00:36:16 xenbsd /netbsd: nfs server
nfsHOST:/LinuxBackup: not
responding
Mar  4 00:36:16 xenbsd /netbsd: nfs server
nfsHOST:/LinuxBackup: is
alive again
Mar  4 00:39:39 xenbsd /netbsd: nfs server
nfsHOST:/LinuxBackup: not
responding
Mar  4 00:39:39 xenbsd /netbsd: nfs server
nfsHOST:/LinuxBackup: is
alive again

Mar  4 00:41:13 xenbsd /netbsd: xennetback: got only 14 new
mcl pages

Mar  4 00:41:53 xenbsd /netbsd: xennetback: can't get new
mcl pages (0)
Mar  4 00:41:53 xenbsd /netbsd: xennetback_ifstart: no
mcl_pages
Mar  4 00:41:53 xenbsd /netbsd: xennetback: can't get new
mcl pages (0)
Mar  4 00:41:53 xenbsd /netbsd: xennetback_ifstart: no
mcl_pages
Mar  4 00:41:53 xenbsd /netbsd: xennetback: can't get new
mcl pages (0)
Mar  4 00:41:53 xenbsd /netbsd: xennetback_ifstart: no
mcl_pages
Mar  4 00:41:53 xenbsd /netbsd: s
Mar  4 00:41:53 xenbsd /netbsd: xennetback: can't get new
mcl pages (0)
Mar  4 00:41:53 xenbsd /netbsd: xennetback_ifstart: no
mcl_pages

> -----Original Message-----
> From: port-xen-ownerNetBSD.org 
> [mailto:port-xen-ownerNetBSD.org] On Behalf Of
Anzi
> Sent: Monday, 5 March 2007 7:06 PM
> To: port-xenNetBSD.org
> Subject: xenbsd /netbsd: xennetback: can't get new mcl
pages (0)
> 
> Mar  4 00:34:01 xenbsd /netbsd: nfs server 
> nfsHOST:/LinuxBackup: not responding Mar  4 00:34:01
xenbsd 
> /netbsd: nfs server nfsHOST:/LinuxBackup: is alive
again Mar  
> 4 00:36:16 xenbsd /netbsd: nfs server
nfsHOST:/LinuxBackup: 
> not responding Mar  4 00:36:16 xenbsd /netbsd: nfs
server 
> nfsHOST:/LinuxBackup: is alive again Mar  4 00:39:39
xenbsd 
> /netbsd: nfs server nfsHOST:/LinuxBackup: not
responding Mar  
> 4 00:39:39 xenbsd /netbsd: nfs server
nfsHOST:/LinuxBackup: 
> is alive again

I've just started seeing this with one of my domu nfs
clients. When there
are a lot of connections all I/O slows to a crawl, almost
unusable.

I've yet to really determine what the cause is. I'm not
seeing the mcl
errors you are. I've tried a lot of things nfs related to
fix it but no
luck. One easy way to replicate the problem is to fire up
mldonkey and point
the temp/download locations at nfs. This normally causes the
failure within
about 10 mins.

Unmounting then remounting seems to fix it for another 10
mins.

Loading up nfs doesn't seem to cause it either, large files
transfer fine.

All other nfs clients with minimal connections continue to
function fine.
This used to be the case with mldonkey also, so I'm not sure
what changed.

Any ideas? I'm not great with nfs anyway so I may be
overlooking something
simple that is nfs related. If someone has the questions,
I'll probably have
the answers  ... Any
help appreciated, there's a lot of people missing out
while this domu is down ;)

Sarton

On Mon, Mar 05, 2007 at 08:25:53PM +1100, Sarton O'Brien
wrote:
> > -----Original Message-----
> > From: port-xen-ownerNetBSD.org 
> > [mailto:port-xen-ownerNetBSD.org] On Behalf Of
Anzi
> > Sent: Monday, 5 March 2007 7:06 PM
> > To: port-xenNetBSD.org
> > Subject: xenbsd /netbsd: xennetback: can't get new
mcl pages (0)
> > 
> > Mar  4 00:34:01 xenbsd /netbsd: nfs server 
> > nfsHOST:/LinuxBackup: not responding Mar  4
00:34:01 xenbsd 
> > /netbsd: nfs server nfsHOST:/LinuxBackup: is alive
again Mar  
> > 4 00:36:16 xenbsd /netbsd: nfs server
nfsHOST:/LinuxBackup: 
> > not responding Mar  4 00:36:16 xenbsd /netbsd: nfs
server 
> > nfsHOST:/LinuxBackup: is alive again Mar  4
00:39:39 xenbsd 
> > /netbsd: nfs server nfsHOST:/LinuxBackup: not
responding Mar  
> > 4 00:39:39 xenbsd /netbsd: nfs server
nfsHOST:/LinuxBackup: 
> > is alive again
> 
> I've just started seeing this with one of my domu nfs
clients. When there
> are a lot of connections all I/O slows to a crawl,
almost unusable.
> 
> I've yet to really determine what the cause is. I'm not
seeing the mcl
> errors you are. I've tried a lot of things nfs related
to fix it but no
> luck. One easy way to replicate the problem is to fire
up mldonkey and point
> the temp/download locations at nfs. This normally
causes the failure within
> about 10 mins.
> 
> Unmounting then remounting seems to fix it for another
10 mins.
> 
> Loading up nfs doesn't seem to cause it either, large
files transfer fine.
> 
> All other nfs clients with minimal connections continue
to function fine.
> This used to be the case with mldonkey also, so I'm not
sure what changed.
> 
> Any ideas? I'm not great with nfs anyway so I may be
overlooking something
> simple that is nfs related. If someone has the
questions, I'll probably have
> the answers  ... Any
help appreciated, there's a lot of people missing out
> while this domu is down ;)

What xen version are you running ?

-- 
Manuel Bouyer, LIP6, Universite Paris VI.          
Manuel.Bouyerlip6.fr
     NetBSD: 26 ans d'experience feront toujours la
difference
--

> -----Original Message-----
> From: port-xen-ownerNetBSD.org 
> [mailto:port-xen-ownerNetBSD.org] On Behalf Of
Anzi
> Sent: Monday, 5 March 2007 7:06 PM
> To: port-xenNetBSD.org
> Subject: xenbsd /netbsd: xennetback: can't get new mcl
pages (0)
> 
> Mar  4 00:34:01 xenbsd /netbsd: nfs server 
> nfsHOST:/LinuxBackup: not responding Mar  4 00:34:01
xenbsd 
> /netbsd: nfs server nfsHOST:/LinuxBackup: is alive
again Mar  
> 4 00:36:16 xenbsd /netbsd: nfs server
nfsHOST:/LinuxBackup: 
> not responding Mar  4 00:36:16 xenbsd /netbsd: nfs
server 
> nfsHOST:/LinuxBackup: is alive again Mar  4 00:39:39
xenbsd 
> /netbsd: nfs server nfsHOST:/LinuxBackup: not
responding Mar  
> 4 00:39:39 xenbsd /netbsd: nfs server
nfsHOST:/LinuxBackup: 
> is alive again

I've just started seeing this with one of my domu nfs
clients. When there
are a lot of connections all I/O slows to a crawl, almost
unusable.

I've yet to really determine what the cause is. I'm not
seeing the mcl
errors you are. I've tried a lot of things nfs related to
fix it but no
luck. One easy way to replicate the problem is to fire up
mldonkey and point
the temp/download locations at nfs. This normally causes the
failure within
about 10 mins.

Unmounting then remounting seems to fix it for another 10
mins.

Loading up nfs doesn't seem to cause it either, large files
transfer fine.

All other nfs clients with minimal connections continue to
function fine.
This used to be the case with mldonkey also, so I'm not sure
what changed.

Any ideas? I'm not great with nfs anyway so I may be
overlooking something
simple that is nfs related. If someone has the questions,
I'll probably have
the answers  ... Any
help appreciated, there's a lot of people missing out
while this domu is down ;)

Sarton

On Mon, Mar 05, 2007 at 08:25:53PM +1100, Sarton O'Brien
wrote:
> > -----Original Message-----
> > From: port-xen-ownerNetBSD.org 
> > [mailto:port-xen-ownerNetBSD.org] On Behalf Of
Anzi
> > Sent: Monday, 5 March 2007 7:06 PM
> > To: port-xenNetBSD.org
> > Subject: xenbsd /netbsd: xennetback: can't get new
mcl pages (0)
> > 
> > Mar  4 00:34:01 xenbsd /netbsd: nfs server 
> > nfsHOST:/LinuxBackup: not responding Mar  4
00:34:01 xenbsd 
> > /netbsd: nfs server nfsHOST:/LinuxBackup: is alive
again Mar  
> > 4 00:36:16 xenbsd /netbsd: nfs server
nfsHOST:/LinuxBackup: 
> > not responding Mar  4 00:36:16 xenbsd /netbsd: nfs
server 
> > nfsHOST:/LinuxBackup: is alive again Mar  4
00:39:39 xenbsd 
> > /netbsd: nfs server nfsHOST:/LinuxBackup: not
responding Mar  
> > 4 00:39:39 xenbsd /netbsd: nfs server
nfsHOST:/LinuxBackup: 
> > is alive again
> 
> I've just started seeing this with one of my domu nfs
clients. When there
> are a lot of connections all I/O slows to a crawl,
almost unusable.
> 
> I've yet to really determine what the cause is. I'm not
seeing the mcl
> errors you are. I've tried a lot of things nfs related
to fix it but no
> luck. One easy way to replicate the problem is to fire
up mldonkey and point
> the temp/download locations at nfs. This normally
causes the failure within
> about 10 mins.
> 
> Unmounting then remounting seems to fix it for another
10 mins.
> 
> Loading up nfs doesn't seem to cause it either, large
files transfer fine.
> 
> All other nfs clients with minimal connections continue
to function fine.
> This used to be the case with mldonkey also, so I'm not
sure what changed.
> 
> Any ideas? I'm not great with nfs anyway so I may be
overlooking something
> simple that is nfs related. If someone has the
questions, I'll probably have
> the answers  ... Any
help appreciated, there's a lot of people missing out
> while this domu is down ;)

What xen version are you running ?

-- 
Manuel Bouyer, LIP6, Universite Paris VI.          
Manuel.Bouyerlip6.fr
     NetBSD: 26 ans d'experience feront toujours la
difference
--