NFS - After an Upgrade from 10.20 to 11.00

[Brian Killeen_1]

I have recently upgraded a machine from 10.20 to 11.00. My problem is that the shares are no longer working. They are still in the /etc/exports file. They are showing as shared. The nfsconf seems to be OK. All the correct 11.00 daemons ae running but the shares are not visible on the network. Has anyone seen this before? all ideas welcome.

[Paula J Frazer-Campbell]

Brian

I would stop and restart the nfsd as a first fix attempt.

Another option is to move the exports file to a bak and the do a completly fresh export - I would use SAM for this.


Paula

[Paula J Frazer-Campbell]

Brian

See man nfsd - remember to start the correct number of deamons.

Paula

[Brian Killeen_1]

Thanks Paula - I has restarted the nfs via
/sbin/init.d/nfs.server stop & then start. This had no luck. Regarding the daemons all are running OK.

rpcbind
nfsd
rpc.mountd
rpc.pcnfsd
rpc.lockd
rpc.statd

I tried your suggestion of renaming the exports and recreating the mounts via SAM to no avail unfortunately. I did though find a patch called

PHNE_22288 B.10.00.00.AA NFS/NIS cumulative megapatch

This looks like a legacy from the upgrade - I might try removing this and see how I get on. Any other ideas are always welcome - Thanks for your help so far.

Brian

[Brian Killeen_1]

I have just noticed that the process

rpc.mountd is running at almost 100% CPU usage.....I have stopped and restarted it to no avail.....?????? Might have something to do with it..

[Paula J Frazer-Campbell]

Brian

Have you tried a reboot.


Paula

[Brian Killeen_1]

Reboot has been tried again to no avail...Could it be anything to do with the versions of NFS?

[Paula J Frazer-Campbell]

Brian

11.00 has by default ver 3 and 10.20 ver 2.

Are the other servers 10.20?

If so you can down load v3 for 10.20.

Paula

[Paula J Frazer-Campbell]

Brian

I have just spotted the word "UPGRADE"


Did you upgrade this 10.20 server?

If you search the forums you will find many many threads advising a fresh install as opposed to an upgrade.

I heard of a HP course where 12 students tried an upgrade - qty 10 had problems.


If you did upgrade I would seriously think about doing a fresh instal of 11.00.

The forum has many threads on how best to do this.


Paula

[Mark Vollmers · Accepted solution]

Brian-

are you using NIS as well? When I switched over from 10.20 to 11.0, I forgot that we had NIS setup, and I couldn't get my shares to share until I went back in and setup the master server, etc. Just a thought. (never quite figured out why the one needed the other, but I never really looked into it either)

Mark

[pap]

Hi,

1. Check that your machine is correctly set up for the network.
2. Execute following commands
exportsfs -a
showmount -e
3. If they are executing correctly that means your machine is set up correctly for NFS.

4. Re check the /etc/rc.config.d/nfsconf file.

NFS_CLIENT=1
NFS_SERVER=1

if you are using automouter please check that options as well. Check if NIS is running properly if you are using it.

5. Try to run following commands
nfsstat -n
nfsstat -r
This will give you info regarding nfs system.
You will have idea regarding FNS running successfully or not.

Hope this helps.

-Piyush

[Brian Killeen_1]

Mark,

I think you could be onto something with the NIS .... I am awaiting the machines availability to remove that PHNE_22888 patch which relates to NIS/NFS

Pap,
Cheers for the help - All as you ask is as it should be and all reporting correctly....

[pap]

If possible copy following files from your old server (10.20)to new server (11.0).

/etc/services
/etc/protocols

in /etc/services look for the following entries
and make sure that it is not marked.

nfsd-keepalive 1110/udp # Client status info
nfsd-status 1110/tcp # Cluster status info
nfsd 2049/udp # NFS remote file system

Hope this helps.

-Piyush.

[Brian Killeen_1]

pap,

I cannot copy the files over as it is the same machine and have already upgraded it...Regarding the /etc/services there are four not three entries in there

There are two nfs entries....One for UDP and one for TCP. Could this be causing the trouble...?

[pap]

Hi,

Yes, it requires all 3 entries in it.

I have checked my all NFS servers and all of them having these entries.
So, just put those entries and try it then how it goes.
Hopefully your problem should resolve.

Thanks,
-Piyush.

[Brian Killeen_1]

sorry pap - I am confusing you with my answers - All three entries are already in there - I was saying that in fact there were four enties in there - ie.

nfsd-keepalive 1110/udp # Client status info
nfsd-status 1110/tcp # Cluster status info
nfsd 2049/udp # NFS remote file system



and

nfsd 2049/tcp # NFS remote file system

[pap]

Hi,
Sorry about the confusion.

Please remove any one of two identical entry from /etc/services file.

Thanks,
-Piyush

[Brian Killeen_1]

I have been having another look and if I run

/usr/bin/rpcinfo -p

it is showing

portmapper as a running process - My understanding is that this is a 10.20 process replaced by rpcbind in version 11.00

Does anyone know any way I can replace this