Always many queues baseline_engine and preditcion_engine

[Miller Echagarreta]

Hi everyone.

 

I have been see that my CA UIM always have many queues from probes baseline_engine and prediction_engine.

 

 

And this case, my capture show you how the .sds occupied all the disk.

 

 

I am trying resolved this (deactivate and activate) the probe baseline_engine and empty the queues from IM.

Too I applied this procedure but (https://support.ca.com/us/knowledge-base-articles.TEC1353703.html) the problem is the same.

My UIM-Server have 20gb of memory and I have been increment the memory max and init to 3 gb for this probe baseline_engine and prediction_engine and this not resolved the problem, rather needs more memory. This is a problem, becaused I can't increment more memory.

 

Do you have any suggerence or exist a best practice for this type situation?

 

My system is:

 

CAUIM 8.51

baseline_engine 2.76

prediction_engine 1.34

 

Thank you for your help and comments.

[Britta Hoffner]

Is this a baseline engine on the primary or secondary hub ? Please have a look at Secondary (remote) hub's baseline_engine QOS MESSAGE queue turned yellow and is backing up 

 

Kind regards,

Britta Hoffner

CA Support

[Miller Echagarreta]

Britta.

I have with my environment 1 year functioning, and this problem started from a 3 months ago. I have many problems that I have been resolved, but this "baseline and prediction" is a recurring big problem.

 

I have 1 Primary Hub and 1 Proxy Hub. This is my environment.

 

                                                                  Primary Hub (queues-typeGet)

                                                                                       |

                                   Proxy Hub: queues-typeGet (cutomerN), Tunnel-Server. queues-typeAttach

                      |                                                               |                                                            |

      Hub-Customer01:                                  Hub-Customer02:                                        Hub-Customer03:

(queue-typeAttach, Tunnel-Client)          (queue-typeAttach, Tunnel-Client)            (queue-typeAttach, Tunnel-Client)

 

Right now the system working very fine. But I don't know when problem starting again

 

M.E

[Yu Ishitani]

Hello.

Please let us know which monitoring probes are in use on the Primary HUB Robot.

 

You may be able to deactivate baseline_engine / prediction_engine in the Primary HUB Robot, depending on which monitoring probes are in use on the Primary HUB Robot.

 

Regards,

Yu Ishitani

[Miller Echagarreta]

Hi Yu.

 

May be is a solution (deactivate baseline_engine / prediction_engine). I spossible this? If deactivate i think that this decrements consume of memory in Primary Hub.

 

UIMServer probes 

 

Continuation UIMServer probes 

 

ProxyServer probes

 

Hub-Customer

[Yu Ishitani]

Hello.

As far as I see your screenshot, I think there are no impacts even deactivating baseline_engine / prediction_engine on the Primary HUB.

baseline_engine / prediction_engine might be required on Customer-HUB at least you are using alarming in SNMPcollector.

 

Regards,

Yu Ishitani

[David LeDeaux]

Miller,

 

Please check this article; it is slightly different:

Baseline_engine.QOS_MESSAGE queue is piling up and is yellow in the hub queue. 

 

It goes much beyond setting loglevel to 0.

[Miller Echagarreta]

Hi David.

 

This solution I applied 1 week ago in Proxy-Hub. Right now I apply in Primary Hub. Although I did not seem to see a difference in Proxy-Hub

 

Will this be necessary in all client hubs?

 

M.E

[David LeDeaux]

It will be necessary to apply to all hubs with baseline_engine.

 

Bear in mind that part of the problem is that the version of baseline_engine that ships with the current release does not actually support to log level 0.  The CASE statement that controls logging had no entry for loglevel 0 so it defaulted to loglevel 1.  That's why the article you cited had no effect.  We should retire that article.

 

Additionally, there are a few logging facilities that are not affected by loglevel; they have their own configuration directives.

 

Part of the problem is that baseline_engine can not log to disk fast enough when the queue gets too large.  Therefore, the messages build up in memory and since they cannot be garbage collected, they reside there indefinitely.

 

The article that I linked describes how to send all logging to console instead of writing to disk.  This effectively turns off logging so it doesn't consume memory to log.

 

https://support.ca.com/us/knowledge-base-articles.TEC1000777… 

[Miller Echagarreta]

Okay David.

 

If I am understand you fine, i should do this:

1- Change in the log4j2.xml file all registrys <AppenderRef ref="console"/>

2- Change loglevel in baseline_engine probe to = 0

3- Change the line from log4j2.xml file, in all hubs with baseline_engine.

 

This is correct?

 

M.E