| Title: | Digital PostScript printers and their associated software |
| Moderator: | REGENT::LASKO HER |
| Created: | Wed Jan 24 1990 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 7230 |
| Total number of notes: | 31971 |
Customer has a large Vax cluster running OVMS 6.2 and has noticed that
the lps daemons keep dying. They were running LPS V5.0-17 and now have
upgraded to LPS V5.1-15. They have seen this problem from both kits.
There are not any errors recorded in the event logs or in OPCOM. The
process is just gone. They have to keep restarting the processes.
Using a "watcher" utility that are now watching for the daemon
processes. If one dies this watcher will restart it. Now the daemons
are even more critical as they now have enabled the "font faulting"
feature for postscript level II for their printers.
I asked them to look into process killers and they are running one.
They said that this process killer will not touch any system processes.
Is there a trace feature for the daemon process to see who, what kills
it?
Renis
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 7166.1 | check that watcher! | REGENT::GALLAGHER | Tue Apr 22 1997 19:06 | 21 | |
Are they sure their watcher utility isn't at fault? How does the
utility decide what processes to kill off? Although the daemon
processes are started by "system" since the command procedure is run
from the system startup procedure, the daemon processes have a userid
of [LPS, LPS$SERVER] . Could this be a problem?
Could they modify their watcher program to send to output info about
the process being killed?
I think if a process is killed on VMS by a STOP command, the process is
not aware of what killed it. The daemon code seems to report those
conditions under which it kills itself, though I've never inspected the
code to see if that is true for all cases.
I would ask to see their watcher program, it just sounds suspicious to
me!
Another clue which might indict the watcher:
Do the daemons seem to stay up and running as long as log messages are
being written to the log file? They could make this obvious by turning
on "reliable" accounting.
| |||||
| 7166.2 | We just run the startup | TAY2P1::HOWARD | Whoever it takes | Wed Apr 23 1997 18:04 | 7 |
Our "watcher" is just @Sys$Startup:LPS$Startup_Services. This runs
every 30 minutes and seems to do the right thing. I have this running
on 3 systems and they all seem happy.
I wonder what the customer runs for a watcher.
Ben
| |||||
| 7166.3 | I used the wrong term | REGENT::GALLAGHER | Thu Apr 24 1997 15:37 | 9 | |
OOPS! I meant to say they should look at the "PROCESS KILLER". I think
that's the problem.
Using lps$startup_services to restart daemons won't get at the
underlying cause.
-cg
| |||||