Monitoring: Watch Citus Worker and Coordinator

[perdelt]

• sum all metrics?
• show them separately?

[perdelt]

as of v0.8.14:

set_metric_of_config()

• for loading and stream
• podname ~ (.*){configuration}-{experiment}(.*)
• if there are workers: configuration=name_worker.lower(), experiment=""
  • thus does only monitor one statefulset in that case
  • but Citus: 1 stateful set and 1 deployment
  • but TiDB: 2 stateful sets and 1 deployment
  • but PGBouncer: 0 stateful sets and 2 deployments
• uses default otherwise
• can be overwritten for special dbms like cloud dbms not managed by bexhoma

set_metric_of_config_default()

• for all components with default bexhoma naming
• podname ~ (.*){configuration}-{experiment}(.*)

metrics.py

• uses special metrics for type loading or stream
• normal metrics otherwise
• if there is a container name given: replace container name in metrics

[perdelt]

they are handled separately as of next release