We are running Sensu Go (non-enterprise) currently and have a Sensu cluster with around 2500 machines pointed to it, and have not had time yet to compile from source and distribute to all machines to get rid of the “exceeded node limit” situation. I’m curious what the expected behavior is when the node limit is exceeded and an enterprise key is not present - we are seeing some odd behavior that I think may be related to this, but wondering what I should expect to see if I point more than the node limit at a non-enterprise cluster in this manner, so I can determine whether that’s what we’re actually experiencing or not.
I think you are most likely seeing etcd reaching its performance limit as an event store in your configuration. If so it’s going to impact both the open source codebase and free tier binaries.
Take a look at https://github.com/sensu/sensu-perf for publicly available performance testing tools.
@molly Also check out the Recommended Hardware guide: https://docs.sensu.io/sensu-go/5.11/installation/recommended-hardware/ particularly the section regarding Backend Configuration. There’s info there on hardware recommendations per checks per seconds and how you can best calculate your requirements.
The Deploying Sensu Guide also has some information with regards to network requirements/performance especially if Sensu Go is clustered: https://docs.sensu.io/sensu-go/5.11/guides/deploying