Moves bundle unpack timeout into OperatorGroup #2952

m1kola · 2023-04-13T11:24:09Z

Description of the change:

Moves operatorframework.io/bundle-unpack-timeout annotation from InstallPlan to OperatorGroup

Motivation for the change:

operatorframework.io/bundle-unpack-timeout is an internal annotation used for E2E testing.

We need to move this out of InstallPlan in preparation to changes in the unpacking process (see #2942): OLM will soon be creating unpack jobs before creating InstallPlan so we need to find a new place where we can set this annotation.

Testing remarks:

Updates existing E2E tests

Reviewer Checklist

openshift-ci · 2023-04-13T11:24:15Z

Skipping CI for Draft Pull Request.
If you want CI signal for your change, please convert it to an actual PR.
You can still manually trigger a test run with /test all

perdasilva

lgtm

ankitathomas

/lgtm

tmshort · 2023-04-20T19:39:32Z

pkg/controller/bundle/bundle_unpacker.go

+
+	ogs, err := ogLister.List(k8slabels.Everything())
+	if err != nil || len(ogs) == 0 {
+		return ignoreTimeout, nil


Shouldn't this return an err?

Also, the check for len(ogs) == 0 is a bit redundant the check below. Are we treating "0" different than "not 1"?

Shouldn't this return an err?

I looked into a similar piece of code where we check whether UnsafeFailForward upgrade strategy is set on an OperatorGroup:

operator-lifecycle-manager/pkg/controller/registry/resolver/fail_forward.go

Lines 12 to 27 in b5a885e

// IsFailForwardEnabled takes a namespaced operatorGroup lister and returns

// True if an operatorGroup exists in the namespace and its upgradeStrategy

// is set to UnsafeFailForward and false otherwise. An error is returned if

// an more than one operatorGroup exists in the namespace.

// No error is returned if no OperatorGroups are found to keep the resolver

// backwards compatible.

func IsFailForwardEnabled(ogLister v1listers.OperatorGroupNamespaceLister) (bool, error) {

ogs, err := ogLister.List(labels.Everything())

if err != nil || len(ogs) == 0 {

return false, nil

}

if len(ogs) != 1 {

return false, fmt.Errorf("found %d operatorGroups, expected 1", len(ogs))

}

return ogs[0].UpgradeStrategy() == operatorsv1.UpgradeStrategyUnsafeFailForward, nil

}

It allows a case when there is no OperatorGroup. I don't know if it still makes sense. I was under the impression that OG is required for resolution to succeed, but I didn't dare to handle timeout differently from fail forward.

Also, the check for len(ogs) == 0 is a bit redundant the check below. Are we treating "0" different than "not 1"?
I see that we can

The idea is that we return either a value specified in the annotation or a default value ignoreTimeout. Given the above logic around a case when OG does not exist (len(ogs) == 0) we do not want to return an error since we can return default value. In case when more than one OG exist - we return an error because we do not know which one to respect.

Basically this is exactly the same pattern as in IsFailForwardEnabled, but I'm happy to change me if you think that special handling of "OG does not exist" situation no longer makes sense.

I wonder if we ought to return the error if the List call fails and remove the || len(ogs) == 0. This can let the caller decide whether they want to retry, or are happy with using the default value. Maybe we could also add a ticket for the old code so we get this rechecked. I'm worried silent errors around this code could lead to hard to determine bugs. wdyt?

I made this function to return an error from List and removed || len(ogs) == 0. From what I know about OLM so far - OperatorGroup is required and without it resolution will fail eventually. In case of these errors - sync will be requeued (no extra code on caller required).

I'll try to the same for old code in IsFailForwardEnabled, but in a separate PR. Will see how it goes.

Created #2957 for IsFailForwardEnabled.

tmshort · 2023-04-20T19:48:01Z

pkg/controller/bundle/bundle_unpacker.go

+	d, err := time.ParseDuration(timeoutStr)
+	if err != nil {
+		logger.Errorf("failed to parse unpack timeout annotation(%s: %s): %v", BundleUnpackTimeoutAnnotationKey, timeoutStr, err)
+		return ignoreTimeout, nil


Logging vs error return? The behavior here is a bit inconsistent.

This works in the same way it was before the refactoring here: we log an error and ignore the value.

The only case when we want to return an error in this func - is where there are two OperatorGroups in which case we can not proceed.

I also decided to change how we handle duration parsing error. In master it is also a silent error (we just log) - now I return parse error to the caller which will requeue.

This error is very unlikely to happen as annotation is internal use only (not documented) and is set programmatically (we only use it for E2Es). So I think other than requeuing - we do not need any special handling for it.

m1kola · 2023-04-24T11:30:04Z

Rebased on top of master to make github happy.

perdasilva · 2023-04-24T13:53:59Z

Rebased from this side

`operatorframework.io/bundle-unpack-timeout` is an internal annotation used for E2E testing. We need to move this out of InstallPlan in preparation to changes in the unpacking process: OLM will soon be creating unpack jobs before creating InstallPlan so we need to find a new place where we can set this annotation. Signed-off-by: Mikalai Radchuk <mradchuk@redhat.com>

test/e2e/installplan_e2e_test.go

tmshort

/lgtm

openshift-ci · 2023-04-24T19:17:22Z

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: ankitathomas, m1kola, perdasilva, tmshort

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

~~OWNERS~~ [perdasilva]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

openshift-ci bot added the do-not-merge/work-in-progress Indicates that a PR should not merge because it is a work in progress. label Apr 13, 2023

m1kola force-pushed the move-bundle-unpack-timeout-to-og branch from eaa0cd8 to e2680ad Compare April 13, 2023 12:03

m1kola mentioned this pull request Apr 13, 2023

Prevent OLM from creating InstallPlans when bundle unpack fails #2942

Merged

11 tasks

m1kola marked this pull request as ready for review April 13, 2023 12:05

openshift-ci bot removed the do-not-merge/work-in-progress Indicates that a PR should not merge because it is a work in progress. label Apr 13, 2023

openshift-ci bot requested review from ankitathomas and asmacdo April 13, 2023 12:05

m1kola requested a review from awgreene April 13, 2023 12:06

perdasilva approved these changes Apr 13, 2023

View reviewed changes

openshift-ci bot added the approved Indicates a PR has been approved by an approver from all required OWNERS files. label Apr 13, 2023

ankitathomas approved these changes Apr 20, 2023

View reviewed changes

openshift-ci bot assigned ankitathomas Apr 20, 2023

openshift-ci bot added the lgtm Indicates that a PR is ready to be merged. label Apr 20, 2023

perdasilva force-pushed the move-bundle-unpack-timeout-to-og branch from e2680ad to b47f9e1 Compare April 20, 2023 16:08

openshift-ci bot removed the lgtm Indicates that a PR is ready to be merged. label Apr 20, 2023

tmshort reviewed Apr 20, 2023

View reviewed changes

m1kola force-pushed the move-bundle-unpack-timeout-to-og branch from b47f9e1 to 0d20d85 Compare April 24, 2023 11:29

m1kola requested review from ankitathomas, perdasilva and tmshort April 24, 2023 12:19

perdasilva force-pushed the move-bundle-unpack-timeout-to-og branch from 0d20d85 to 8438943 Compare April 24, 2023 13:53

m1kola force-pushed the move-bundle-unpack-timeout-to-og branch 4 times, most recently from 09dc7a7 to aea944e Compare April 24, 2023 15:32

m1kola force-pushed the move-bundle-unpack-timeout-to-og branch from aea944e to 03b3747 Compare April 24, 2023 16:14

m1kola mentioned this pull request Apr 24, 2023

Removes error silencing from IsFailForwardEnabled #2957

Merged

11 tasks

tmshort reviewed Apr 24, 2023

View reviewed changes

test/e2e/installplan_e2e_test.go Show resolved Hide resolved

tmshort approved these changes Apr 24, 2023

View reviewed changes

openshift-ci bot assigned tmshort Apr 24, 2023

openshift-ci bot added the lgtm Indicates that a PR is ready to be merged. label Apr 24, 2023

openshift-ci bot merged commit 47d6aa1 into operator-framework:master Apr 24, 2023

m1kola deleted the move-bundle-unpack-timeout-to-og branch April 24, 2023 21:29

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Moves bundle unpack timeout into OperatorGroup #2952

Moves bundle unpack timeout into OperatorGroup #2952

m1kola commented Apr 13, 2023

openshift-ci bot commented Apr 13, 2023

perdasilva left a comment

ankitathomas left a comment

tmshort Apr 20, 2023

tmshort Apr 20, 2023 •

edited

Loading

m1kola Apr 24, 2023

perdasilva Apr 24, 2023

m1kola Apr 24, 2023

m1kola Apr 24, 2023

tmshort Apr 20, 2023

m1kola Apr 24, 2023

m1kola Apr 24, 2023

m1kola commented Apr 24, 2023

perdasilva commented Apr 24, 2023

tmshort left a comment

openshift-ci bot commented Apr 24, 2023

	// IsFailForwardEnabled takes a namespaced operatorGroup lister and returns
	// True if an operatorGroup exists in the namespace and its upgradeStrategy
	// is set to UnsafeFailForward and false otherwise. An error is returned if
	// an more than one operatorGroup exists in the namespace.
	// No error is returned if no OperatorGroups are found to keep the resolver
	// backwards compatible.
	func IsFailForwardEnabled(ogLister v1listers.OperatorGroupNamespaceLister) (bool, error) {
	ogs, err := ogLister.List(labels.Everything())
	if err != nil \|\| len(ogs) == 0 {
	return false, nil
	}
	if len(ogs) != 1 {
	return false, fmt.Errorf("found %d operatorGroups, expected 1", len(ogs))
	}
	return ogs[0].UpgradeStrategy() == operatorsv1.UpgradeStrategyUnsafeFailForward, nil
	}

Moves bundle unpack timeout into OperatorGroup #2952

Moves bundle unpack timeout into OperatorGroup #2952

Conversation

m1kola commented Apr 13, 2023

openshift-ci bot commented Apr 13, 2023

perdasilva left a comment

Choose a reason for hiding this comment

ankitathomas left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

tmshort Apr 20, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

m1kola commented Apr 24, 2023

perdasilva commented Apr 24, 2023

tmshort left a comment

Choose a reason for hiding this comment

openshift-ci bot commented Apr 24, 2023

tmshort Apr 20, 2023 •

edited

Loading