SUMMARY: AMD on DU 3.2c

From: Marc Heinzmann <heinzman_at_informatik.tu-muenchen.de>
Date: Thu, 18 Jul 1996 15:29:11 +0200 (MESZ)

My questions were:

1. Is there a major performace difference between NFS V2 and NFS V3?
2. Is there a patch for AMD to handle NFS V3?


Answers:

1. V3 is a *lot* faster and more relyable than V2.
   So use V3 if you can.

2. a) David Warren (warren_at_atmos.washington.edu) told me, that there
   is code to handle nfs3 in amd. He didn't have many chances to try it,
   but he thinks it's possible.
   I haven't verified this yet, but I'll give it a try and I'll
   post the solution if I succeed.

   b) Rudi Gabler (RUG_at_USM.Uni-Muenchen.DE) has a work around, which is
   working fine:
   This is how his amd.master.map looks like:

   #
   # there comes the data disks on our cluster
   #
   /defaults type:=program;mount:="/sbin/mount mount -o nfsv3,nogrpid
   ${rhost}:/e/${key} /i/${key}";unmount:="/sbin/umount umount
   ${rhost}:/e/${key}";opts:=rq,quota;fs:=/i/${key}
   data1 host!=hal1;rhost:=hal1; \
                   host==hal1;type:=link;fs:=/e/${key}
   data1b host!=hal1;rhost:=hal1; \
                   host==hal1;type:=link;fs:=/e/${key}
   data2 host!=hal2;rhost:=hal2; \
                   host==hal2;type:=link;fs:=/e/${key}

   Rudi's workaround is to use the 'type' 'program' and whenever
   a filesystem gets mounted, '/sbin/mount' mounts this filesystem with
   the option 'nfsv3'.


Thanks to Rudi, David and Martyn for their replies!

        Marc
-- 
Marc Heinzmann		http://www.informatik.tu-muenchen.de/~heinzman/	
Received on Thu Jul 18 1996 - 15:48:36 NZST

This archive was generated by hypermail 2.4.0 : Wed Nov 08 2023 - 11:53:46 NZDT