Bug 1879963 - Clusters seen upgrading from/to same version
Summary: Clusters seen upgrading from/to same version
Keywords:
Status: CLOSED NOTABUG
Alias: None
Product: OpenShift Container Platform
Classification: Red Hat
Component: Cluster Version Operator
Version: 4.6
Hardware: Unspecified
OS: Unspecified
medium
low
Target Milestone: ---
: 4.7.0
Assignee: Jack Ottofaro
QA Contact: Johnny Liu
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2020-09-17 13:01 UTC by Jack Ottofaro
Modified: 2020-12-08 13:48 UTC (History)
6 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2020-12-08 13:48:39 UTC
Target Upstream Version:
Embargoed:


Attachments (Terms of Use)

Description Jack Ottofaro 2020-09-17 13:01:29 UTC
Description of problem:

Seeing some clusters with type="updating",from_version=X,version=X:

{_id="1b1eb908-c095-4a94-bc7c-4bde7942af7a", from_version="4.1.41", type="updating", version="4.1.41"}

Ref: https://coreos.slack.com/archives/CEGKQ43CP/p1600260341388300

And more details and example clusters linked here: https://issues.redhat.com/browse/SDA-2743

Duplicate entiries should not be in version history but will check if there is a bug in the merge logic.

Comment 1 Beni Paskin-Cherniavsky 2020-09-25 09:04:40 UTC
Even if fixed in CVO I assume this is too minor to deserve backporting, so we'll still see those in telemetry for a long time.

Externally visible impacts:

- [x] The OCM UI https://cloud.redhat.com is hiding such upgrades.

- [ ] The API api.openshift.com is reporting X->X upgrades when present in telemetry.
      Do you think we should filter them, or just leave it as is, until eventually fixed?

Comment 2 W. Trevor King 2020-10-02 23:17:52 UTC
Sprint is up, but we want to fix this.  Hopefully next sprint.

Comment 3 Jack Ottofaro 2020-12-03 21:32:53 UTC
Upgrading from/to the same version, and having in history, happens when the "--to-image" to upgrade to differs in any way from the currently loaded image, e.g. different repo or using "<repo>:semvar" vs. "<repo>@digest".

For example, a cluster at 4.5.16 with image "registry.svc.ci.openshift.org/ocp/release@sha256:adb5ef06c54ff75ca9033f222ac5e57f2fd82e49bdd84f737d460ae542c8af60":

- upgrade will occur for `--to-image quay.io/openshift-release-dev/ocp-release:4.5.16-x86_64`
- from there `--to-image quay.io/openshift-release-dev/ocp-release@sha256:adb5ef06c54ff75ca9033f222ac5e57f2fd82e49bdd84f737d460ae542c8af60` results in another upgrade
- the same image in the same repo but referenced by `4.5.16-x86_64` vs. digest or vice-versa results in an update

However using `--to 4.5.16` is blocked by oc client with "Cluster is already at version 4.5.16".

Comment 4 Jack Ottofaro 2020-12-08 13:48:39 UTC
Any changes to the image will require the upgrade logic to run. Since I can only recreate this issue by making the from/to image differ as described in comment https://bugzilla.redhat.com/show_bug.cgi?id=1879963#c3, closing this as not a bug.


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