T.R | Title | User | Personal Name | Date | Lines |
---|
4259.1 | Fixed in next release | IOSG::MAURICE | Six Programmers in search of an analyst | Wed Jun 15 1994 13:24 | 10 |
| Hi,
The problem is that the shutdown AST is not delivered to users in the
sub-process. This problem is fixed in V3.1 where the shutdown program
will kill off processes still in ALL-IN-1 even if they are in the
sub-process.
Cheers
Stuart
|
4259.2 | How do we kick the users off now? | COPCLU::ELIN | Elin Christensen @DMO, DTN 857-2406 | Wed Jun 15 1994 14:09 | 13 |
| Thanks Stuart,
I knew that there was a problem with subprocesses in earlier version,
but I had hoped that it had been solved in v3.0.
Is there any recommended (preferably automated) way of throwing these
users off in the current version?
Or should I tell our ALL-IN-1 manager to log in at 1:50 every Friday,
Saturday and Sunday night, do $SHOW DEVICE/FILES and $STOP PROC/ID on
users who lock ALL-IN-1 data files?
Elin
|
4259.3 | | IOSG::MAURICE | Six Programmers in search of an analyst | Wed Jun 15 1994 14:23 | 10 |
| Hi Elin,
If you look at an EFT kit of the next release you will see in there
command procedures called OA$LIB:OA$SM_USER_KILL_ALL.COM &
OA$LIB:OA$SM_USER_KILL_NODE.COM. What the rules are about you giving
these to a customer, I don't know.
HTH
Stuart
|
4259.4 | thanks | COPCLU::ELIN | Elin Christensen @DMO, DTN 857-2406 | Wed Jun 15 1994 14:56 | 7 |
| Thanks,
It is our inhouse production cluster in Digital Denmark that has the
problem, so the rulse should not be a problem this time.
Elin
|
4259.5 | What this rulse all about ? :-) | COPCLU::GLARGAARD | Allan Glargaard, DS @DMO | Thu Jun 16 1994 12:33 | 8 |
| >> It is our inhouse production cluster in Digital Denmark that has the
>> problem, so the rulse should not be a problem this time.
Elin, what is a "rulse" - anything relating to p�lse?
Just kidding, of course,
Allan
|
4259.6 | What I did | SHRMSG::HOWARD | Yes it is | Thu Jun 16 1994 18:34 | 14 |
| It sounds like you have the solution. But I used the technique in .3
about a year ago in batch with great success. I just did a
$ SHOW DEV/FILES/OUT=FILE.TMP DISKNAME, then looked at the process id
for each one, checking it for "0000000" and my own process id. There
was some issue that if the process id was null:
$STOP/ID=
then the current process (the batch job) was killed. I believe that I
also checked to see if the user was still active, as most had multiple
files open and might have been killed for having an earlier file open.
As far as I know, nobody ever escaped being killed.
Ben
|