Hazelcast in A12 Camunda Service

In A12 Dataservices Hazelcast is active by default.

As A12 Camunda Service now is / or has some aspects of a dataservice: is it recommend to activate Hazelcast here (esp. in a setup with multiple service instances)? It does not seem to be active by default. What would I have to do?

Thank you.

Hi @anon67142020,

thank you for your post! We’ve forwarded your request to our workflows team and they are looking into it in A12WF-1041.
As they are currently very busy, there may be some delays in processing. We apologise for this and hope to be able to provide an answer soon.

Your A12 discourse team

I’d like to extend my original question: Is Hazelcast active by default in a12-camunda? It does not seem so.

Background: we usually have multiple instances per service. If Hazelcast was active, we’d have to reconfigure it for a multi-instance setup. For starters, I’d be fine with a setup without Hazelcast / caching, but I am not sure, what the default is (and what is recommended).

@peter-smooth-cursor maybe you could give some input? Thanks!

Hi @anon67142020,

I’m really sorry for the late reply. We were finally able to investigate this together with the A12 Data Services team.

We found that there is a somewhat confusing “error” message when starting up Camunda:

2023-03-14 14:21:01.917 DEBUG 18812 — [main] .i.c.AbstractYamlConfigRootTagRecognizer : An exception is encountered while processing the provided YAML configuration

com.hazelcast.internal.yaml.YamlException: An error occurred while loading and parsing the YAML stream

Is this the reason why you think it doesn’t seem like Hazelcast is started? We found that this is just a DEBUG log message that actually also occurs in Data Services itself on startup and doesn’t seem to prevent the Hazelcast startup.

The following logs indicate that Hazelcast is running:

2023-03-14 14:22:11.743 DEBUG 18812 — [hz.A12S.cached.thread-4] c.h.i.p.InternalPartitionService : [<INTERNAL_IP>]:5701 [dev] [5.2.1] Checking partition state, stamp: -8661523421455686299

There, you can see the default cache name “A12S” coming from Data Services.


Regarding the multi-instance setup: yes, it is recommended to use Hazelcast. The configuration of the cluster is out of the scope of A12WF but the A12S team said that the Hazelcast docs are supposed to be very good, see Network Configurations | Hazelcast Documentation.

If you already did this successfully for Data Services, you should be able to do it the same way for Camunda. Possibly, the A12 Build & Deployment team can also support in this.

Hope that this helps!

Hi @peter-smooth-cursor, thank you for the clarification.
It turns out that for some reason - related to logging configuration - we simply did not see the Hazelcast Logs in a12-camunda. Now that we do, we can go about setting up multi-instance support.
Best regards, Alex

Hi @anon67142020,

Okay great, glad I could help!