[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference azur::mcc

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

3631.0. "HISTORIAN PROBLEMS" by TAV02::KALAI () Wed Aug 26 1992 13:19

Shalom,
We have a very big network that we try to manage and control with DECMCC
As I mention in some other note this is the first project it is running
now on real but i still have some problems.
it's running on 4000/60
          	vms 5.4
		56mb
		DECMCC 1.2
	
I thing we have some degradation in the performance and I fell it can
be something with the historian.
for recording purpose the network is divided to three domain ,the polling rate
is 15 m for counter and status of line and circuit 2 h for node .
we keep information for 14 days.the repository file are now between 
250000 to 310000 blocs.I started them at 29 of July and we are doing
the compress tonight .
Since a couple day I notice that even that I have no error message from 
the background process the polling is not happening that it suppose to run.
when I do SHO RECORD NODE4..... I find tow thinks :
	- the last successful pool was a few hour or a day before no
	  fails pool since.
	-or from time to time we get fails pool with SCHEDULE TIME PASSED.

Any idea what I can do to resolve this problems?
We need the history our customer like doing investigation on the past,and
he is losing confidence on our capability to give him The right tools.

thanks
Kalai Yael
T.RTitleUserPersonal
Name
DateLines
3631.1TOOK::SHMUYLOVICHWed Aug 26 1992 16:4064
	Let's numerate your problems:
	   
	  - #1 - "the last successful poll was a few hour or a day before no
	          fails poll since";
          - #2 - "from time to time we get fails pool with SCHEDULE TIME PASSED".

	Problem #2 happens when the system is busy and there is 
	no time to start the  polling operation in time (the "in time" window is
	about 5 seconds). It can be if at the time of the current poll your
	system is doing other MCC actions, for example, a lot of alarms or 
	exportings. Or it can be because of other activities in your system.
	Or it can be because the background restart procedure forces many
	recordings to start simultaneously ( see details about restart procedure
	at the and of this reply).	


	The problem #2 is more difficult to understand and I need your
	answers on the following questions:

	1. How many recordings do you have for each domain?

	2. Does situation #1 happen in all domains or only 
	   in one of them? How many recordings are in this domain(s)?

	3. Do you have at least one successful pool (close to the current time)
	   for any other recording in the domain with problem #1?

	4. Did you restart the background process for the domain(s) with
	   problem #1? If you did, did you specify a second argument to
	   the background?

	5. What other MCC activities do you have? Alarms,exportings (how many?).

	6. Is it possible to send me (took::shmuylovich) the result of 
	   show recording for the domain with the problem #1? Please, send me 
	   the result for as many recorded entities as you can. It may give 
	   an idea what is going on.

	Before I receive your answers I can see the only explanation for 
	the problem #1. This is the background procedure restart without 
	second argument and a big number of active recordings.

	The second argument to the background procedure is a delay (in seconds)
	between restarting of a recordings. If for example, the delay is 10 sec
	the first recording restarts (makes a poll) immediately , the second
	recording in 10 seconds and so on. If the delay argument is not 
	specified the default ( 30 seconds) is used. Please note that if you
	have 100 recordings and you are using a default delay the last recording
	starts in about 50 minutes after the background procedure starts! If you
	do Show Recording during this 50 min it shows "last successful poll"
	from previous session of the background. 
	Is it possible that this is your situation? 
	If yes and a default delay is not acceptable for you, try to specify 
	smaller delay.
	
	Why in V1.2 we introduced the delay argument? This is done to prevent
	the system overload when all recordings start immediately (of course,
	you can specify delay = 0).

	Hope this helps,

	Sam Shmuylovich