Process Mapping Logo

Process Mapping - Forums

Sharing 19 years of knowledge and experience

 
Metastorm BPM forums
Sign up Latest Topics
 
 
 


Reply
  Author   Comment  
paulsiu

Senior Member
Registered:
Posts: 57
Reply with quote  #1 

Recently, our dev box has started to 100% cpu. We are using Metastorm 7.6 SR3. The process sucking up all of the cpu time is dllhost.exe.

We have talked to our DBA and he said there is no unusual load on the db, no locking, and no fts.

How can we go about tracing down this problem?

Paul


0
CraigO

New Member
Registered:
Posts: 6
Reply with quote  #2 

This is still an ongoing issue.  More info: If the event manager is disabled in the registry, CPU usage drops to near nothing.  So it seems to be in the event manager, not the alert generator.  If I start another engine on another server connected to the same metastorm db, the high CPU usage also starts on that server as well.  Nothing ever in the eraisedflag table (or the earlertrequesttable).  The eevent table is also not adding events during the CPU usage (just a few legitimate events when needed).  So what is that Event Manager working on?  It just runs at 80% to 100% CPU every day with no break.  Our other environments (test, beta, production, training) are fine...   Running out of ideas here and hoping for some help from the community.

0
Jerome

Avatar / Picture

Guru
Registered:
Posts: 5,507
Reply with quote  #3 
Are you sure there is no process on that box running amok? It may be cloning recursively, looping endlessly, or just trying to raise a flag for thousands of folders (this last can be very hard to diagnose).

__________________
Post an example, and we will have a much better idea what the problem is. In about 90% of posts, the problem is one of communication. Examples bridge that gap.
0
LJ

Avatar / Picture

Senior Veteran
Registered:
Posts: 452
Reply with quote  #4 

We get this on our Live box. It has been up to 79% but is currently running at 20-29%. Is this likely to be something running on the engine (such as a poller) or an external process (xml data pass or external database reading)?


__________________
LJ
0
Jerome

Avatar / Picture

Guru
Registered:
Posts: 5,507
Reply with quote  #5 
It is usually what I suggested in the previous post in our experience.

__________________
Post an example, and we will have a much better idea what the problem is. In about 90% of posts, the problem is one of communication. Examples bridge that gap.
0
jerry1970

Member
Registered:
Posts: 13
Reply with quote  #6 
We have the same problem. We've already found out that the Metastorm tables are far from optimized. Indexing is pretty horribly, namely non-existent. That's pretty bad, to be honest. We've been adding indexes to a lot of tables.

We also got a better performance by removing the Ticker from e-ServiceDesk. That one called a stored procedure called DashboardBanner, which was run 2 billion times (!) a day.

Removing the Ticker was not an easy task. We had to add the WebPartsAdministrator role to a user by hand (it does not exist as a role itself, but the scripts do check for it). Then you can remove web parts from the Shared interface instead of Private interface. Then you also need to remove a line from an ASPX page to stop people from adding a ticker to their private interface.

Next up is optimisation of the stored procedure that uses @MonthDates. This uses not very efficient query's.


0
raminder

Member
Registered:
Posts: 36
Reply with quote  #7 
Hi
We are facing the same problem. Disabling event manager brings CPU usage to normal. But still cannot figure out what is the root cause of the issue and how to fix the event management thing.

__________________
shady
0
Previous Topic | Next Topic
Print
Reply

Quick Navigation:


Create your own forum with Website Toolbox!