Skip to content

Commit e2d1422

Browse files
committed
update to SIG Cloud Provider input.
1 parent 04ff1ce commit e2d1422

File tree

1 file changed

+8
-3
lines changed
  • keps/sig-cloud-provider/2436-controller-manager-leader-migration

1 file changed

+8
-3
lines changed

keps/sig-cloud-provider/2436-controller-manager-leader-migration/README.md

Lines changed: 8 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -397,11 +397,13 @@ N/A. This feature is never used by any user workloads.
397397
###### How can someone using this feature know that it is working for their instance?
398398

399399
- [X] Other (treat as last resort)
400-
- Details: The cluster administrator SHOULD have access to logs and metrics during cluster upgrade.
400+
- The `Lease` resource used in the migration can be watched for transition of leadership and timing information.
401+
- logs and metrics can directly indicate the status of migration.
401402

402403
###### What are the reasonable SLOs (Service Level Objectives) for the enhancement?
403404

404-
N/A. This feature is short-lived and only active during cluster upgrade.
405+
Leader Migration is designed to ensure availability of controller managers during upgrade,
406+
and this feature will not affect SLOs of controller managers.
405407

406408
###### What are the SLIs (Service Level Indicators) an operator can use to determine the health of the service?
407409

@@ -414,6 +416,9 @@ N/A. This feature is short-lived and only active during cluster upgrade.
414416
It would help if every controller that the controller manager hosts expose metrics about their availability.
415417
However, per-controller metrics are out of scope of this KEP.
416418

419+
Status of the migration lease, provided by the API server, can help observe the transition of holders
420+
if exposed as resource metrics.
421+
417422
### Dependencies
418423

419424
###### Does this feature depend on any specific services running in the cluster?
@@ -453,7 +458,7 @@ No.
453458
###### Will enabling / using this feature result in non-negligible increase of resource usage (CPU, RAM, disk, IO, ...) in any components?
454459

455460
Both `kube-controller-manager` and `cloud-controller-manager` runs another leader election process,
456-
which may slightly increase CPU usages during leader changes, and it needs extra memory to hold another lease.
461+
which cause negligible increases of CPU and memory usages, both during upgrade and under normal operations.
457462

458463
### Troubleshooting
459464

0 commit comments

Comments
 (0)