| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 463.1 | Workaround | VISA::BIJAOUI | Tomorrow Never Knows | Tue May 05 1987 18:29 | 4 |
|
Why don't you check if BANNER is runnin' or not ?
Pierre.
|
| 463.2 | | EDWIN::WATERS | Lester Waters | Tue May 05 1987 18:45 | 20 |
| Well, that is a great idea, but I haven't been able to figure
out how to get the PID for BANNER. If BANNER was started from
SYSTEM account, then F$PID() will not return a PID number that
you can check, unless you have priv's. This is going in SYLOGIN.COM
What I tried:
$ CONTEXT = ""
$loop:
$ pid = f$pid(CONTEXT)
$ if pid .eqs. "" then goto done
$ pidname = f$getjpi(pid,"PRCNAM")
$ write sys$output "Process ''pid' is ''pidname'"
$ goto loop
$done:
Of course, if f$pid() let me go through all pid's, then I could
check for BANNER running. Is there a function that will return
the PID or TRUE/FALSE given a process name?
|
| 463.3 | | ALBANY::KOZAKIEWICZ | You can call me Al... | Tue May 05 1987 19:48 | 12 |
| Does the mailbox writer hang every time, or only after a while? The reason
that this usually happens is that not reading the mailbox eventually consumes
whatever dynamic memory was set aside for the device. This can be verified
by looking at the process of the writer - it will be in a RWAST (resource
wait) state. If I was doing the write from a program, I would be doing
2 things. First, use a QIO instead of a QIOW. Second, call the system service
SYS$SETRWM. This will cause the QIO to return an error instead of waiting
for the resource to become available. I don't know what your options might
be from DCL. Since SYS$SETRWM sets the mode for the process, perhaps you
can write a short program to set this and run it inside your command
procedure. Hope this helps somewhat....
|
| 463.4 | | CAFEIN::PFAU | Now where did I leave my marbles? | Tue May 05 1987 22:51 | 6 |
| F$GETDVI(device,"REFCNT") might help. It tells you how many channels
are assigned to the device. If the banner program is normally the
only one having a channel assigned, the reference count will be
1 if it is running, 0 if not.
tom_p
|
| 463.5 | Hmmm... but... | EDWIN::WATERS | Lester Waters | Wed May 06 1987 02:05 | 8 |
| F$GETDVI(device,"REFCNT") is a help, but the reference count may
still be non-zero (as in the case where DISPMSG - display message
thing) is running. So a non-zero refcnt could mean that BANNER
is running *or* BANNER died and the dispmsg is still running. It
is not guaranteed that BOTH (REFCNT=2) will always be running.
Interesting challenge...
|
| 463.6 | hack, hack, hack,... | CHOVAX::YOUNG | Back from the Shadows Again, | Wed May 06 1987 03:34 | 17 |
|
$!
$! Determine if BANNER is running
$!
$ Banner_running := TRUE
$ Temp_name := 'f$pid(0)'
$ Temp_file = Temp_name + ".temp"
$ Show System /output='temp_file'
$ Search 'temp_file' BANNER /nooutput /nolog /nohead
$ If ($severity .nes. "1") then $ Banner_running := FALSE
$ Delete 'temp_file';0
$!
$! Now take action depending on whether Banner_running is TRUE or FALSE
$!
$...
-- Barry
|
| 463.7 | Funny I can tell with privs off | FROST::HARRIMAN | Too much talk, Small Talk | Wed May 06 1987 13:30 | 12 |
|
Just 'cause you got no privs doesn't mean you can't find out if
a process exists via DCL. F$GETJPI is nice enough to tell you "%W,
no privilege for attempted operation" if the process exists, and
"%W, no such process" if it isn't there in the first place. Since
the errors are still trappable (and you can get $STATUS back for
them at least this version of VMS) you could get away with your
original try as long as you inhibit message output during SYLOGIN.COM.
just a suggestion
/pjh
|
| 463.8 | how about SPAWN/NOWAIT | VIDEO::OSMAN | type video::user$7:[osman]eric.six | Wed May 06 1987 14:45 | 10 |
| Two suggestions to avoid hanging on mailbox operations from DCL:
o Experiment with the /TIMEOUT=0 switch on dcl's READ command.`
o Use SPAWN/NOWAIT @WRITE_TO_MAILBOX instead of just @WRITE_TO_MAILBOX.
That way, if it hangs, so what. It's just the subprocess that hangs,
not your entire terminal.
/Eric
|
| 463.9 | SPAWN/NOWAIT & undesirable side effect | UTRTSC::ROBERTS | UTO-23.9b, DTN 838-2679 | Thu May 07 1987 08:14 | 3 |
| In V4.3 and V4.4 there were "certain disadvantages" with the
SPAWN/NOWAIT DCL command. I don't know if V4.5 fixed the problem
or not.
|
| 463.10 | You can play with resource wait from DCL | MAKO::NELSON | Jeff E. Nelson | Thu May 07 1987 12:58 | 28 |
| RE: a couple of replies back.
You can enable and disable resource wait mode from DCL. Here's
the info from HELP:
SET
PROCESS
/RESOURCE_WAIT
/RESOURCE_WAIT
/NORESOURCE_WAIT
Enables or disables resource wait mode for the current process.
If you specify /NORESOURCE_WAIT, the process will receive an error
status code when system dynamic memory is not available or when the
process exceeds one of the following resource quotas:
Direct I/O limit
Buffered I/O limit
Buffered I/O byte count (buffer space) quota
Doesn't look like it will help, but it's worth a try.
JENelson
|