On Wed, 2005-11-09 at 11:16 +0000, Jonathan Brooks wrote:
Roger Oberholtzer wrote:
On Tue, 2005-11-08 at 18:55 -0800, Joseph Loo wrote:
There is an option on the /etc/fstab that does the mounting automtaically. It has to do with the last 2 numbers on the list. I suggest you add 1 n where n is the order of mounting. If you have 0 in the first number, it will not automatically moun the partition. Naahh..
The fifth number on the line (I get this from the fstab man page) is related to the dump() command, and the sixth (last number on the line) is related to fsck(). I think you are thinking of the key word 'auto' or 'noauto' that you can provide in the comma separated list of mount options in item four on the line. To understand those, look at the 'mount' man page. Still, this is only automatic relative to when the system boots. It does not mean that they will be automatically mounted on demand at a later time. 'autofs' seems to be for that.
Okay, so is there a problem with NFS, SuSE10 and AMD64? It seems like my fstab was correct, and things should be mounted automagically at boot time. Are there any particular commands/logs I can look at on the server when the client boots, to see why it's not being allowed to connect? Is there something different with NFS under SuSE10? I.e. does it want to use nfsvers=4, and the server barfs when it gets these requests?
Roger: is this why setting nfsvers=2 fixed your problems?
Yes. The system I really needed this for was a vxWorks NFS server. It exported the directory and only understood NFS version 2. As the 2.6 kernel (SUSE 9.3 and later) does NFS version 3 internally and by default, this option (and the 'udp' option) was needed to be able to talk to the vxWorks server. Why NFS version 3 does not have a mechanism to fall back on version 2 is anyone's guess.
However, I'm thinking that this isn't something SuSE 10 specific, more something related to AMD64, since the other AMD64 machine with this problem is on SuSE 9.3.
Meaning kernel 2.6. All Linux with this version of the kernel would exhibit this behavior.
Hope someone has some ideas out there? :)
Best wishes,
Jon.
Hmm, our i586 (athlonXP/pentium4) machines running SuSE 9.3 (kernel 2.6) never have this problem. So I think there must be something different in
Roger Oberholtzer wrote: the x86_64 kernel implementations of NFS (?) Or, our SLES server doesn't like the way our AMD64 machines talk to it. Any ideas? Jon. -- Jonathan Brooks (Ph.D.) Research Assistant. PaIN Group, Department of Human Anatomy & Genetics, University of Oxford, South Parks Road, Oxford, OX1 3QX tel: +44(0)1865-282654 fax: +44(0)1865-282656 web: http://www.fmrib.ox.ac.uk/~jon
On Wed, 2005-11-09 at 12:12 +0000, Jonathan Brooks wrote:
On Wed, 2005-11-09 at 11:16 +0000, Jonathan Brooks wrote:
Roger Oberholtzer wrote:
On Tue, 2005-11-08 at 18:55 -0800, Joseph Loo wrote:
There is an option on the /etc/fstab that does the mounting automtaically. It has to do with the last 2 numbers on the list. I suggest you add 1 n where n is the order of mounting. If you have 0 in the first number, it will not automatically moun the partition. Naahh..
The fifth number on the line (I get this from the fstab man page) is related to the dump() command, and the sixth (last number on the line) is related to fsck(). I think you are thinking of the key word 'auto' or 'noauto' that you can provide in the comma separated list of mount options in item four on the line. To understand those, look at the 'mount' man page. Still, this is only automatic relative to when the system boots. It does not mean that they will be automatically mounted on demand at a later time. 'autofs' seems to be for that.
Okay, so is there a problem with NFS, SuSE10 and AMD64? It seems like my fstab was correct, and things should be mounted automagically at boot time. Are there any particular commands/logs I can look at on the server when the client boots, to see why it's not being allowed to connect? Is there something different with NFS under SuSE10? I.e. does it want to use nfsvers=4, and the server barfs when it gets these requests?
Roger: is this why setting nfsvers=2 fixed your problems?
Yes. The system I really needed this for was a vxWorks NFS server. It exported the directory and only understood NFS version 2. As the 2.6 kernel (SUSE 9.3 and later) does NFS version 3 internally and by default, this option (and the 'udp' option) was needed to be able to talk to the vxWorks server. Why NFS version 3 does not have a mechanism to fall back on version 2 is anyone's guess.
However, I'm thinking that this isn't something SuSE 10 specific, more something related to AMD64, since the other AMD64 machine with this problem is on SuSE 9.3.
Meaning kernel 2.6. All Linux with this version of the kernel would exhibit this behavior.
Hope someone has some ideas out there? :)
Best wishes,
Jon.
Hmm, our i586 (athlonXP/pentium4) machines running SuSE 9.3 (kernel 2.6) never have this problem. So I think there must be something different in
Roger Oberholtzer wrote: the x86_64 kernel implementations of NFS (?)
Odd. Our 9.3 were the first to exhibit the problem. So, you have YAP (yet another problem). -- Roger Oberholtzer OPQ Systems AB
participants (2)
-
Jonathan Brooks -
Roger Oberholtzer