@SDunlap has been chatting with @colewhite in #wikimedia-observability about Catalyst sending its logs to Observability's Beta-Logs ELK stack. The Observability team has expressed openness to this idea, with the following considerations:

1. We can only consume from a Kafka instance hosted in your project
   1. TLS is optional
   2. A firewall rule might be needed and maintained - if we change Logstash hosts, the IP will change breaking the connection
   3. We'll need to know the Kafka topic(s) you intend to produce to
2. We cannot provide custom log normalization in this env - production and beta filters are shared and should not diverge
3. We cannot provide any availability guarantees - the environment is for our testing and secondarily benefits users of deployment-prep
   1. Service outages are handled on a best-effort basis
   2. Logs may be deleted at any time
4. Logs on this instance are considered "public" in the sense that they are accessible by anyone with deployment-prep access
5. We try to keep logs for two weeks. This could be extended if need be and depending on cluster space available.

This task is to:

• with PMs input, decide whether each consideration listed above is acceptable to our use-case
• identify and catalogue steps needed to begin sending meaningful logs to Beta-Log