Oracle RAC Backend handling

[Lutz Mader]

Hi all folks,

we use Oracle RAC with two member for some application, if I have a look to the Backends metrics tree, I find two entries in the tree. But only one contain some usefull metric data, the other one contain "gray" entries only (see below).

 

Does someone know, how to merge this two tree branches into one branch.

Is this the normal behaviour to add matric data to one of the RAC member branches only.

 

A question of understanding only,

Lutz

 

p.s.

We use APM 10.5.1 Agents and EMs in a IBM Java 1.7.1 and Websphere 8.5.5 environment.

 

[Haruhiko Davis]

Guenter_Grossberger will your NF work in this case?

[Lutz Mader]

No idea or suggestion,

I open a ticket to get some detail information to find out what's going wrong.

 

Collecting metrics for Oracle RAC DBs works well in the past, based on Oracle 11 and Introscope 9.5.2 or 9.5.5 Agents. Unfortunately it doesn't work well based on Oracle 12 and APM 10.5.1.

 

With regards, Lutz

[Hallett German]

Hi Lutz:

     I did let those in the internal Communities and APM Support know about your request. Since it is a specialized topic, only a limited number of people could possibly respond 

 

Thanks

Hal German

[Sergio Morales Correa]

Hello Lutz

This is how it works:

 

1. We track every SQL that is triggered and determine performance metrics.
2. Then we determine the metric path. Agent Nameformatters help us determine the metric path which contain “hostname + PortNum + (Oracle DB)”. In RAC case, we get the session properties from the connection.
3. If Oracle maintains wrong info in those properties, we show the wrong hostname.

 

Our code deals with different configurations Oracle is allowing, but from version to version, Oracle modifies their logic and accordingly we need to modify our code.
If this configuration worked before but now now, it is definitely better to open a case, a few debug jars should be able to help Sustaining team identify the problem and fix the issue if any.

 

Regards,

Sergio

[gamsa03]

Hi Lutz,

From the screenshot it looks like two different nodes of the cluster. ---d100stul0249--- and ----d100stul0749--- . In such case the metrics tree looks fine, why to merge these two.

This indicates that the current requests are all served by ---d100stul0249---- node of the cluster as the other node is greyed out.

 

--

Thanks,

Santosh Gampa