High system resources consumption detected for datasources

[AntonioSousa]

I have an environment with ActiveGates that are used mainly for extensions v2, with SNMP & traps. Lots of them...

I'm trying to optimize the number of AGs, but as I reduce them, I bump into these SFM logs:

• High system resources consumption detected for datasources (hard limit). Rejecting configuration ef008f30-b5c2-33d1-ad87-9020f3c892a8
• Failed to assign monitoring configuration to ActiveGate. Reason: High system resources consumption detected for datasources (soft limit)

Now, monitoring the dsfm metrics, I don't see high CPU usage, nor memory. JVM heap memory and CPU doesn't seem very different before/after changes...

How would I discover which resource is most being needed, so I can eventually give it more? CPU? Memory?

[AntonioSousa]

I forgot to say that the ActiveGates are already in "High limits" for their performance profile.

In the meantime, I also discovered this page,
https://www.dynatrace.com/support/help/extend-dynatrace/extensions20/dedicated-performance-profile

that has a neat custom.properties example for disabling things that might not be needed.

I also can't go for "Dedicated limits" because I also need extensions v1...

[jegron]

Hi @AntonioSousa !

Interesting. Did you check AG limit ?
https://www.dynatrace.com/support/help/extend-dynatrace/extensions20/extensions-concepts#limits

[AntonioSousa]

@jegron,

Thanks for pointing this out. I had already seen this when I had changed to "High limits", but had not associated it this time.

So the soft/hard limits should be the ones you point to, but now I'm trying to figure out what of the two, CPU or/and memory, are affecting me...

[AntonioSousa]

I have also managed to reduce ActiveGate memory usage by inserting the following two lines in launcheruserconfig.conf

-java.xmx.absolute_part=2000
-java.xmx.relative_part=0

I got the idea from https://www.dynatrace.com/support/help/extend-dynatrace/extensions20/dedicated-performance-profile

but I really had to put the two lines, because in my case Xmx would be around ~3500M if I only put the first one. Now it's -Xmx2000M 

[Yosi_Neuman]

HI @AntonioSousa 

Did you get answer for this one from support?

We have see the same error message from Elasticsearch extension (1.0.7) running on OA   

Thanks for any input on what can be done ! 

Yos 

[AntonioSousa]

@Yosi_Neuman,

I have not been working with OA EF2 extensions.

Regarding this AG context, I did not go through Support, as these type of tweaks are not easy to discuss.
I went for the "dedicated" performance profile, and tweaked the ActiveGate memory, as referenced above. Please note that when I refer to ActiveGate here, is to the main AG module, which typically does the routing. As mine AGs are only for extensions, I reduced the gateway memory, as it is not needed, thus giving more memory for EF2 extensions.

[arild_palencia3]

@AntonioSousa this adjustment is essential for the stability of monitoring of 3rd parties with the Extension AGs. Thank you for this tip, it really improved our monitoring of MQ and DB2