NFS Mount write issue...

[OldSchool]

I cannot recreate the problem and I have other clients mounted to the exact same directory, which eliminates the NFS server...


If any of the other servers noted above are HP-UX, can you (or somebody) check the fstab entry for this mount and compare it w/ the one on the problem box?

[jmckinzie]

I have already done this...the monut on this particualr server will not work unless the packet size is 1024.

WE are checking Networking at this time.

[A. Clay Stephenson]

I am probably wrong but since you seem to be saying that you have other HP-UX boxes at the same OS version and presumably patched the same that are working normally, I'm going to ask that you look at one kernel tunable, timeslice. If someone has set that to 1 rather than 10 then all kinds of incredible behavior that is far from obvious can occur. For example, I have seen cases where the left-arrow key (which might be the 3 character sequence, ,"[","A", for example would occasionally not be correctly handled by curses and instead be read as 3 distinct characters rather than the single pseudocharacter KEY_LEFT. Other times, KEY_LEFT would be detected as expected. If timeslice is set to a small value, it's just possible to be whacky enough to cause your problems.

However, I am still leaning towards an underlying network problem (and one not-so-obvious candidate that springs to mind is cable length -- the total length of your segment may be just on the hairy edge). Your next move may be to become much more familiar with Ethereal than you ever wanted to be.

[Dave Olker]

Hi Jody,

Sorry, I'm confused:

_____________________

I have already done this...the monut on this particualr server will not work unless the packet size is 1024.
_____________________


It's not clear to me which suggestion you were referring to when you said you'd already tried. Which things have you tried already?

[jmckinzie]

Sorry, I already compared the /etc/fstab.

THe settings on one server will not work on the other server.

[jmckinzie]

I don't have admin rights on the netapp.
I cannot change it to use tcp, so thats out...

Any other ideas as to what to try?

I really appreciate ALL the responses.

Thanks,

Jody

[Dave Olker]

Have you collected the lanadmin or netstat -s output yet? Are you using a Gigabit interface or a 100BT interface? There are known issues with NFS/UDP and certain GigE interfaces. The issues are fixed in the latest GigE patches.

Dave

[OldSchool]

"I already compared the /etc/fstab. The settings on one server will not work on the other server."

Thanks...I didn't realize that was done already.

I agree w/ Clay and Dave, the deeper you go, the more this sounds like a network issue.

What hardware model is this box, and what kind of interface card is in use?

Also, is the card set to auto-negotiate or are the speed/duplex hard coded?

(Probably not the case here, but some of the older versions, like 11.0 IIRC, had problems w/ auto-neg vs. non-auto-neg).

You didn't say (or maybe I missed it). The hosts that work, are they in the same datacenter, or located elsewhere? If the are in the same location, try "ping -o" from working and non-working host and see if for some reason they are routing differently.

[jmckinzie]

exported using the rw and 1024 as suggested.