Hi Stephan!
Quoting <sk...@ithnet.com> (22.10.10 15:47):
> you are talking of the problem with identification field being only 16 bit,
> right?
Right. I had some pretty bad experience ~2006 before I switched to TCP.
> We experienced this scenario to be far less severe than TCP busted by packet
> drops. In fact we were not able to run NFS over TCP for more than 2 days
> without a complete service breakdown. Whereas UDP runs for several years now
> without seeing the corruption issue.
You must have an interesting setup :-) I'm using NFS over TCP since the
early 2.6 times in hundreds of installations. Range from 2-2000 clients,
usually GigE. Of course packet loss may happen, but it is recovered by the
typical mechanism in the kernel. All clients have above 2G of memory which
sets some TCP tuning parameters to reasonable values. Boxes with a lower
amount of memory could have issues when handling a large amount of TCP
connections.
NFSv4 gives me still some headache. SLES is currently completely broken and
will crash on a regular base, RHEL and it's clones have still some bugs in
the locking. NFSv3 is rock solid on both.
I should go to the weekend. I'm talking about NFS on the Gluster mailing
list. Completly off topic. Sorry for that! :-)
Beat
--
\|/ Beat Rubischon <b...@0x1b.ch>
( 0^0 ) http://www.0x1b.ch/~beat/
oOO--(_)--OOo---------------------------------------------------
Meine Erlebnisse, Gedanken und Traeume: http://www.0x1b.ch/blog/
_______________________________________________
Gluster-users mailing list
Gluster-users@gluster.org
http://gluster.org/cgi-bin/mailman/listinfo/gluster-users
