Description of problem: when rpc.statd is started with a -o option it is ignored and random port is used instead Version-Release number of selected component (if applicable): # rpm -q nfs-utils nfs-utils-1.0.6-70.EL4 How reproducible: Start rpc.statd with a -o option and argument, look at netstat to see what really happened Steps to Reproduce: 1. pkill statd 2. rpc.statd -p 31002 -o 31003 Actual results: # netstat -lep --numeric-ports |grep statd tcp 0 0 *:31002 *:* LISTEN root 861353 11065/rpc.statd udp 0 0 *:641 *:* root 861342 11065/rpc.statd udp 0 0 *:31002 Expected results: # netstat -lep --numeric-ports |grep statd tcp 0 0 *:31002 *:* LISTEN rpcuser 861567 11078/rpc.statd udp 0 0 *:31002 *:* rpcuser 861564 11078/rpc.statd udp 0 0 *:31003 *:* root 861556 11078/rpc.statd Additional info: This seems to work ok when using rpc.netstat from nfs-utils 1.0.9 (http://prdownloads.sourceforge.net/nfs/nfs-utils-1.0.9.tar.gz)
I've come across this too. Looks like upstream fixed it with this patch: http://marc.theaimsgroup.com/?l=linux-nfs&m=110891787507544&w=2 This doesn't apply to the RHEL4 source. I'll attach a patch that works for me.
Created attachment 139317 [details] Patch to fix statd -o option
*** This bug has been marked as a duplicate of 210283 ***