Bug 1986215

Summary:	cluster-storage-operator needs to handle API server downtime gracefully in SNO
Product:	OpenShift Container Platform	Reporter:	Naga Ravi Chaitanya Elluri <nelluri>
Component:	Storage	Assignee:	Fabio Bertinatto <fbertina>
Storage sub component:	Operators	QA Contact:	Rohit Patil <ropatil>
Status:	CLOSED ERRATA	Docs Contact:
Severity:	high
Priority:	high	CC:	aos-bugs, jsafrane, nelluri, ropatil, vgrinber
Version:	4.9
Target Milestone:	---
Target Release:	4.9.0
Hardware:	Unspecified
OS:	Linux
Whiteboard:	chaos
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:
Clones:	1992255 (view as bug list)		Environment:
Last Closed:	2021-10-18 17:41:26 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:
Bug Depends On:
Bug Blocks:	1984730, 1992255

Description Naga Ravi Chaitanya Elluri 2021-07-27 00:02:40 UTC

Description of problem:
cluster-storage-operator is crashing/restarting and is going though leader elections during the kube-apiserver rollout which takes around ~60 seconds with shutdown-delay-duration and gracefulTerminationDuration is now set to 0 and 15 seconds ( https://github.com/openshift/cluster-kube-apiserver-operator/pull/1168 and https://github.com/openshift/library-go/pull/1104 ). cluster-storage-operator leader election timeout should be set to > 60 seconds to handle the downtime gracefully in SNO.

Recommended lease duration values to be considered for reference as noted in https://github.com/openshift/enhancements/pull/832/files#diff-2e28754e69aa417e5b6d89e99e42f05bfb6330800fa823753383db1d170fbc2fR183:

LeaseDuration=137s, RenewDealine=107s, RetryPeriod=26s.
These are the configurable values in k8s.io/client-go based leases and controller-runtime exposes them.
This gives us
1. clock skew tolerance == 30s
2. kube-apiserver downtime tolerance == 78s
3. worst non-graceful lease reacquisition == 163s
4. worst graceful lease reacquisition == 26s

Here is the trace of the events during the rollout:
http://dell-r510-01.perf.lab.eng.rdu2.redhat.com/chaos/sno/cluster-storage-operator/cerberus_cluster_state.log. As we can see for the log: http://dell-r510-01.perf.lab.eng.rdu2.redhat.com/chaos/sno/cluster-storage-operator/cluster-storage-operator-pod.log, there are leader lease failures. The leader election can also be disabled given that there's no HA in SNO.

Version-Release number of selected component (if applicable):
4.9.0-0.nightly-2021-07-26-031621

How reproducible:
Always

Steps to Reproduce:
1. Install a SNO cluster using the latest nightly payload.
2. Trigger kube-apiserver rollout/outage - $ oc patch kubeapiserver/cluster --type merge -p '{"spec":{"forceRedeploymentReason":"ITERATIONX"}}' where X can be 1,2...n
3. Observe the state of cluster-storage-operator.

Actual results:
cluster-storage-operator is crashing/restarting and going through leader elections.

Expected results:
cluster-storage-operator should handle the API rollout/outage gracefully.

Additional info:
Logs including must-gather: http://dell-r510-01.perf.lab.eng.rdu2.redhat.com/chaos/sno/cluster-storage-operator/.

Comment 1 Jan Safranek 2021-07-29 08:39:01 UTC

Reused this BZ to fix cluster-csi-snapshot-controller-operator leader election timeouts.

Comment 2 Jan Safranek 2021-07-29 08:39:46 UTC

CSO + CSI driver operators still need to be fixed.

Comment 5 Rohit Patil 2021-08-09 05:37:00 UTC

Tested on build: 4.9.0-0.nightly-2021-08-07-175228
Failed: There is increase in the number of restarts for csi-snap-shot-controller after triggering the patch command and moving to leaderElection.
 
1) Created sno cluster
rohitpatil@ropatil-mac cluster % oc get nodes
NAME                                         STATUS   ROLES           AGE   VERSION
ip-10-0-189-194.us-west-1.compute.internal   Ready    master,worker   29m   v1.21.1+8268f88

2) Observed the logs:
NAME                                               READY   STATUS    RESTARTS   AGE
cluster-storage-operator-6bbf5f9d9d-g2mt9          1/1     Running   1          30m
csi-snapshot-controller-b95b686f9-fq24t            1/1     Running   5          28m
csi-snapshot-controller-operator-565bf56b7-4fc8k   1/1     Running   0          30m
csi-snapshot-webhook-5f8d9b45f-cp9g5               1/1     Running   0          28m

#Event log observation before triggering the oc patch command: 
21m    Normal    LeaderElection          lease/snapshot-controller-leader          csi-snapshot-controller-b95b686f9-fq24t became leader
13m    Normal    LeaderElection          lease/snapshot-controller-leader          csi-snapshot-controller-b95b686f9-fq24t became leader
11m    Normal    LeaderElection          lease/snapshot-controller-leader          csi-snapshot-controller-b95b686f9-fq24t became leader

3) Apply the patch
rohitpatil@ropatil-mac cluster % oc patch kubeapiserver/cluster --type merge -p '{"spec":{"forceRedeploymentReason":"ITERATION1"}}'
kubeapiserver.operator.openshift.io/cluster patched

4) Check the pod status and restart counts
NAME                                               READY   STATUS    RESTARTS   AGE
cluster-storage-operator-6bbf5f9d9d-g2mt9          1/1     Running   1          34m
csi-snapshot-controller-b95b686f9-fq24t            1/1     Running   7          33m
csi-snapshot-controller-operator-565bf56b7-4fc8k   1/1     Running   0          34m
csi-snapshot-webhook-5f8d9b45f-cp9g5               1/1     Running   0          33m

#Event logs 
18m    Normal    LeaderElection       lease/snapshot-controller-leader       csi-snapshot-controller-b95b686f9-fq24t became leader
16m    Normal    LeaderElection       lease/snapshot-controller-leader       csi-snapshot-controller-b95b686f9-fq24t became leader
58s    Normal    LeaderElection       lease/snapshot-controller-leader       csi-snapshot-controller-b95b686f9-fq24t became leader

Comment 6 Fabio Bertinatto 2021-08-09 18:08:34 UTC

*** Bug 1960120 has been marked as a duplicate of this bug. ***

Comment 7 Fabio Bertinatto 2021-08-09 19:13:22 UTC

The restarts are happening in csi-snapshot-controller, which is an upstream project. I've submitted a PR to address this in upstream [1].

Once that's merged in upstream we'll be able to pull those changes in our downstream fork.

[1] https://github.com/kubernetes-csi/external-snapshotter/pull/575

Comment 8 Fabio Bertinatto 2021-08-10 20:16:03 UTC

@Rohit, can we address only CSO in this ticket?

I created bug 1992255 to cover the csi-snapshot-controller issue, since this ticket was originally created for CSO.

Comment 13 errata-xmlrpc 2021-10-18 17:41:26 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 (Moderate: OpenShift Container Platform 4.9.0 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.

https://access.redhat.com/errata/RHSA-2021:3759