- Metrics about DB and "host" the DB running on are wrong (we use node, not pod metrics). This makes some data where we compare MySQL parameters and "host" wrong (for DBA, Developer)
- There is no way to get any understanding about capacity and usage of K8s clusters (for DBA)
- K8s clusters are added to PMM but there is no way to manage them except one special page. They are not entities in PMM, so we can't create good dashboards for them or combine K8s and DB entities. For example, build some graph of relations between k8s clusters and DB's in PMM dashboards and inventory. (for Developer, DBA)
- Allow user to drill down from Environment ->K8s -> pod -> DB (example) : K8s Overview, K8s Summary, Containers overview, container summary > MySQL/mongoDB summary )
- Execute Advisors (Checks) against K8s cluster. WE wand to be able to operate with it as we have Services/nodes we have in advisors.
In PMM terms K8s cluster is more related to Services if we speak about "What we are monitoring " and more Node if we look at it as a container for other objects.
As for now, PMM assumes Node and Services are only related objects. ad There is no inheritance for nodes (no way to have Node1-> node2->node3->servece1 )
Considering K8s can change forms it looks like more as Services as Nodes usually more 'containers' not applications
- pmm-admin add k8s -
- User able to add K8s cluster into PMM and specify label(s) (Environment) using UI and CLI
- User able to see metrics about NODES in K8s cluster (Autodiscovery? )
- User able to see Container metrics (Autodiscovery? )
- Containers and Services are presented on the same dashboards as bare-metal services and nodes so the user able to see DB metrics together with Node/container metrics
Out of scope:
- services metrics (will be added by operators )
How to test: