Recently I had difficulty in restarting portmapper service following a failure in RPC listeners.
Background info: this is a nis server, which doesn't export out any file systems, i.e. server doesn't work as NFS server.
So do I expect to have 'portmapper' running on this server?
why did the portmapper failed abruptly?
Appreciate clarification and answers.
portmaper daemon fails
Re: portmaper daemon fails
Yes you need RPC services for NIS.
As to why the daemon died, I guess you need to post the relevant logs.
As to why the daemon died, I guess you need to post the relevant logs.
Re: portmaper daemon fails
Thanks aks for taking time to give me a feedback.
I guess what I would like to know is if my /etc/exports file is empty, i.e. no filesystems are being exported out and no nfs server running do I still get the 'portmapper' started when I start nis?
I looked at the logs, and it doesn't produce any clues.
Couldn't stop and start rpcbind, failed.
Then I restarted the server, then it started gracefully, in results I did see rpcinfo -p output with 'portmapper' running.
My feeling is some sort of cache maxed out. Could this be the socket saturation?
In my server cat /etc/sysconfig/nfs value for RPCNFSDCOUNT is 8. I think that is the default.
Could this be a lead? Should I increase the value from 8-to-16?
But this is not a nfs server, so does it still required to bump up the RPCNFSDCOUNT to larger number?
Appreciate your feedback.
I guess what I would like to know is if my /etc/exports file is empty, i.e. no filesystems are being exported out and no nfs server running do I still get the 'portmapper' started when I start nis?
I looked at the logs, and it doesn't produce any clues.
Couldn't stop and start rpcbind, failed.
Then I restarted the server, then it started gracefully, in results I did see rpcinfo -p output with 'portmapper' running.
My feeling is some sort of cache maxed out. Could this be the socket saturation?
In my server cat /etc/sysconfig/nfs value for RPCNFSDCOUNT is 8. I think that is the default.
Could this be a lead? Should I increase the value from 8-to-16?
But this is not a nfs server, so does it still required to bump up the RPCNFSDCOUNT to larger number?
Appreciate your feedback.
Re: portmaper daemon fails
NIS is not NFS. While they come from the same (sadly missed) company (Sun), they are not the same thing. They use some of the same technologies (RPC), nut an NFS doesn't have to be a NIS server and vice versa.
The idea of portmapper is that RPC calls can dynamically map ephemeral ports (short lived) ports to help the system scale. Connect to a well known port and then "negotiate" a port for the services required.
Really? Type cat /etc/exports to find out!I guess what I would like to know is if my /etc/exports file is empty
Why? You give absolutely no clues.Couldn't stop and start rpcbind, failed.
The idea of portmapper is that RPC calls can dynamically map ephemeral ports (short lived) ports to help the system scale. Connect to a well known port and then "negotiate" a port for the services required.
I don't know what that is! Do you mean you have more then the 65535 (minus 1024, although modern systems like to minus +-32768 from that number, so more than 32K) sockets available in constant use?Could this be the socket saturation?
Re: portmaper daemon fails
Thanks for the update. I know for sure /etc/exports has no entries.
Does it make any sense in increasing the rpcnfsdcound number to 16?
Does it make any sense in increasing the rpcnfsdcound number to 16?
Re: portmaper daemon fails
Do you mean rpcnfsdcount?
That will increase the number of threads available for NFS.
Is that what you want to do?
That will increase the number of threads available for NFS.
Is that what you want to do?