Analytic engine cluster configuration

Hi all

I want to configure the analytic engine in a cluster environment in 9.6 version of webMethods with Terracotta, but the documentation is not very clear, it will be helpfull if you could give me a guide of how it must be do.

currently I have configured and environment with two Analytic engine servers, but I’m not sure if it is working correctly because when I check the terracotta console it does not show nothing :shock: so I’m not sure if an extra configuration in terracotta is needed.

Thanks in advice.

Can you export your environment configuration and attach it here? I just want to take a quick look to make sure there’s nothing out of the ordinary.

Having said this, several months ago, I also setup a 9.6 AE cluster and ran into issues. If I recall correctly, one of the issues was precisely what you mentioned: the environment seemed correct, deployed fine, but nothing Optimize-related showed up on the Terracotta Dev console.

I had to open a ticket with Software AG for my issue. One of the problems was that the file sag.opt.clusterable.caches.xml file was not properly configured out-of-the-box in 9.6. This was months ago, and if I’m not mistaken, a fix should have been released for it by now. So, one thing you can try is to apply the latest Optimize fix.

If that doesn’t work, let me know and I can share my files with you.

Percio

Now… I’m going to play devil’s advocate for a second.

Is clustering AE absolutely necessary in your case?

I ask because since clustering AE, I noticed my environment is not as stable. I’ve had Terracotta and AE crash on me multiple times. To be perfectly fair, AE is not yet playing a critical role in our infrastructure so I have not spent too much time doing root cause analysis. It could very well be an issue that is easily fixed.

Regardless, I can say that out-of-the-box, compared to a non-clustered AE, the clustered configuration seems less stable, and it is obviously, more complex. Add to this the fact that the AE guide itself states that clustering may hinder performance and I’m beginning to wonder whether an active-active AE cluster is even worth it.

Perhaps using an active-passive OS-based clustering or allowing VMWare, for example, to handle fail-over is a better way to achieve availability, stability, performance, and simplicity.

Food for thought.

Percio

Thanks for your high value comments, a clustered AE is required for our environment, at the moment I can’t apply the latest fix for optimize but I will check the sag.opt.clusterable.caches.xml

Thanks in advice.

Please check the 9-6_Configuring_BAM.pdf (available from Empower or TechCommunity or locally installed).

It has a chapter describing how to cluster several Analytic Engines.

What I am missing there is to which Analytic Engine the Optimize-Package in IS should be pointed and which one should be assigned in MWS for that IS (under My webMethods -> Servers) for Monitoring.

We are currently preferring the approach to seggregate our application to smaller chunks and spread them over several IS instances, which in turn have their dedicated Analytic Engines.
This reduces the load for the AE and avoids to cluster the AE.
One of the AE additionally handles O4I-Data.

Regards,
Holger

I believe the proper configuration is to add a load balancer in front the Analytic Engines’ 12503 port. In this configuration, you’d point the IS’s and MWS’s to the load balancer instead of a specific AE. That’s how our Production environment is currently configured.

Percio

Hi Percio,

If I understand, you are suggesting to remove the Terracotta from the architecture, and in the environment configuration point to the load balancer instead of an AE?

Any way, I’m attaching my environment configuration and the sag.opt.clusterable.caches.xml file.

Regards.
Filiberto

ExportedEnvironments.xml (153 KB)
sag.opt.clusterable.caches.xml (7.25 KB)

No, I was simply replying to Holger on the question of what Analytic Engine to point to from the IS and MWS for HTTP-based communications. Terracotta is still needed, with or without the load balancer.

Let me know if you need me to clarify further or if this makes sense.

I’ll take a look at your file and I’ll get back with you.

Percio

1 Like

It make sense for me, I’ll be waiting for your response.

Many thanks
Filiberto.

Filiberto,

Indeed, it looks like your config file is missing some things, one of them being the terracottaConfig line. I’m attaching the file that I ended up with after some exchanges with support. If you scroll all the way to the bottom, you should see the line I’m referring to.

Let me know if adding that line helps. Of course, substitute your own server names. If it doesn’t help, try making some of the other changes as well, but I recommend making them one at a time. Also, be sure to keep a back up of the original file just in case.

Percio
sag.opt.clusterable.caches.xml (9.47 KB)

Many thanks Percio, I appreciate too much your help. I 'll test the configuration and back to post the results.

Filiberto
Regards