search

airshipit / treasuremap

Reference Airship manifests, CICD, and reference architecture.
http://openstack.org
Apache License 2.0
52 stars 39 forks source link

Create Phase for LMA Subcluster Workload #71

Open sb1975 opened 3 years ago

sb1975 commented 3 years ago

As a developer I would like to create an airshipctl phase called LMA-Subcluster Workload Phase. This phase will use the LMA-Stack Composite. The LMA Stack composite uses the functions (ES, Kibana, ES-Exporter & Thanos-operator, Grafana(enable the grafana in the KPS, and Minio-S3 function) earlier covered in #61

Acceptance Criteria :

  1. create a LMA-Stack Composite that can reused for Consolidating the LMA data from other clusters in a LMA Subcluster
  2. use the Elasticsearch function for Elasticearch-Ingest nodes & Elasticearch-Data nodes.
  3. This phase will include the LMA CRs for the LMA Subcluster components which includes - Flow,Output,Podmonitor,ServiceMonitor,PrometheusRules,AlertmanagerConfig,etc)

Note: In case its not feasible to use the same function for Elasticearch-Ingest nodes & Elasticearch-Data nodes, then it may need two additional composites - Elasticearch-Ingest Composite & Elasticearch-Data composite. (#57 and #58 )

Dependency List :

61 is closed because the scope is addressed as part of this Issue.

70 this phase runs on top of LMA-Base Composite Phase

StevTheDev commented 3 years ago

This phase entrypoint is created in This PS for #67, but requires the composites from #83 and #69 (PS here) all to be merged before this issue can be marked complete.

StevTheDev commented 3 years ago

The 'lma-stack' phase is now present in treasuremap v2's test-site/target cluster, and deploys the monitoring stack composite (#67).

These PS add the other composites which will complete this phase:

lb4368 commented 3 years ago

Notes from email discussion for LMA manifest integration for sub-clusters

  1. Disassociate airship-core from multi-tenant (#116). Create composites (like the existing “infra” composite) that represent reusable bundles of functions, as appropriate, and the independent types will consume them.
  2. Integrate the LMA composites into the airship-core workload (for operators) and workload-config (for CRs) phases, and get rid of the site-level lma-* phases a. Steven/Graham, is there any reason we can’t collapse the three LMA phases into these two^?
  3. Integrate the LMA composites into the multi-tenant’s LMA subcluster’s workload and workload-config functions a. Steven/Graham, as we move components into the LMA subcluster, do any pieces need to remain in the undercloud / target cluster?