Operating System - HP-UX
1753522 Members
8883 Online
108795 Solutions
New Discussion юеВ

Re: Running Two NFS Packages

 
Robert Gamble
Respected Contributor

Running Two NFS Packages

When I took my MC/SG I class last year, I thought my instructor had mentioned that you could only run 1 NFS package per cluster.

My customer would like to add a CIFS (a samba-like NFS type application) package to a cluster that already has an NFS package.

Has anyone tried this before ?
Is this possible ?

TIA
3 REPLIES 3
melvyn burnard
Honored Contributor

Re: Running Two NFS Packages

Rob, to my knowledge we have only ever supported one NFS package, but with the old versions of the NFS Toolkit there were reasons for this, due to the way it set up and started various rpc's etc.

The newer toolkit looks like it might be able to handle more than one, but there are a few things to consider. In the newer scripts, one of the things that happens is that we attempt to kill and restart rpc.statd and rpc.lockd when the package goes down. How this would affect another running NFS package, I am not sure.

I guess the simple answer would be to say combine the two as one package.

My house is the bank's, my money the wife's, But my opinions belong to me, not HP!
Byron Myers
Trusted Contributor

Re: Running Two NFS Packages

Robert, SAMBA doesn't use NFS. It has it's own daemons (smbd,nmbd). In one of my Service Guard packages, I run both NFS and SAMBA with no problems. Hope this helps.
If you can focus your eyes far and straight enough ahead of yourself, you can see the back of your head.
Wodisch
Honored Contributor

Re: Running Two NFS Packages

Hello Robert,

first: NFS and Samba=CIFS use different RPC (Remote
Procedure Calls: NFS uses Suns RPC, portmapper at
port 111, versa SMB (M$ Server Message Block) uses
DCE RPC, portmapper at 135), hence there is no conflict.

second: if you keep portmapper/rpcbind, and the other
NFS daemons out of your MC/SG packages (but running
on all your nodes), you just have to modify "/etc/exports"
and "/etc/xtab" (using "exportfs -a") and sending signal
HUP to the status and lock daemons.
That is quiet different form the training scenario, and from HP's NFS package, but it should be enough.

HTH,
Wodisch