Scale OpenNMS service

[vincenzodnp]

Hello folks and thanks for your work.

I have a doubt. I have a Docker Swarm cluster by 4 nodes and I'd like to run a docker service replicated on 4 nodes. Each container running on each node, has its own persistent volumes to store etc, data and log.
I have a single PostgreSQL endpoint.

My questions are:

• first, is it possible to run OpenNMS scaled in this way?
• are there any services like vacuumd, eventd, pollerd, alarmd, etc that cannot be parallelized?
• which issues could be face by using this infrastructure?

I'm using docker image: opennms/horizon-core-web:22.0.4-1

Thanks a lot.
Vincenzo

[indigo423]

Hello @vincenzodnp Currently you can't scale out all components of OpenNMS. Here is a list of parts which you can scale out:

• Cassandra/ScyllaDB when you use a Newts backend
• ElasticSearch when you use it for Events/Alarms/Flows
• Telemetryd when you process flows and streaming telemetry running in Sentinel
• Minions as failover and load balancing

Vacuumd is deprecated in favour of our Alarm Correlation project Sextant.

Everything else can't be easily scaled out right now but is a goal of the ongoing project Sentinel.

[vincenzodnp]

Thanks @indigo423
I think the same considerations could be done by talking about "legacy" infrastructure without container.
Isn't it?
Thanks
Vincenzo