Problem
I get the following message on my screen:
WARNING: NLM: RPC call failed: RPC error: xxxxxx, errno
where 'xxxxxx' could be either of the following:
rpc_mount failed
RPC_PMAP_FAILUREor, on SCO OpenServer 5.0.0:WARNING: NLM: RPC call failed: RPC error: RPC_PMAPFAILURE, errno 0
or, on SCO OpenServer 5.0.2:WARNING: pormapper on server [IP] is not respondingThese messages are found verbatim in /usr/adm/syslog.
Note: This is not a fatal error, but only a warning. The message
is generated when NFS is started and the "lockd" command is invoked,
which occurs in the startup rc scripts, or if NFS is started
manually. Cause
A remote procedure call (RPC) query, involving the network lock
manager (nlm) and the portmapper, failed. This may happen when
the machine's IP address and the entry in the hosts file don't
agree. lockd attempts to connect back to the host by resolving
the hostname. If the address returned doesn't match the real
address, the connection fails.It could also be that the NFS server is down or its portmapper is
dead or hung. The problem may be evidenced if the network card
has yet to be configured (when NFS is chained only to the TCP/IP
of the loopback driver), or if the network card is reconfigured.Another possible cause of the problem is the timing of when the
various daemons are started using the rc2 startup scripts when
booting up to multiuser mode. This timing issue has been noted
on MP systems.Also, when a user changes the IP on an interface that has already been configured using netconfig, the /etc/hosts file gets updated with the new IP, but the old IP is not removed.
Solution
There are several things you can check to determine the cause and resolve the problem:
A. The portmapper is not running.With the network functional, "ps -ef" output should include
"portmap" and "lockd" daemons. To make sure that the client
and server network connections still work, from the client, run:# rpcinfo -p <servername>The rpcinfo output should include the portmapper and nlockmgr programs. If the portmapper (portmap) is not running on the server, you will get a message similar to:
rpcinfo: can't contact portmapper: RPC: Remote system error - Connection refused
To restart the portmapper, stop and restart NFS as:
# nfs stop
# nfs startB. /etc/hosts needs adjustment.Here are some of the specific ways that the /etc/hosts would cause the problem along with the solutions to each:
1. Duplicate entries in /etc/hosts.Use your favorite editor to remove the duplicate entries.
2. Inconsistent I/P address between what is specified in /etc/hosts and what "ifconfig -a" shows.
Use your favorite editor to change the IP address of the machine to match what "ifconfig -a" shows.
3. Missing entry in /etc/hosts.Use your favorite editor to add the entry for the network
connection to the /etc/hosts from the information provided
by: "ifconfig -a", "uname -n", and "hostname". Also, be sure
that you have an entry for localhost defined as:127.0.0.1 localhostNOTES: Acronyms defined:RPC=remote procedure call
PMAP=portmapper
NLM=network lock manager See also
Networking Guide, Chapter 15: Configuring the Network File System, Troubleshooting NFS, sections:
- Remote mount failed; - Troubleshooting failed lock requests
/usr/adm/syslogManual pages from SCO OpenServer Networking Reference:
- mount(NADM) - rpcinfo(NADM) - statd(NADM) - lockd(NADM) - hosts(SFF)
ta105779, "OpenServer 5, How to change your server's IP address, netmask, or broadcast."