NFS mount - Slow when attempting ll command for example

[Nick Wickens]

We have two HP-UX boxes with a couple of nfs mounts between each. The servers were moved to a new site a few weeks ago and are now sited together on the same network (as they was before). However it was noticed that NFS was rather slow particularly when attempting an ll command whilst ls works fine. The problem then "disappeared" for a week but has now come back.

nfsstat command is showing badcalls 221 badxids 217 and timeouts 122 on the client but nothing on the server.

Have spent a while trailing the forums without solution so far - Any ideas ?

[Pete Randall · Accepted solution]

Hi Nick,

Is there any chance you've got a duplex mismatch? Your 100 Mb NICs and the switches they connect to need to be forced to 100 Mb/Full Duplex. If left to auto-negotiate, you can end up with strange anomalies like this one.


Pete

[Gordon Morrison]

As these servers were physically moved (I presume that their IP addresses have not changed), I would start by checking your routing tables and default gateways. Although the IP addresses may not have changed, the data traffic needs to take a different physical route to get to the new location.

[Pete Randall]

I'm sure you know this, but just in case -

To check your NIC run "lanadmin -x N" where N is the card's instance number obtained from "lanscan".


Pete

[harry d brown jr]

I agree with Pete and Gordon, but of course the best answer is: don't do ll's :-))

live free or die
harry d brown jr

[Mark Ellzey]

Nick,

In addition to the above, check your name resolution. Make sure you can nslookup each of the hosts from each other.

Check your /etc/auto_master or /etc/auto_direct files for any 'dead' mounts. I had a server that was trying to mount filesystem from hosts that had been turned off long ago, but no one had changed the /etc/auto_direct file. I had exactly the same symptoms you describe (i.e. ls works fine, ll takes forever).

Regards,
Mark