NFS server not responding - NFS server OK - I'm not

[Igaduma]

Hi all,

For some dark & mysterious reason one of my desktops is behaving bad.
I'm nfs mounting a drive from a server (sunOS 5.7) to numerous other desktops (sunOS 5.7), never had an nfs problem untill now.

1 desktop is complaining about:
NFS server not responding, still trying
and after a while
NFS server OK

It used to work for months without a problem and nothing has changed in the config.

I checked the good 'ol Rwin & settings related to the network and they seem fine, fairly default settings as they appear on all other desktops.
Server's HappyMEal0 (hme0) related settings seem also OK
rebooting doesn't help.
No congestion is appearing on the network (or so i'm told by IT) and metatool doesn't report outrageous drive stats.
vfstab isn't any different from other vfstabs on "working" or "non complaining" desktops.
the hard drive that needs to be mounted is a concat dev external 2 x 54 gb discpacks, all set up & working perfect on the serverside.

argh..

Thanks for *any* illumination ascii people


Iga

 

[ayjaycee]

If it's just the one workstation that's suffering (and you, of course ;-)), then what's different about it?

Is it on a separate LAN/Segment/sub-net/hub/switch/patch panel than the working clients?

Does ping & traceroute show higher times than for the other clients?

Any other dark & mysterious networking problems appeared on the client?

http://www.tanda.f2s.com

One by one, the penguins steal my sanity.

 

[Igaduma]

Hi there,

I checked the ping & traceroute stats and they are virtually the same.
0.2 - 0.3 ms on The Infected desktop & the other desktops
from the server-side it's the same story, no indications on a bad network.
Although it *could* have something to do with the network.
I should traceroute it when the nfs server is not responding.
Although I already did that with ping, not with traceroute.
Didn't indicate any problems.

When you come to mention it, yes, something did happen on that desktop.

When logging in on CDE some while ago it hanged and finally went into openwindows, i repaired that by recopying the initialization files to a certain user who was experiencing that problem. Didn't occur anymore afterwards, although I suspect the CDE hanging problem was directly caused by the nfs server not responding, as the user home dir was on that mounted drive.

The Saga Continues...

iga




"Things should be made as simple as possible, but not any simpler" - Einstein

 

[Thief]

have u checked if there is any ip conflict in the network. that might also be one of the reasons y CDE fails. have u added any new machines or made any nodename/hostname/ip changes to any of the machines....?? Thief................
(I think the surest sign that intelligent life exists elsewhere in the universe is that none of it has tried to contact us .)

 

[Igaduma]

Hiya,

yeaup, no IP clashes, not in adm message

"Help Someone with mac xxxxx is stealing my IP adress"

IT here has been helpfull and assigned a range of IP adresses to me, so there is no risks of clashes.

Again, i have to say, nop, nothing changed for the worse.
Some machines have been added but care has been taken to give them unique names & IP.
Since the errors started showing i didn't change anything on the config of any machines or server.

And the search continues...


"The pleasure of being mad is only known to a madman"
shakespeare.

 

[trunix]

Have you used the nfsstat utility ?
This will tell you exactly what is happening, but you will need to study the man pages to understand and interpret the output.

 

[vguneta]

Hi
Can u do the foll.. (log in as root)...

1. On the server:

# cd /etc/init.d
# nfs.server start

2. On the server:

# cd /etc/init.d
# nfs.client start

Best Regards,
- Vijay Guneta

 

[Igaduma]

Hi list,

nfsstat is giving some intresting information although
I have to look at the man pages first to fully understand its output.

in the nfs client part there are 16 calls of which 10 are logged as bad calls...unless i'm high on chocolate & coffee this can't be good...

coffee, man pages & headphones .. ..

on and on we go!

 

[bluedevils]

what is your NIC duplex setup as? If there is a mismatch, you might not see anything wrong with pings and traceroutes.

 

[Kegnut]

What ever happened to this one? Cause now I am having the same problem. Really getting annoying. Anyone else run into this before?

Thanks in advance for any help!

**Maybe it's a Christmas thing***

J

 

[Feline]

check hardware

1. is the lan cable ok? intermittent problems may be coming through loose crimping or whatnot. make sure the user isn't twiddling with the cable. reminds me of a site i went to years ago that layed those spiky chair mats over the co-ax lan cable!

2. try a different switch port for that machine. flaky switch ports have been known to happen.

3. replace the NIC. maybe swap it with another machine and see if the problem migrates to the new machine (don't tell the users tho)

i used to get a lot of NFS Server Not Responding Messages on my LAN but thats because some brainiac with more clout than me decided a wireless LAN (11mb shared bandwidth) would be cool way to transfer NFS and MVFS traffic among 25 pc's doing compiles. heh. i ended up cabling the friggin site myself.

every now and then i get some NFS timeouts between my local servers and the NIS+ masters on the other side of the country. i am running a 128kb frame relay (only shared between 50 people :/ ) so the problems are mostly intermittent and clear themselves up.

when i was doing my research as to why my wireless lan sucked so much ass i discovered how intolerant NFS is of dropped packets. changing the block size helped somewhat but then the air conditioning would click on and clobber the whole lan anyways

i am running solaris 8 tho so there may be a little known issue with solaris 7. you have checked out sunsolve.sun.com yeah?

good luck

 

[Annihilannic]

I use an NFS mount across a WAN which experienced a similar problem due to loss of packets; switching from UDP to TCP on the NFS server did the trick.

Doesn't explain why it suddenly started happening to Igaduma's machine though. Annihilannic.