-
Task
-
Resolution: Done
-
Normal
-
None
-
None
-
None
-
None
-
5
-
False
-
None
-
False
-
Yes
-
MGDSRVS-346 - Autoscaling of OpenShift Streams Data Plane
-
---
-
---
-
5
-
MK - Sprint 221
WHAT
Communicate the max number of nodes for each supported instance size to the fleetshard via the ManagedKafkaAgent spec. The Fleetshard will send back the maxUnits and remainingUnits per instance type in the ManagedKafkaAgent status, we’ll have to store maxUnits per instance type in the database as the information can be used in the placement strategy to determine if there is capacity for a given instance type
WHY
The Fleetshard needs this information to compute the maxUnits and remainingUnits which are need by the control plane for placement strategy
HOW
- The control plane will need to send the "maxNodes" information via the ManagedKafkaAgent spec part as:
capacity: standard: maxNodes: <max-nodes-for-standard> developer: maxNodes: <max-nodes-for-developer>
The maxNodes will be in the database and will be done as part of https://issues.redhat.com/browse/MGDSTRM-8740
- Store the maxUnits and remainingUnits for each supported instance type. The value will be sent by the Fleetshard as part of ManagedKafkaAgentCR status.
capacity: standard: remainingUnits: <value> maxUnits: <value> developer: remainingUnits: <value> maxUnits: <value>
DONE
Include the following where applicable:
- maxNodes per instance type sent to the fleetshard
- maxUnits and remainingUnits per instance type store in the database when the status report is sent by the fleetshard
- blocks
-
MGDSTRM-8742 Modify the kafka cluster placement strategy to take the maxUnits into account
- Closed
- is blocked by
-
MGDSTRM-8740 Expose a configuration to read the max number of nodes for autoscaling configuration per instance type
- Closed
-
MGDSTRM-8582 Update the managedkafkaagent spec and status for capacity reporting
- Closed