topic Re: NFS problem in Operating System - HP-UX

NFS problem

Leen Biesemans — Wed, 04 Feb 2004 04:05:41 GMT

I have a small network of HP workstations and one workstation is configured as server, but this workstation has a small harddisk, so I have put the software of a few programs on another workstation. When I want to start the daemons of these programs on the server, then I get the following error:

NFS read failed for server xxxx: RPC: Timed out
PID xxxx killed due to text modification or page I/O error.
Bus error.

It's strange because it has always worked untill a few days ago.
And mounting all the disks is possible. So, I can mount the disk with the programs and I can open it from the server.

Can someone please help me, because tomorrow I need the programs for my students.

Thanks in advance,

Leen Biesemans

Re: NFS problem

T G Manikandan — Wed, 04 Feb 2004 04:58:47 GMT

How much memory you have on these workstations and on the one where the errors are popping out.

What is the usage of memory?

Re: NFS problem

T G Manikandan — Wed, 04 Feb 2004 05:06:07 GMT

RPC: Timed out

With the above messages I would also suggest to

1.Load all the neccessary NFS patches
2.use NFSv3 for nfs mounts

Re: NFS problem

Leen Biesemans — Wed, 04 Feb 2004 05:16:52 GMT

...and how can I do all this....I have not so much experience with HP-UX.

Thanks!

Leen Biesemans

Re: NFS problem

T G Manikandan — Wed, 04 Feb 2004 05:26:19 GMT

What version of HPUX you are running!

Re: NFS problem

Leen Biesemans — Wed, 04 Feb 2004 05:31:35 GMT

HP-UX 11.0

Re: NFS problem

Leen Biesemans — Wed, 04 Feb 2004 05:38:04 GMT

server: harddisk: 1,75 GB in total and 833 MB free
memory: 42252K real, 41604K virtual and 22192K free

other workstation: harddisk: 12,5 GB in total and 6,3 GB free
memory: 35832K real, 28036K virtual and 130660K free

Re: NFS problem

T G Manikandan — Wed, 04 Feb 2004 06:15:42 GMT

Can you post the output of

# echo phys_mem_pages/D | adb -k
/stand/vmunix /dev/kmem

Re: NFS problem

Leen Biesemans — Wed, 04 Feb 2004 06:22:08 GMT

vmunix is not a directory on my server.
So, I cannot do the command you wrote....vmunix is a file.

Re: NFS problem

T G Manikandan — Wed, 04 Feb 2004 06:25:01 GMT

That's ok.It should work.I was just trying to get the right amount of memory on the system.

Can you post the memory value from

SAM--->performance-->system properties

Re: NFS problem

Leen Biesemans — Wed, 04 Feb 2004 06:46:36 GMT

on the server:
Physical memory: 128.1 MB
Real memory:
Active: 15531.0 KB
Total: 49778.3 KB
Virtual memory:
Active: 19156.6 KB
Total: 46092.7 KB
Free memory pages: 2716 at 4KB/page
Swap space:
Avail: 256 MB
Used: 129 MB
Free: 127 MB

on the workstation:
Physical memory: 256.2 MB
Real memory:
Active: 26215.8 KB
Total: 46088.7 KB
Virtual memory:
Active: 21149.5 KB
Total: 33839.2 KB
Free memory pages: 29322 at 4KB/page
Swap space:
Avail: 512 MB
Used: 80 MB
Free: 432 MB

Re: NFS problem

T G Manikandan — Wed, 04 Feb 2004 07:01:22 GMT

For me it looks like the main cause of this error is the resource crunch on the machines.

1.check whether the nfs client mount is a hard mount and not a soft mount.

Hard mount keeps on querying the server even incase of timeout.

2.check the nfs version.you can make the vers 3 on both sides using vers=3 in the -o option of the mount command.

you can check the version using

server:
rpcinfo|grep

client:
netstat -m

Re: NFS problem

Leen Biesemans — Wed, 04 Feb 2004 07:47:09 GMT

After rpcinfo | grep nfs I get the following

100003 2 udp 0.0.0.0.8.1 nfs superuser
100003 3 udp 0.0.0.0.8.1 nfs superuser

But netstat -M has no information

Re: NFS problem

Geoff Wild — Wed, 04 Feb 2004 08:44:27 GMT

You say it was working fine a few days ago - maybe something has changed?

You can mount, and read the files - it's only starting daemons over NFS that's the issue?

What is the value of dbc_max_pct and dbc_min_pct?

kmtune |grep dbc

Looks like evrything is going into swap - as you only have 128MB of ram - in my experience, NFS hates to be in swap...

$ export UNIX95=1
$ ps -C nfsd -o comm,flags,pcpu
COMMAND F %CPU
nfsd 0 0
nfsd 0 0
nfsd 0 0
nfsd 0 0
...

If the flag (F) is 0 then the nfsd's are currently not in memory! Then the server has not enough memory / there's a memory leak

If dbc_max_pct is high, say 50 - try decreasing to maybe 20, or 10....

netstat -s -p udp (* overflows high ? Not enough nfsd daemons; start more in nfsconf *)

What's the server uptime? NFS is pretty finicky - if you have been stop/starting NFS several times - you may have a locking issue - which will be cleared up by a reboot...

Rgds...Geoff

Re: NFS problem

T G Manikandan — Wed, 04 Feb 2004 08:47:50 GMT

Just stop the unneccessary sevices on the machine and then reboot should free up most of the temporarily used resources.

the resources on your machines looks very low for which the application from nfs just struggles.

Also,another thought would be to check the network speed of both the machines
Make sure they match in speed/duplex.

Re: NFS problem

Leen Biesemans — Wed, 04 Feb 2004 09:01:15 GMT

I can mount and read the files on the other workstations, that's not the problem. I cannot start the daemon of the software that's on another workstation than the server.

After the command kmtune | grep dbc I get

dbc_max_pct 50
dbc_min_pct 5

After the command ps -C nfsd -o comm, flags, pcpu I get

nfsd 141 0
nfsd 1 0
nfsd 1 0
nfsd 1 0

After the netstat command I get the following:
0 incomplete headers
0 bad checksums
0 socket overflows

Rebooting the system is not a solution because I have tried it a few times in the past days and the problem stays the same.

The problem showed up when there was a main voltage shut down through some cause.