[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

3127.0. "A big Working Set for MCC" by BLG03::ROSSI () Wed Jun 03 1992 12:44

Hi there,

	the context is:

	microVAX3900   
	32MByte
	2 x 150000 blks pagefiles on different disk drives
	LAVC Boot
	Client MOTIF    
	VMS 5.5    
	MCC T1.2.7
	MCC_AUDIT.COM OK
	
	VAXSTATION 3100/48   
	LAVC Satellite  
	Server MOTIF  (MCC runs on 3900, only display here)
	
Our problems are:

	1 - Using MCC (Iconic Map) for network monitoring and
	    defining notification requests, the working set of
	    MCC process increases countinously. Defining a lot
	    of notification requests (300) and no rules defined,
	    the process working set goes to WSMAX (38000 pages) 
	    and there is a crash of the process. Increasing WSMAX
	    the process increase to the new working set limit
	    value. VIRTUALPAGECNT is enought bigger. The process
	    goes to 130000 virtual pages. Never the Working set 
	    is decreased. We tried to use PFRATL <> 0 just to
	    avoid crashes.
	Qestions:
		  Is this a bug ?
		  What about this and release 1.2 ?

	2 - About retargeting: is it possible to retarget on 
	    entities of other domains ?  If yes, How can we do it ?

	3 - How can we save filter informations ? 

Thanks in advance

Bye

Giuseppe Rossi
T.RTitleUserPersonal
Name		DateLines
3127.1SSB kit will be betterTOOK::MINTZErik Mintz, dtn 226-5033Wed Jun 03 1992 16:436
On question 1:

  We have continued to plug memory leaks in the areas where you are seeing
problems, and the alarms/notification area is much more robust in the
current baselevel than it the EFTU kit.  Expect the SSB kit to be better.
3127.2fixes/enhancements to notification/alarmsTOOK::CALLANDERMCC = My Constant CompanionMon Jun 08 1992 16:0623
    More info...
    
    Alarms and Notification just got CLEAN BILLS OF HEALTH!!! I have
    been running 400 rules for three days with **NO** memory leaks in
    either the notificaiton or alarms FM. This doesn't mean there aren't
    leaks in DECwindows or the iconic map, but we are make significant
    leaps forward. For example the 400 rules only user 75520 pages
    when run from the FCL with all the rules in this same process.
    
    As to the rest of your problem...in the V1.1 and T1.2 kits the
    notification FM made 1 request per domain being monitored for events.
    In V1.1 it also made 1 request per entity in the entity list; and in
    T1.2 it made 1 request per entity class. With the SSB kit we have
    removed the overhead of calling alarms, and instead have replaced
    it with a direct request to the internal event manager. The effect
    of this is a speed up in performance, a drastic reduction (in larger
    environments) in resource consumption (or a minor reduction if only 
    using 1 domain).
    
    Hope this helps.
    
    jill