Nonstop error message on rx660 console: “rtable_main.c: unable to register”

[Adel_H]

A production RX6600 was plagued with nonstop messages that make impossible to use it.

                ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

MP MAIN MENU:

         CO: Console

        VFP: Virtual Front Panel

         CM: Command Menu

      SMCLP: Server Management Command Line Protocol

         CL: Console Log

         SL: Show Event Logs

         HE: Main Help Menu

          X: Exit Connection

[mp0014c29c1237] MP>  co

- - - - - - - - - - - - Live Console - - - - - - - - - - - -

rtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_main.c: unable to registerrtable_

                ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

I checked “sar –v”, syslog, searched for table mail.c, and stopped the audit system (audsys) without  any progess.  The server is up but I am hesitant to reboot it for maintenance purpose without knowing what this message means.  Hoping somebody have seen this message before or know what it means and how to stop it. 

Appreciate any feedback.

Adel

ahabashy@g6hospitality.com

[Dennis Handly]

I think the error message is really: rtable_main.c: unable to register

There is one old topic without a solution:

http://h30499.www3.hp.com/t5/System-Administration/q-concole-display-rtable-main-c-unable-to-register/td-p/3278776

google also finds it in some CDE source fragment:

CDE - Common Desktop Environment
svcmain.c

        if (!svc_register(udp_transp, ph->program_num, version, program,
                IPPROTO_UDP)) {
               (void)fprintf(stderr, "rtable_main.c: unable to register");

(And they messed up and forgot to add a newline to the end. ;-)

[Patrick Wallek]

Hi Adel,

No messages in syslog.log related to the times these messages are occurring?

Bryan and I were talking about this today and there is not much information anywhere about this message.

Does the 'kcusage' command show anything that appears abnormal for kernel parameters?

# kcusage

I know this is HP-UX 11.31... What other software is running on this system?

[Duncan Edmonstone]

I think Dennis is on the right lines here - a bit more digging and there's some indication that this error message is coming from rpc.cmsd which is the calendar manager service daemon in CDE.

It seems unlikely that you are actually using this tool (although you should check - it is part of the CDE Calendar Manager), so you might want to look in /etc/inetd.conf and comment out the following line:

rpc dgram udp wait root /usr/dt/bin/rpc.cmsd 100068 2-5 rpc.cmsd

then execute "inetd -c" to re-read the inetd.conf file

finally you might want to check whether you still have any instances of rpc.cmsd running and terminate them.

Probably wouldn't do any harm to see if you have the latest CDE Applications patch which has rpc.cmsd in it as well... that is PHSS_36407 - there are dependencies for that patch as well.

[Adel_H]

Thanks for the feedback. 

Dennis,  earlier I ran to the old link about this message without a solution but I will research the CDE source segment that you found and has a reference to this message.

Patrick, thanks for the reply and indeed the syslog has no reference to this message. kctune doesn't shaow anything out out of normal.  The server has tomcat, Informix, and HP-UX audit (audsys) which I temporary stopped.    

Ducan, inetd.conf had all rpc lines commented out except:

rpc xti tcp swait root /usr/dt/bin/rpc.ttdbserver 100083 1 /usr/dt/bin/rpc.ttdbserver.

BTW, we don't use CDE.  I searched for PHSS_36407 and it is not on the server.

Thanks,

Adel

[Matti_Kurkela]

If the message comes from the kernel, it should also appear in the output of the "dmesg" command.

If it doesn't, then it's probably created by an userspace process that has its standard output and/or standard error directed to /dev/console. In that case, "fuser /dev/console" should display one or more process IDs: the culprit should be among them.

[Adel_H]

Good point Matti.  dmesg didn't have the error.  fuser /dev/console led me to a list of pid's mainly of the database (informix) start up command which doesn't tell us much other than they are db processes. 

I am thinking though to request an outage to rebounce the database and see if this will stop the messages.

[Bryan_Payne]

Adel,

Since CDE is one possibility (because it does reference rtable_main.c in its source code)... and it was mentioned that it was an issue with the calendar... the latest CDE applications patch for 11.31 is PHSS_42341.  In the notes for this patch there is a reference for a problem with the calendar manager RPC service that was addressed in PHSS_41788 (Same patch family tree).

Can you look for the following patches:

Either one should do:

PHSS_42341 (Latest)
 or

PHSS_41788 Superceded

If instead you have PHSS_39409 or no patch at all... either of the two patches mentioned above may address your issue.

The other option was already mentioned... to simply disable it if you have no need of it.

[Adel_H]

The problem was resolved when I deleted these two lines at  /etc/inetd.conf :-

rpc xti tcp swait root /usr/dt/bin/rpc.ttdbserver 100083 1 /usr/dt/bin/rpc.ttdbserver.

rpc dgram udp wait root /usr/dt/bin/rpc.cmsd 100068 2-5 rpc.cmsd

looking closely I found out that these two line were commented out before I updating the server from 11.23 to 11.31 recently.  It seems the update process appended several lines to inetd.conf. 

Thanks a lot for the feedback and responses.  

In a side note, I also was running HP-UX system audit and I found out while I was working on this problem that update-ux put new audit configuration files.