Bug 1958885

Summary:	4.7.6 -> 4.7.9 upgrade: leader election stuck
Product:	OpenShift Container Platform	Reporter:	Vadim Rutkovsky <vrutkovs>
Component:	Node Tuning Operator	Assignee:	Jiří Mencák <jmencak>
Status:	CLOSED ERRATA	QA Contact:	Simon <skordas>
Severity:	high	Docs Contact:
Priority:	high
Version:	4.7	CC:	jmencak, scuppett, sejug, wking
Target Milestone:	---	Keywords:	Reopened
Target Release:	4.7.z	Flags:	scuppett: needinfo-
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:		Doc Type:	No Doc Update
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2021-07-06 11:38:38 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:	1962219
Bug Blocks:

Description Vadim Rutkovsky 2021-05-10 11:28:43 UTC

Description of problem:
vsphere build cluster updated from 4.7.6 to 4.7.9 and stuck on node tuning operator not progressing to 4.7.9. The new pod got stuck on leader election:

```
I0508 13:55:12.095058       1 main.go:25] Go Version: go1.15.7
I0508 13:55:12.095391       1 main.go:26] Go OS/Arch: linux/amd64
I0508 13:55:12.095442       1 main.go:27] node-tuning Version: v4.7.0-202104250659.p0-0-g1d2e014-dirty
I0508 13:55:12.101977       1 controller.go:954] trying to become a leader
```

Version-Release number of selected component (if applicable):
4.7.6

Comment 2 Jiří Mencák 2021-05-10 13:17:28 UTC

Without properly investigating must-gather yet, this reminds me of: https://bugzilla.redhat.com/show_bug.cgi?id=1916865
Were there other operators also stuck during the upgrade?

Comment 3 Vadim Rutkovsky 2021-05-10 14:02:57 UTC

(In reply to jmencak from comment #2)
> Without properly investigating must-gather yet, this reminds me of:
> https://bugzilla.redhat.com/show_bug.cgi?id=1916865

Looks similar, I guess it could be closed as a dupe

> Were there other operators also stuck during the upgrade?

Marketplace also got stuck similarly

Comment 4 Jiří Mencák 2021-05-10 14:15:48 UTC

As other operators are stuck in a similar way, I'm closing this BZ as duplicate of https://bugzilla.redhat.com/show_bug.cgi?id=1916865
There is a Jira card to re-assess the current implementation of leader election in NTO not to rely on ConfigMaps and the garbage collection mechanism as it doesn't seem reliable.
Currently, the plan is to have the rewrite done in OCP 4.9.

*** This bug has been marked as a duplicate of bug 1916865 ***

Comment 5 W. Trevor King 2021-05-10 17:17:09 UTC

Can you link the Jira card, or point us at something else where we can subscribe to hear about the fix?

Comment 6 Jiří Mencák 2021-05-10 17:36:43 UTC

(In reply to W. Trevor King from comment #5)
> Can you link the Jira card, or point us at something else where we can
> subscribe to hear about the fix?

Sorry, I meant to include it in comment 4.
https://issues.redhat.com/browse/PSAP-314

Note this is for NTO only, other operators will have to do a similar fix.

Comment 7 Stephen Cuppett 2021-05-10 23:03:27 UTC

Deleted node-tuning-operator-lock ConfigMap created April 20 in the openshift-cluster-node-tuning-operator Namespace

I0510 11:51:01.870067       1 controller.go:954] trying to become a leader
I0510 22:54:32.086704       1 controller.go:959] became a leader
I0510 22:54:32.092826       1 controller.go:966] starting Tuned controller
I0510 22:54:32.496588       1 controller.go:1017] started events processor/controller

Comment 14 Simon 2021-06-29 10:20:41 UTC

Upgrade verified on 4.7.6 -> 4.7.0-0.nightly-2021-06-26-014854

$ oc get cm/node-tuning-operator-lock -o yaml
apiVersion: v1
kind: ConfigMap
metadata:
  annotations:
    control-plane.alpha.kubernetes.io/leader: '{"holderIdentity":"cluster-node-tuning-operator-6bc548dcc5-n779h_cd1248a0-a728-4189-ab62-152e74f966d5","leaseDurationSeconds":30,"acquireTime":"2021-06-29T09:31:29Z","renewTime":"2021-06-29T10:19:23Z","leaderTransitions":1}'
  creationTimestamp: "2021-06-29T09:08:06Z"
  managedFields:
  - apiVersion: v1
    fieldsType: FieldsV1
    fieldsV1:
      f:metadata:
        f:annotations:
          .: {}
          f:control-plane.alpha.kubernetes.io/leader: {}
    manager: cluster-node-tuning-operator
    operation: Update
    time: "2021-06-29T09:08:06Z"
  name: node-tuning-operator-lock
  namespace: openshift-cluster-node-tuning-operator
  resourceVersion: "79282"
  selfLink: /api/v1/namespaces/openshift-cluster-node-tuning-operator/configmaps/node-tuning-operator-lock
  uid: 5d4fe7ed-02e5-4f75-b95b-dc00b7869bf2

Comment 16 errata-xmlrpc 2021-07-06 11:38:38 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 (OpenShift Container Platform 4.7.19 bug fix 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/RHBA-2021:2554