I have a script on my NIM Master that runs with crons and performs mksysb backups of all NIM Clients once a week.
There is a problem with one of the clients as it stucks and never finishes the mksysb backups.
I tried to unmount the NFS file system created by the NIM in order to perform the mksysb but i cannot do it.
I killed all processes from NIM Master and client about the nim operation but i still cannot unmount the NFS.
I even tried umount -f but i still get the message below:
NFS server <NIM> not responding still trying
i don't know what else to try. I searched through smitty the nfs filesystems for this server but it doesn't appear anywhere. I cannot understand why it still try to mount it.
command to perform these backups and it is the client that is not able to unmount.
To debug - you could start with using rpcinfo -p <NIM> | grep nfs to see if the nfs demons are recognized.
The other question I would have is whethr this client has more than one IP address and/or nim route back to the nim server. If it is, the way I debugged the interface being used by default is the telnet/ssh to the nim server (use the <NIM> name as hostname, and then after login, use the command "who" to show which IP address/hostname is recognized as originating IP address.
Compare these values to what a "working" client returns.
My gut feeling, if I read corrently it is the same client that fails, is that there is a configuration difference between this client and the one that succeed.
is the one that i use to perform the mksysb backups in my script
you are right i use more than 1 ip addresses on this client. i compared
characteristics of the client throught smit nim and i noticed that on the
field Network Adapter Hardware Address the value is set to 0 and also
in the field Network Adapter Logical Device Name the value is set to ent.
I tried to correct these values but i get the error
0042-023 m_chmac: "sem_nim_network e_rea " is not a valid NIM interface stanza
i don't know why i get this error as i have the same values in the
respective field of the "correct" client
When we had physical servers, we had the same error and it ended up being caused by our switch being configured to 1000 full and our adapter was set to auto/auto. Once we changed our adapter settings, the error went away.
Also, look in your rc.net and rc.tcpip for any special customizations. We had one server with an "/usr/sbin/ifconfig..." line in it that should not have been there.
A "simple" solution, at least to try, is to delete the machine definition in the nim server, then on the client, use "smitty nim" to recreate the client.
If the nim master is missing a network resource, that is, thinks it is, you should get a message, and/or nim master will make one for you.
When you telnet/ssh to the nim master IP address, does "who" give you the expected hostname/IP address for the client?
sorry i didn't mention that before. yes i tried to ssh to NIM from the client and did who and i get the correct output.
in the meantime i already performed a reboot on the client just to clear it from the NFS.
i guess it would be a good idea to re-configure the server as NIM client.
it is strange though as it worked before and i experience this problem just this week.
but i think this is what i am going to do.