It's stable if you do not overload it. If you go over the boundary that it can handle with the existing infrastructure, it becomes a nightmare. Otherwise, it just runs smoothly.

I would give this solution a six out of ten for stability. It has quite a few hiccups. I usually try to avoid them the best I can, but sometimes they just happen and you have to deal with it.

This solution is definitely stable. 

The solution is stable.

Graylog is a stable solution. However, while using some microservices it may go down.

Yes, with Elasticsearch.

Graylog is quite stable, and the only issue is the index rotation.

The solution's current version that I am using is stable.

I do have a multinode deployment, with only one Graylog node. As we rely more on Graylog permanently and consume more of its collected data, I will transition to a Graylog HA installation, as and when we come to require it without outage. We are moving more to IoT, and those streams will be mandated to not have any gaps. They will be responders to events that can't have any outages. 

No issues.

I am very proud of how very stable the solution is. One time I had an entire node on my VxRail VMware collapse, so I basically restored the template, gave it the same IP address and everything was working again.

The only issue we had was during the Java patch. Graylog's search DB was not able to start up after the upgrade to Java 9, so we returned back to v.8. With that only exception, we have any issues with application or its components.

No issues.

Over six months, I had two similar issues where searches were performed on field "messages". It exhausted all the memory of the ES node causing an ES crash and a Graylog halt.

No issues.

It's a stable solution. 

It is stable.