| Title: | DECmcc user notes file. Does not replace IPMT. |
| Notice: | Use IPMT for problems. Newsletter location in note 6187 |
| Moderator: | TAEC::BEROUD |
| Created: | Mon Aug 21 1989 |
| Last Modified: | Wed Jun 04 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 6497 |
| Total number of notes: | 27359 |
Hi all and happy new year,
I have a problem at a customer site overthere which concerns the
collection AM. It seems to be related to a performance problem or
ability of DECMCC to handle event burst.
Here is the problem description:
Dropped collector events in DEC MCC under heavy load
=====================================================
Products: DEC MCC 1.4, DECnet/OSI 6.2 VMS 6.1,
Facility: Collection AM
Problem Description:
We have an environment where up to 50 collector events from remote
nodes are regularly sent *AT THE SAME TIME* from around 20 remote nodes to
an MCC Station. The events are sent by system managemet tool ROBOMON using
the EVC API, transport is DECnet / NSP.
We observe that in this setup about 10% of the events are dropped quietly
by DEC MCC. These dropped events are logged by ROBOMON with
'connect to network object rejected'. When firing single collector events
to MCC these are recognized correctly. Until now it seems totally random
which and how much of the events get dropped
Monitoring DECnet (NSP local nsap,remote nsap), no conn rejects could
be observed.
Doing a CFT trace "NSP port *" however didn't show all (even successful)
events.
My questions are:
1. Is there any limitation on the number of simultaneous events that
mcc can handle ?
2. If there is no limitation or the limit is higher than 50, can you
made any suggestion to adjust some mcc parameters like some stack sizes
or the event pool size or ... assuming that the operating system's
quotas are correct?
Thanks in advance for your help.
Best regards.
Paul
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 6417.1 | That should work ! | DSVB03::TABONE | Mon Jan 15 1996 16:39 | 18 | |
Hello Paul, > We observe that in this setup about 10% of the events are dropped quietly > by DEC MCC. MCC should display an error message. Do you see it ? You can try to increase these parameters : MCC_EVENT_POOL_SIZE 524288 (default value) to 1048576 MCC_EVENT_EDQ_SIZE_LIMIT 50 (default value) to 200 If the problem continues, thank you to logg an IPMT case. Hope this help you. Regards, Philippe Tabone - DECmcc Maintenance Team - | |||||
| 6417.2 | any typing mistake ? | AXPMCS::MOINET | Tue Jan 23 1996 11:03 | 23 | |
Hello Philippe,
Sorry for the late answer but I was not expecting any reply after a few
days.
I already played around this two logicals without any success, but I
will try to apply the values you mentionned. I just have a doubt
because you told me to "increase" two logicals values and the first one
is lower than the default. Is it correct or did you do any typing
mistake?
Could you confirm this?
>You can try to increase these parameters :
>
> MCC_EVENT_POOL_SIZE !!524288!! (default value) to 1048576
> MCC_EVENT_EDQ_SIZE_LIMIT 50 (default value) to 200
>
>If the problem continues, thank you to logg an IPMT case.
I can also answer to your first question: MCC does display any error
messages.
Best regards.
Paul
| |||||
| 6417.3 | No, reading mistake ! | DSVB03::TABONE | Wed Jan 24 1996 09:48 | 9 | |
Hello Paul, > MCC_EVENT_POOL_SIZE !!524288!! (default value) to 1048576 1048576 = 2 * 524288 Bonne journee, Philippe | |||||