Bug 2100347 - KASO retains old config values when switching from Medium/Default to empty worker latency profile
Summary: KASO retains old config values when switching from Medium/Default to empty wo...
Alias: None
Product: OpenShift Container Platform
Classification: Red Hat
Component: kube-apiserver
Version: 4.11
Hardware: All
OS: All
Target Milestone: ---
: 4.11.0
Assignee: Swarup Ghosh
QA Contact: Deepak Punia
Depends On:
TreeView+ depends on / blocked
Reported: 2022-06-23 07:12 UTC by Swarup Ghosh
Modified: 2022-08-10 11:19 UTC (History)
5 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Last Closed: 2022-08-10 11:19:11 UTC
Target Upstream Version:

Attachments (Terms of Use)

System ID Private Priority Status Summary Last Updated
Github openshift cluster-kube-apiserver-operator pull 1360 0 None open Bug 2100347: Bump(openshift/library-go) latency profile observer, controller 2022-06-23 07:17:03 UTC
Red Hat Product Errata RHSA-2022:5069 0 None None None 2022-08-10 11:19:19 UTC

Description Swarup Ghosh 2022-06-23 07:12:27 UTC
Description of problem:
Kube API Server Operator managed observed config retains the old set of latency config arg val pairs when switching from a latency profile to an empty profile. (Empty latency profile is attainable when nodes.config.openshift.io/cluster object has empty spec, i.e. `spec: {}`)

Version-Release number of selected component (if applicable): 4.11

How reproducible: Always

Steps to Reproduce:
1. `oc edit nodes.config.openshift.io`, change `workerLatencyProfile` to `MediumUpdateAverageReaction`
2. wait for sometime to allow the rollout to happen in `oc get KubeAPIServer -o yaml` under the `observedConfig` section
3. `observedConfig` will contain default-not-ready-toleration-seconds and default-unreachable-toleration-seconds with a value of 60.
4. `oc edit nodes.config.openshift.io`, change profile back to empty, i.e. by setting `spec: {}`

Actual results:
5. seems like `observedConfig` will still retain default-not-ready-toleration-seconds and default-unreachable-toleration-seconds with a value of 60 which is Medium profile, though medium profile was already unset

Expected results:
`observedConfig` shouldn't have default-not-ready-toleration-seconds and default-unreachable-toleration-seconds arg values when empty profile is re-activated on the cluster.

Additional info: This problem was also seen in KCMO (cluster-kube-controller-manager-operator) but due to the latest bump in library-go for KCMO, that problem is resolved now. For KASO also, library-go bump should help resolve the problem.

[1] Enhancement proposal: https://github.com/openshift/enhancements/blob/master/enhancements/worker-latency-profile/worker-latency-profile.md#workerlatencyprofile
[2] Problem description: https://github.com/openshift/library-go/pull/1372/#discussion_r901719238
[3] KCMO PR that already fixed the problem for KCM: https://github.com/openshift/cluster-kube-controller-manager-operator/pull/629

Comment 5 errata-xmlrpc 2022-08-10 11:19:11 UTC
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory (Important: OpenShift Container Platform 4.11.0 bug fix and security update), and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.


Note You need to log in before you can comment on or make changes to this bug.