From: Cohen, Andy (Andy.Cohen@cognex.com)
Date: Tue Nov 25 2003 - 14:56:21 EST
No replies to this from the list but I found a solution.
The culprit seems to have been the MTU size of 9000 -- as soon as I set it back to 1500 the problem cleared up. Ironically, and mysteriously, there are 2 other NFS mounts to similar machines mounted on this DS20 that had no problems with this setting. These are the settings in fstab:
This one didn't like the larger setting: nfsv3 rw 0 0
This one was fine with the larger setting: nfsv3 rw,tcp,intr 0 0
This one was fine with the larger setting: nfsv3 rw,intr 0 0
So whether it was the way they were mounted or something about the remote server's configuration to make them behave differently I haven't been able to determine yet ... and probably won't get to.
Thanks,
Andy
ORIGINAL QUESTIONS
==================
Sorry, I failed to mention that I also just installed a DEGPA gigabit card on the DS20. I also made the following changes per tuning guides, tech support, and advice from this list:
fifo-do-adaptive=0
new_wire_method=0
(to increase default system socket buffer sizes)
tcp_sendspace=524288
tcp_recvspace=524288
(to increase MTU size to use jumbo frames)
ifconfig alt0 imptu 9000
Are any of these incompatible with nfsv3?
Thanks again.
Andy
ORIGINAL QUESTION
=================
Help!
We've got a DS20E running 5.1A that we just applied the latest patch kit (PK6) to. There's an nfsv3 mount (to a solaris box) that doesn't seem to be functioning. I can navigate to the directory that serves as the mount point and issue 'ls' ok. However, if I go to a sub-directory any command I issue just hangs indefinitely and I can't even kill it. Has anybody experienced this type of behavior? Anybody have any suggestions for fixes? I can't unmount because the hung command was issued while in the mounted file system.
Thank you!
Andy
This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:49:45 EDT