Page 1 of 1
Startup silently fails on queue errors
Posted: Tue Aug 22, 2023 1:48 pm
by tlovern
was having a problem where startup did not complete and SSH never started, leaving me with only the console port.
by inserting a bunch of "write sys$output" statements in the systartup_vms.com file, I was able to track it down to starting the sys$print queue in error (it had been made into a generic queue with autostart).
The procedure aborted at this point with NO error messages or informational text.
Don't know where to put this, but startup should not fail silently....
I realize the fault was entirely mine for having a bad statement in the procedure.
Re: Startup silently fails on queue errors
Posted: Tue Aug 22, 2023 3:10 pm
by cct
Errors in SYSTARTUP_VMS were always like this!
I recall a DEC engineer saying that you should reboot after ANY change in it - not easy on VAX that took 30 minutes (min) for a reboot, in a production environment...
You can set verify in startup, and log to a file
Chris
Re: Startup silently fails on queue errors
Posted: Tue Aug 22, 2023 4:35 pm
by bobwilson
@tlovern
Do you have a reproducer?
Re: Startup silently fails on queue errors
Posted: Tue Aug 22, 2023 6:45 pm
by tlovern
I don't think I do at the moment, but I can probably re-create....
basically when I first brought up VMS I created a single print queue and used a raw telnet printer. This was my sys$print queue. This worked fine, and was basically a stop-gap until I could re-visit printing.
Went back to revisit printing:
I installed DCPS and created a LPD queue on my networked printer Then I modified sys$print to be generic with autostart on, with it pointing to my new queue.
I left the start/queue sys$print in my startup_vms file, and this is where it silently died.
like I said, entirely my fault. It would have been nice for the "start/queue" to complain loudly.
Re: Startup silently fails on queue errors
Posted: Mon Sep 04, 2023 1:10 pm
by tlovern
so, my old-school solution was to simply put:
"write sys$output ===== entering <section name>"
and
"write sys$output ===== leaving <section name>"
wrappers around startup code I touch.
very old school, but effective - I know I'll fat finger something again.