|
Description
|
When nfsd starts up, it forks a child process to continue
initialization and then immediately exits (nfsd.c:423). However,
there are a number of failures the child can experience that
result in no daemons running. Under these circumstances, the
SMF service network/nfs/server:default is in the online state
even though the service failed. The only record of the failure is
in the system log, which is also bad in the SMF universe.
Instead of exiting, the parent process needs to wait for a
notification from the child that it has started successfully, or
alternately read an error code or message from the child and display
it on standard error so that it ends up in the service log rather
than just the system log. Finally, the nfs-server.sh start method
script also needs to examine the exit status from nfsd and act on it
so that the instance can be placed in the maintenance state pending
administrative action.
mountd suffesrs from the same problem and can use the same solution.
|
