Skip to content

OTA-541: enhancements/update/do-not-block-on-degraded: New enhancement proposal #1719

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 3 commits into from
Closed

OTA-541: enhancements/update/do-not-block-on-degraded: New enhancement proposal #1719

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
9498fb9
enhancements/update/do-not-block-on-degraded-true-clusteroperators: N…
wking Nov 25, 2024
111c8fe
enhancements/update/do-not-block-on-degraded-true-clusteroperators: A…
wking Jan 14, 2025
d4a4682
enhancements/update/do-not-block-on-degraded-true-clusteroperators: M…
wking Jan 31, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion dev-guide/cluster-version-operator/dev/clusteroperator.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -233,7 +233,7 @@ Conditions determine when the CVO considers certain actions complete, the follow
| Begin upgrade(patch) | any | any | any | any | any
| Begin upgrade(minor) | any | any | any | any | not false
| Begin upgrade (w/ force) | any | any | any | any | any
| Upgrade completion[2]| newVersion(target version for the upgrade) | true | false | any | any
| Upgrade completion[2]| newVersion(target version for the upgrade) | true | any | any | any

[1] Install works on all components in parallel, it does not wait for any component to complete before starting another one.

Expand Down
3 changes: 2 additions & 1 deletion dev-guide/cluster-version-operator/user/reconciliation.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -157,7 +157,8 @@ The ClusterOperator builder only monitors the in-cluster object and blocks until
```

would block until the in-cluster ClusterOperator reported `operator` at version 4.1.0.
* Not degraded (except during initialization, where we ignore the degraded status)
* [Prior to 4.19](/enhancements/update/do-not-block-on-degraded-true-clusteroperators.md), not degraded (except during initialization, where we ignore the degraded status).
Since 4.19, ClusterOperator `Degraded` conditions are no longer blocking.

### CustomResourceDefinition

Expand Down
169 changes: 169 additions & 0 deletions enhancements/update/do-not-block-on-degraded-true-clusteroperators.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,169 @@
---
title: do-not-block-on-degraded-true-clusteroperators
authors:
- "@wking"
reviewers:
- "@PratikMahajan, update team lead"
- "@sdodson, update staff engineer"
approvers:
- "@PratikMahajan, update team lead"
api-approvers:
- None
creation-date: 2024-11-25
last-updated: 2025-01-09
tracking-link:
- https://issues.redhat.com/browse/OTA-540
---

# Do not block on Degraded=True ClusterOperators

## Summary

The cluster-version operator (CVO) uses an update-mode when transitioning between releases, where the manifest operands are [sorted into a task-node graph](/dev-guide/cluster-version-operator/user/reconciliation.md#manifest-graph), and the CVO walks the graph reconciling.
Since 4.1, the cluster-version operator has blocked during update and reconcile modes (but not during install mode) on `Degraded=True` ClusterOperator.
This enhancement proposes ignoring `Degraded` when deciding whether to block on a ClusterOperator manifest.

## Motivation

The goal of blocking on manifests with sad resources is to avoid further destabilization.
For example, if we have not reconciled a namespace manifest or ServiceAccount RoleBinding, there's no point in trying to update the consuming operator Deployment.
Or if we are unable to update the Kube-API-server operator, we don't want to inject [unsupported kubelet skew][kubelet-skew] by asking the machine-config operator to update nodes.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

we observed another kind of upgrade blocker here. Applying the infrastructures.config.openshift.io manifest failed as the CRD had introduced some validations and that needed the apiserver to be upgraded to support it. Unfortunately, the upgrade didn't progress and we had to manually step in to update the kube-apiserver to let the upgrade proceed. Is there a way to enhance these cases to at least let the apiserver upgrade before blocking?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is there a way to enhance these cases...

I've been trying to talk folks into the narrow Degraded handling pivot this enhancement currently covers since 2021. I accept that there may be other changes that we could make to help updates go more smoothly, but I'd personally rather limit the scope of this enhancement to the Degraded handling.


However, blocking the update on a sad resource has the downside that later manifest-graph task-nodes are not reconciled, while the CVO waits for the sad resource to return to happiness.
We maximize safety by blocking when progress would be risky, while continuing when progress would be safe, and possibly helpful.

Our experience with `Degraded=True` blocks turns up cases like:

* 4.6 `Degraded=True` on an unreachable, user-provided node, with monitoring reporting `UpdatingnodeExporterFailed`, network reporting `RolloutHung`, and machine-config reporting `MachineConfigDaemonFailed`.
But those ClusterOperator were all still `Available=True`, and in 4.10 and later, monitoring workloads are guarded by PodDisruptionBudgets (PDBs)

### User Stories

* As a cluster administrator, I want the ability to defer recovering `Degraded=True` ClusterOperators without slowing ClusterVersion updates.

### Goals

ClusterVersion updates will no longer block on ClusterOperators solely based on `Degraded=True`.
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

does it mean, if no operator is unavailable, then the upgrade should always complete?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

ClusterOperators aren't the only CVO-manifested resources, and if something else breaks like we fail to reconcile a RoleBinding or whatever, that will block further update progress. And for ClusterOperators, we'll still block on status.versions not being as far along as the manifest claimed, in addition to blocking if Available isn't True. Personally, status.versions seems like the main thing that's relevant, e.g. a component coming after the Kube API server knows it can use 4.18 APIs if the Kube API server has declared 4.18 versions. As an example of what the 4.18 Kube API server asks the CVO to wait on:

$ oc adm release extract --to manifests quay.io/openshift-release-dev/ocp-release:4.18.0-rc.0-x86_64
Extracted release payload from digest sha256:054e75395dd0879e8c29cd059cf6b782742123177a303910bf78f28880431d1c created at 2024-12-02T21:11:00Z
$ yaml2json <manifests/0000_20_kube-apiserver-operator_07_clusteroperator.yaml | jq -c '.status.versions[]'
{"name":"operator","version":"4.18.0-rc.0"}
{"name":"raw-internal","version":"4.18.0-rc.0"}
{"name":"kube-apiserver","version":"1.31.3"}

A recent example of this being useful is openshift/machine-config-operator#4637, which got the CVO to block until the MCO had rolled out a single-arch -> multi-arch transition, without the MCO needing to touch its Degraded or Available conditions to slow the CVO down.

Copy link

@jiajliu jiajliu Dec 4, 2024 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

so could I say, if failing=true for an upgrade, the reason should not be ClusterOperatorDegraded only.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No, we'll still propagate ClusterOperator(s)Degraded through to Failing, it just will no longer block the update's progress. So if the only issue Failing is talking about is ClusterOperator(s)Degraded, we expect the update to be moving towards completion, and not stalling.


### Non-Goals

* Adjusting how the cluster-version operator treats `Available` and `versions` in ClusterOperator status.
The CVO will still block on `Available=False` ClusterOperator, and will also still block on `status.versions` reported in the ClusterOperator's release manifest.

* Adjusting whether `Degraded` ClusterOperator conditions propagated through to the ClusterVersion `Failing` condition.
As with the current install mode, the sad condition will be propagated through to `Failing=True`, unless outweighed by a more serious condition like `Available=False`.

## Proposal

The cluster-version operator currently has [a mode switch][cvo-degraded-mode-switch] that makes `Degraded` ClusterOperator a non-blocking condition that is still propagated through to `Failing`.
This enhancement proposes making that an unconditional `UpdateEffectReport`, regardless of the CVO's current mode (installing, updating, reconciling, etc.).
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

openshift/cluster-version-operator#482 is in flight with this change, if folks want to test pre-merge.


### Workflow Description

Cluster administrators will be largely unaware of this feature.
They will no longer have ClusterVersion update progress slowed by `Degraded=True` ClusterOperators, so there will be less admin involvement there.
They will continue to be notified of `Degraded=True` ClusterOperators via [the `warning` `ClusterOperatorDegraded` alert][ClusterOperatorDegraded] and the `Failing=True` ClusterVersion condition.

### API Extensions

No API extensions are needed for this proposal.

### Topology Considerations

#### Hypershift / Hosted Control Planes

HyperShift's ClusterOperator context is the same as standalone, so it will receive the same benefits from the same cluster-version operator code change, and does not need special consideration.

#### Standalone Clusters

Yes, the enhancement is expected to improve the update experience on standalone, by decoupling ClusterVersion update completion from recovering `Degraded=True` ClusterOperators, granting the cluster administrator the flexibility to address update speed and operator degradation independently.

#### Single-node Deployments or MicroShift

Single-node's ClusterOperator context is the same as standalone, so it will receive the same benefits from the same cluster-version operator code change, and does not need special consideration.
This change is a minor tweak to existing CVO code, so it is not expected to impact resource consumption.

MicroShift updates are managed via RPMs, without a cluster-version operator, so it is not exposed to the ClusterVersion updates this enhancement is refining, and not affected by the changes proposed in this enhancement.

### Implementation Details/Notes/Constraints

The code change is expected to be a handful of lines, as discussed in [the *Proposal* section](#proposal), so there are no further implementation details needed.

### Risks and Mitigations

The risk would be that there are some ClusterOperators who currently rely on the cluster-version operator blocking during updates on ClusterOperators that are `Available=True`, `Degraded=True`, and which set the release manifest's expected `versions`.
As discussed in [the *Motivation* section](#motivation), we're not currently aware of any such ClusterOperators.
If any turn up, we can mitigate by [declaring conditional update risks](targeted-update-edge-blocking.md) using the existing `cluster_operator_conditions{condition="Degraded"}` PromQL metric, while teaching the relevant operators to set `Available=False` and/or without their `versions` bumps until the issue that needs to block further ClusterVersion update progress has been resolved.

How will security be reviewed and by whom?
Unclear. Feedback welcome.

How will UX be reviewed and by whom?
Unclear. Feedback welcome.

### Drawbacks

As discussed in [the *Risks* section](#risks-and-mitigations), the main drawback is changing behavior that we've had in place for many years.
But we do not expect much customer pushback based on "hey, my update completed?! I expected it to stick on this sad component...".
We do expect it to reduce customer frustration when they want the update to complete, but for reasons like administrative siloes do not have the ability to recover a component from minor degradation themselves.

## Test Plan

**Note:** *Section not required until targeted at a release.*
Copy link
Contributor

@DavidHurta DavidHurta Dec 6, 2024 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The enhancement and the tracking card OTA-541 are not targeted at a release. However, changes in the dev-guide/cluster-version-operator/user/reconciliation.md file suggest that the enhancement is targeted at the 4.19 release, and thus the Test Plan section should be addressed.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm not strongly opinionated on what the test plan looks like. We don't do a lot of intentional-sad-path update testing today in CI, and I'm fuzzy on what QE does in that space that could be expanded into this new space (or maybe they already test pushing a ClusterOperator component to Degraded=True mid update to see how the cluster handles that?).

Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

test pushing a ClusterOperator component to Degraded=True mid update to see how the cluster handles that?

+1, that's also what I want to explore during test. I also had some other immature checkpoints in my mind when I read this enhancement doc at the first time, but I still need some inputs from @wking to help me tidy up them. For example #1719 (comment).
I asked this because there's already some cv.conditions check in CI, I'm thinking about if we could update the logic to help catching issues once the feature implemented.


Consider the following in developing a test plan for this enhancement:
- Will there be e2e and integration tests, in addition to unit tests?
- How will it be tested in isolation vs with other components?
- What additional testing is necessary to support managed OpenShift service-based offerings?

No need to outline all of the test cases, just the general strategy. Anything
that would count as tricky in the implementation and anything particularly
challenging to test should be called out.

All code is expected to have adequate tests (eventually with coverage
expectations).

## Graduation Criteria

There are no API changes proposed by this enhancement, which only affects sad-path handling, so we expect the code change to go straight to the next generally-available release, without feature gating or staged graduation.

### Dev Preview -> Tech Preview

Not applicable.

### Tech Preview -> GA

Not applicable.

### Removing a deprecated feature

Not applicable.

## Upgrade / Downgrade Strategy

This enhancement only affects the cluster-version operator's internal processing of longstanding ClusterOperator APIs, so there are no skew or compatibility issues.

## Version Skew Strategy

This enhancement only affects the cluster-version operator's internal processing of longstanding ClusterOperator APIs, so there are no skew or compatibility issues.

## Operational Aspects of API Extensions

There are no API changes proposed by this enhancement.

## Support Procedures

This enhancement is a small pivot in how the cluster-version operator processes ClusterOperator manifests during updates.
As discussed in [the *Drawbacks* section](#drawbacks), we do not expect cluster admins to open support cases related to this change.

## Alternatives

We could continue with the current approach, and absorb the occasional friction it causes.

## Infrastructure Needed

No additional infrastructure is needed for this enhancement.

[ClusterOperatorDegraded]: https://github.com/openshift/cluster-version-operator/blob/820b74aa960717aae5431f783212066736806785/install/0000_90_cluster-version-operator_02_servicemonitor.yaml#L106-L124
[cvo-degraded-mode-switch]: https://github.com/openshift/cluster-version-operator/blob/820b74aa960717aae5431f783212066736806785/pkg/cvo/internal/operatorstatus.go#L241-L245
[kubelet-skew]: https://kubernetes.io/releases/version-skew-policy/#kubelet