STARTUP process hang

[Willem Grooters]

Open3D is, AFAIK, needed for some graphical packages, and known not to work with certain graphical cards (don't ask me for details, please)...

Extracted from the log (comments are mine):

The OpenVMS system is now executing the site-specific startup commands.
...
(So process 434 is started in SYSTARTUP_VMS.COM)
...
Starting TCPware(R) for OpenVMS Version 5.6-2 ...
%RUN-S-PROC_ID, identification of created process is 0000041D
%RUN-S-PROC_ID, identification of created process is 0000041E
%DCL-W-IVKEYW, unrecognized keyword - check validity and spelling
\UP\
( - error in TCPWARE startup?? Could that cause a problem?)
...
(then WBEM startup, I guess:)

This command procedure assists with starting the OpenVMS
management web agents after installation.

Run the subagents/mibs
...
%DCL-I-SUPERSEDE, previous value of WBEM$DCL has been superseded
Run the webserver
%RUN-S-PROC_ID, identification of created process is 00000432
Run the DCL show application
%RUN-S-PROC_ID, identification of created process is 00000433
%RUN-S-PROC_ID, identification of created process is 00000434

%AMDS-I-EXLOGFILE, Executing AMDS$AM_AMDS_SYSTEM:AMDS$LOGICALS.COM
...

If the process causing problems was indeed PID 434, could it be what is showing itself as "the DCL show application"? When running in startup, that may block OPA0 if it's waiting for input ("TYPE/PAGE"? READ SYS$COMMAND?).
It might be a good idea to scan *.COM for the this string (just a hint)

Willem

[Wim Van den Wyngaert]

1) Syntax error during startup of TCPWARE.

Starting TCPware(R) for OpenVMS Version 5.6-2 ...
%RUN-S-PROC_ID, identification of created process is 0000041D
%RUN-S-PROC_ID, identification of created process is 0000041E
%DCL-W-IVKEYW, unrecognized keyword - check validity and spelling
\UP\

2) PID 00000434 died when you stopped STARTUP. In startup.log :
%DCL-I-SUPERSEDE, previous value of WBEM$DCL has been superseded
Run the webserver
%RUN-S-PROC_ID, identification of created process is 00000432
Run the DCL show application
%RUN-S-PROC_ID, identification of created process is 00000433
%RUN-S-PROC_ID, identification of created process is 00000434 <==

So, it was WDEM that caused it. Whatever it may be (I thought it was WEBES which I saw misbehaving on our nodes regularly).

Wim

[Wim Van den Wyngaert]

Oeps, Willem was first.

Wim

[Wim Van den Wyngaert]

Could it be that /detached was missing when the RUN was done of the 3rd process ? The other 2 didn't terminate when startup was stopped.

Wim

[Martin Vorlaender]

>>>
%DCL-W-IVKEYW, unrecognized keyword - check validity and spelling
\UP\
<<<

This was just a comment missing its "!" (in TCPWARE:ROUTING.COM). As it's -W- it shouldn't do any harm.

>>>
PID 434, could it be what is showing itself as "the DCL show application"?
<<<

AFAIK, no. Process WBEM$DCLSHOW - this is for WBEM DCL SHOW (CPU, CLUSTER, DEVICE, ...)support.

>>>
So, it was WBEM that caused it.
<<<

Menawhile, I think so, too. The course of action is

STARTUP.COM -> @TCPWARE:STARTNET.COM -> @TCPWARE:SERVERS.COM -> @SYS$SPECIFIC:[WBEM]WBEM$STARTUP.COM and $RUN/PROCESS=HR_MIB SYS$SYSTEM:TCPIP$HR_MIB
and then
STARTUP.COM -> @SYS$STARTUP:AMDS$STARTUP

Guess I'll move the WBEM startup out of TCPware startup and back into SYSTARTUP_VMS.COM.

>>>
Could it be that /detached was missing when the RUN was done of the 3rd process
<<<

I always thought that RUN/PROCESS implied /DETACH - I could be wrong. I'll add it to the HR_MIB startup.

Thanks, all. I'll keep you updated.

cu,
Martin

[Robert Gezelter]

Martin,

I would not recommend moving it to SYS$STARTUP. If there is a problem with the WBEM startup, it will cascade STARTUP_VMS.COM (which would give us a WHOLE different set of problems)

Instead, create a file that can be found using the SYS$STARTUP logical (either in the node specific or System common [SYSMGR] or [SYS$STARTUP] directories; a question of personal taste question). Then add it to the STARTUP database FOLLOWING the phase that invokes STARTUP_VMS.COM (which is in phase LPBEGIN). The phase following LPBEGIN is LPMAIN. So:

- create SYS$COMMON:[SYSMGR]WBEM$STARTUP.COM as follows:

$ FILENAME = "SYS$SPECIFIC:[WBEM]WBEM$STARTUP.COM"
$! As insurance against accidents, check that the file exists.
$ IF F$SEARCH(FILENAME) .NES. ""
$ THEN
$ @'FILENAME'
$ ELSE
$ WRITE SYS$ERROR "''FILENAME' not found"
$ ENDIF

- add it to the Layered Products STARTUP database

$ MCR SYSMAN ADD FILE WBEM$STARTUP.COM/PHASE=LPMAIN/MODE=SPAWN/NODE=

If there is a problem, the fact that it is a SPAWN will isolate the problem from the rest of the startup.

- Bob Gezelter, http://www.rlgsc.com
$ MCR SYSMAN ADD FILE

[Martin Vorlaender]

Hi,

a big hardware rebuild of the computer room gave me the opportunity to reboot the machine a few times, testing what the f**k the cause was.

And the loser was...

TCPIP$HR_MIB !

(which I have to start explicitly when using WBEM under TCPware, see the Insight Management Agents Installation instructions at http://h71000.www7.hp.com/openvms/products/mgmt_agents/install.html )

It seems that a simple RUN is not enough to fully detach the process. I replaced it by RUN/DETACH and have a working OPA0: after startup.

Thanks, everybody, for the insightful hints leading me to the solution.

cu,
Martin