Jaeger: Determine initial configuration for jaeger-operator
This issue is part of &210 (closed).
We previously developed a proof-of-concept (https://gitlab.com/gitlab-com/gl-infra/infrastructure/-/issues/9829) for the jaeger operator connected to GDK.
By default, the operator uses an all-in-one setup where everything runs in a single deployment. For scalability reasons, we'll want to separate out the different jaeger components.
The next step is to review the jaeger operator's documentation and decide on the configuration and topology we want to use.
This includes:
- Where to deploy the agent (injected sidecar, manual sidecar, or DaemonSet)
- Which Elasticsearch to use for persistence (logging cluster, new elastic-managed, self-managed on k8s)
- How to interface with Elasticsearch (do we use the supplied cron jobs or integrate properly with ILM?)
The goal of this issue is to track the work and document the initial decisions. We can (and probably will) iterate on this config later. But let's aim to have something more production-ready as a starting point.