| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 1517.1 | More info... | WELCLU::CRIDDLE | Graham Criddle, DS Tech Consultant, 853-4015 | Mon Sep 23 1991 12:56 | 36 |
| Hi,
Having originally raised this with Bipin, I have been able to find a
way to reproduce this 50% (or so) of the time.
This is a VAX 9000 running VMS 5.4-1A
I have an alarm (severity level critical) which always satisifies it's
test criteria. (It is testing a nodes inactivity timer = 60 which is
true)
I have a command procedure which does the following:
$ loop:
$ man/ent
enab mcc 0 alarm rule critical, in domain uk.hinckley.domain
show mcc 0 alarms rule critical, in domain uk.hinckley.domain, -
at start = (+00:00:04)
exit
$ goto loop
If I start this procedure running, and then start the IMPM from a
Fileview window, 50% of the time notifications is disable on the domain
uk.hinckley.domain with the window popping up as Bipin described in .0
The domain uk.hinckley.domain is a dynamic subdomain of the domain I open
when starting MCC.
I assume that there is some timing window at IMPM startup during which
it has enabled notification but is not ready to receive them, or some
such thing??
Any thoughts/comments?
Rgds,
Graham
|
| 1517.2 | known problem | TOOK::HAO | | Mon Sep 23 1991 13:24 | 12 |
| Graham,
You are correct. The V1.1 IMPM had problems when alarms are enabled
before the IMPM is fully up. This is because the notification and map
window code access some of the same structures, and those structures
may not be fully set up by the map code before notification tries to
use them.
This has been fixed for V1.2.
Christine
|
| 1517.3 | Thanks for the update. | WELCLU::CRIDDLE | Graham Criddle, DS Tech Consultant, 853-4015 | Tue Sep 24 1991 06:11 | 18 |
| Hi, Christine...
Many thanks for the reply...
Is it possible that the same problem occurs some time after the map has
been setup and in use for some time?
Why I ask is that I am getting exactly the same error condition
occuring at random times (seemingly) during normal operation. However,
it was possible to provide a reproducible scenario for the startup
problem but not for the case where it occurs in 'normal' operation.
If it is likely that I am seeing the same symptoms for a different
cause, can you think of anyway in which I can gain sufficient
information to provide analysis of this aspect of the problem?
Rgds,
Graham
|
| 1517.4 | rare but possible | TOOK::HAO | | Tue Sep 24 1991 15:14 | 15 |
| Another situation where the problem could come up would be when you're
double-clicking into something that takes a long time, and notification
was received. It really depends on the timing. Most of the time, the
above will work without problems; however, if the notification was
received at the time the map code is still building the data structures
for the new view, the same problems will appear.
So I guess the answer is yes, it may happen even after the IMPM has
been fully up. Since the real problem is that we do not have locks on
these data structures, any situation you can get into where map code
is modifying data structures that notification code happens to want to
look at, will be a problem.
Christine
|
| 1517.5 | A note about your script... | TOOK::ORENSTEIN | | Tue Sep 24 1991 15:25 | 7 |
|
I suggest that you add a DISABLE RULE * to your command procedure.
The EXIT will abort the rules running and this may leave the event
manager in a funny state. The DISABLE will properly clean things
up.
aud...
|
| 1517.6 | Many thanks... | WELCLU::CRIDDLE | Graham Criddle, DS Tech Consultant, 853-4015 | Wed Sep 25 1991 09:57 | 15 |
| re .4
Thanks, Christine.
The scenario you describe seems to tie in with what I have seen. I
guess we wait for 1.2
re .5
Hi, Aud.
Thanks for the point about DISABLE. In the procedure I use for
prodcution work I do have the appropriate DISABLE commands, just did
not include it in this test procedure.
Thanks to all...
Rgds,
Graham
|