Hi, I developed my microservices based solution and I am using zeebe to orchestrate my services. I am wondering if someone has some advice concerning the production environment. I am actually using docker-compose but it is a test/dev environment. Thank you
1 Like
Hi @nizare, many people use self-managed K8s with some variation of https://github.com/zeebe-io/zeebe-helm.
You can also use Camunda Cloud - a hosted, managed Zeebe SaaS that is in beta. I’ve done both, and Camunda Cloud is improving all the time, and has become useable for me for my use-case. I recently switched off a self-hosted system and moved it to a Camunda Cloud-hosted cluster.
2 Likes
Hi @jwulf, thank you for your answer. In my case, I cannot use Camunda Cloud because the customer wants to use his data center. I guess that the best solution will be to install a bare metal kubernetes cluster and deploy a zeebe cluster with persistent volumes. What do you think ? Thank you
I think so. I ran a docker compose of Zeebe + Operate + ES on a 4GB VM on AWS, but the VM itself would die every month or two, after a CPU/memory spike.
Having multiple brokers in a K8s cluster means that one broker stopping won’t halt the system, and K8s will restart nodes that fail for whatever reason.
These Zeebe Nation episodes might be of interest:
2 Likes
I have already faced this problem when I tested the clustered version using Azure K8S. I will keep an eye on the issue because I guess that this is related to Grpc. I am having issues with Grpc even when I use the docker-compose version.
1 Like
What issues do you have with the docker-compose version?