SFS Client Breaking in LInux 2.6.22?

"Kevin Berry" <[email protected]> Thu, 11 Oct 2007 17:34:23 -0500
Newsgroups gmane.comp.file-systems.sfs.general
Message-ID <[email protected]>
------=_Part_6876_33534190.1192142063872
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

I originally tried to send this back in August, but the mailing lists seemed
to be down then.  Trying again now as I noticed some traffic:

Hello.  I just upgraded 2 different machines (i386 and amd64) to Linux
2.6.22.2 kernel, from 2.6.21.5 and 2.6.21.1.  Since then, sfs client has
been breaking rather badly.  I can get small directory listings okay (less
than 5 files) but anything larger than that or trying to access any files in
the sfs mount seems to deadlock the process involved.  In my logging, I've
seen only the following in my logs from sfscd to give any indication as to
what the problem might be:

2007-08-13_18:19:38.69605 sfscd: unknown named protocol socket
2007-08-13_18:20:25.77349 sfscd: afsdir::xdr: bad cookie 0xf

This was approximately 2 minutes after the initial mount of the sfs server.
sfssd and its friends seem to run just fine on this kernel.  I've noticed
that the NFS client code seems to have changed significantly between
2.6.21.5 and 2.6.22.2 by just diffing the linux/fs/nfs/ directories of each
tree.  I'm not sure if that's where I should start to dig to see what's
broken this.

Can anyone provide some input as to where I might go to fix this?

-- 
Kevin Berry

------=_Part_6876_33534190.1192142063872
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

I originally tried to send this back in August, but the mailing lists seemed to be down then.&nbsp; Trying again now as I noticed some traffic:<br><br>Hello.&nbsp; I just upgraded 2 different machines (i386 and amd64) to Linux <a href="http://2.6.22.2/" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
2.6.22.2</a> kernel, from <a href="http://2.6.21.5/" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">2.6.21.5</a> and <a href="http://2.6.21.1/" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
2.6.21.1</a>.&nbsp;
Since then, sfs client has been breaking rather badly.&nbsp; I can get small
directory listings okay (less than 5 files) but anything larger than
that or trying to access any files in the sfs mount seems to deadlock
the process involved.&nbsp; In my logging, I&#39;ve seen only the following in
my logs from sfscd to give any indication as to what the problem might
be:
<br><br>2007-08-13_18:19:38.69605 sfscd: unknown named protocol socket<br>2007-08-13_18:20:25.77349 sfscd: afsdir::xdr: bad cookie 0xf<br><br>This
was approximately 2 minutes after the initial mount of the sfs server.&nbsp;
sfssd and its friends seem to run just fine on this kernel.&nbsp; I&#39;ve
noticed that the NFS client code seems to have changed significantly
between <a href="http://2.6.21.5/" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">2.6.21.5</a> and <a href="http://2.6.22.2/" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
2.6.22.2</a>
by just diffing the linux/fs/nfs/ directories of each tree.&nbsp; I&#39;m not
sure if that&#39;s where I should start to dig to see what&#39;s broken this.
<br><br>Can anyone provide some input as to where I might go to fix this?<br><br>-- <br>Kevin Berry

------=_Part_6876_33534190.1192142063872--