OCPBUGS-1731: Merge 1.10.0

[gcs278]

Rebased to CoreDNS 1.10.0 so that we use the Kubernetes 1.25 packages.

I made this PR using the same process as #73, #52, #60, and #70:

1. git remote update
2. git checkout v1.10.0
3. git checkout -b merge-1.10.0-tmp
4. git checkout origin/master
5. echo 'merge coredns/coredns v1.10.0' | git commit-tree 'merge-1.10.0-tmp^{tree}' -p HEAD -p merge-1.10.0-tmp -F -
ead3b6507d7bd2fc3e67b79d8ff0dff296396bac
6. git branch merge-1.10.0 ead3b6507d7bd2fc3e67b79d8ff0dff296396bac
7. git checkout merge-1.10.0
8. git rev-list --reverse --no-merges 555dd549ee36ba7f626cf8c53707c74af318ec09..origin/master | git cherry-pick --stdin
   Note: 555dd549ee36ba7f626cf8c53707c74af318ec09 is the merge commit from the last merge Bug 2090827: Merge 1.9.2 #73
9. My merge commit from the above process was commit ead3b65.

Before the rebase, we had the following commits:

% git --no-pager log --oneline --no-merges 555dd549ee36ba7f626cf8c53707c74af318ec09..origin/master
ad24eacf6 Updating coredns images to be consistent with ART Reconciling with https://github.com/openshift/ocp-build-data/tree/2e5eba0108299087402c119ba0dc979a0a6c0bd1/images/coredns.yml
349db7a33 UPSTREAM: <carry>: openshift: Revert "Revert "add wildcard warnings (#5030)" (#5167)"
461cbbf60 UPSTREAM: <carry>: openshift: Revert "remove wildcard query functionality (#5019)"
7a41d2b87 UPSTREAM: <carry>: openshift: Disable dependabot
1d2d4f767 UPSTREAM: <carry>: openshift: Vendor required modules
3dc197d15 UPSTREAM: <carry>: openshift: .gitignore: Properly ignore coredns binary
9176cd5bf UPSTREAM: <carry>: openshift: Make coredns exclusion more specific
488cc73b1 UPSTREAM: <carry>: openshift: Add a `make test` target
b648be195 UPSTREAM: <carry>: openshift: Add product build config
1e6995352 UPSTREAM: <carry>: openshift: Add OWNERS
% 

I edited the "Vendor required modules" commit by running go mod vendor and go mod tidy and ammend commit 1d2d4f767 UPSTREAM: <carry>: openshift: Vendor required modules.

I squashed/merged the following commit into the existing "Add product build config" commit:

ad24eacf6 Updating coredns images to be consistent with ART Reconciling with https://github.com/openshift/ocp-build-data/tree/2e5eba0108299087402c119ba0dc979a0a6c0bd1/images/coredns.ym

Git merge conflicts for 461cbbf60 UPSTREAM: <carry>: openshift: Revert "remove wildcard query functionality (#5019)":
plugin/kubernetes/kubernetes.go

• Line 630: matchPortAndProtocol function comment update, but matchPortAndProtocol function has been removed so I ignored the updates (see PR Delta)

plugin/kubernetes/external.go

• Line 52: changes to variables declaration conflicted because they added endpoint as a string. I merged by taking upstream variable declarations, but added port = "*" and protocol = "*" assignment below for reversion (see PR Delta)
• Line 114 & 134: Refactoring moved calls to matchPortAndProtocol and added it in an additional location for resolving headless services (see PR delta)
  • I'm assuming we want wildcard support for headless and that wildcard support is remove via matchPortAndProtocol, but we don't have that function anyways so just using the match function.

[brandisher]

Overall the changes look good and match what we discussed yesterday. I left one question for @Miciah about whether we need a release note for the change in SERVFAIL response caching; I'm assuming that SERVFAILs were not cached previously and will now be cached for 5s.

[brandisher]

@Miciah this could be a nit so I'll defer to you, but would it be worth including in the release notes that SERVFAIL responses will now be cached for 5s (the value of minNTTL)?

[Miciah]

SERVFAIL was cached for 5s in CoreDNS 1.9.2.
https://github.com/coredns/coredns/blob/092c144491e7f62d71eda98a07e4ce815d9df035/plugin/cache/cache.go#L159-L161
This change only makes it configurable, but the default is 5 seconds, and we aren't changing the default.

[brandisher]

Ah okay, thanks for catching that and the link! That's what I get for making an assumption :-)

[gcs278]

Just OVN alerts now.

[openshift-ci]

@gcs278: The /retest command does not accept any targets.
The following commands are available to trigger required jobs:

• /test e2e-aws
• /test e2e-aws-upgrade
• /test e2e-gcp-serial
• /test images
• /test unit

The following commands are available to trigger optional jobs:

• /test e2e-metal-ipi
• /test e2e-metal-ipi-dualstack

Use /test all to run the following jobs that were automatically triggered:

• pull-ci-openshift-coredns-master-e2e-aws
• pull-ci-openshift-coredns-master-e2e-aws-upgrade
• pull-ci-openshift-coredns-master-e2e-gcp-serial
• pull-ci-openshift-coredns-master-e2e-metal-ipi
• pull-ci-openshift-coredns-master-images
• pull-ci-openshift-coredns-master-unit

Details

In response to this:

/retest ci/prow/e2e-metal-ipi
May be fixed now.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[gcs278]

/test ci/prow/e2e-metal-ipi
may be fixed now.

[openshift-ci]

@gcs278: The specified target(s) for /test were not found.
The following commands are available to trigger required jobs:

• /test e2e-aws
• /test e2e-aws-upgrade
• /test e2e-gcp-serial
• /test images
• /test unit

The following commands are available to trigger optional jobs:

• /test e2e-metal-ipi
• /test e2e-metal-ipi-dualstack

Use /test all to run the following jobs that were automatically triggered:

• pull-ci-openshift-coredns-master-e2e-aws
• pull-ci-openshift-coredns-master-e2e-aws-upgrade
• pull-ci-openshift-coredns-master-e2e-gcp-serial
• pull-ci-openshift-coredns-master-e2e-metal-ipi
• pull-ci-openshift-coredns-master-images
• pull-ci-openshift-coredns-master-unit

Details

In response to this:

/test ci/prow/e2e-metal-ipi
may be fixed now.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[gcs278]

/test e2e-metal-ipi
may be fixed now.

[cybertron]

/test e2e-metal-ipi

I'm seeing a lot of io timeouts in the coredns logs and it looks like the whole deployment timed out. Maybe the test node was having issues?

[brandisher]

OVNKubernetesControllerDisconnectedSouthboundDatabase
/test e2e-gcp-serial

[gcs278]

e2e-metal-ipi worked ✔️ thanks @cybertron

e2e-gcp-serial is a lost cause given the outstanding OVN and disruption bugs. I think we need an override now.

[Miciah]

There's some weird re-indentation going on here.

[gcs278]

I think you mean spaces vs. indents? Great catch, I didn't see that my edits used spaces in github diff or in my editors.
fixed

[Miciah]

In the course of reviewing the wildcard carry patch (where @gcs278 had to resolve some conflicts for this rebase), I found and reported coredns#5657. If I apply coredns#5657 on top of #76, then this test case for TestExternal fails:

	{
		Qname: "*._not-udp-or-tcp.svc1.testns.example.com.", Rcode: dns.RcodeSuccess,
	},

However, the same test fails if I check out the v1.8.7 tag (the last upstream version that supported wildcard lookups) and fix TestExternal there. I believe that this test case was always incorrect: The test case has a query for a wildcard with no match, so it should expect an error response.

Additionally, I tried the following two test cases:

	{
		Qname: "any._tcp.svc1.testns.example.com.", Rcode: dns.RcodeSuccess,
		Msg: []msg.Service{
			{Host: "1.2.3.4", Port: 80, TTL: 5, Key: "/c/org/example/testns/svc1"},
		},
	},
	{
		Qname: "_any._tcp.svc1.testns.example.com.", Rcode: dns.RcodeSuccess,
		Msg: []msg.Service{
			{Host: "1.2.3.4", Port: 80, TTL: 5, Key: "/c/org/example/testns/svc1"},
		},
	},

These test cases both pass on v1.8.7 with TestExternal fixed and on coredns#5657, so I believe there has always been a bug in the wildcard logic (that is, both "any" and "_any" are treated as wildcards), and this PR maintains bug-for-bug compatibility.

Let's get this PR merged and let it soak.

If upstream accepts coredns#5657 and someone asks for us to fix the bug in openshift/coredns, then we can treat it as a bug like any other and fix it with a carry patch for coredns#5657.

/lgtm
/approve

[openshift-ci]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: Miciah

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

The pull request process is described here

Details

Needs approval from an approver in each of these files:

• OWNERS [Miciah]

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

[Miciah]

e2e-aws-upgrade failed with the following failure:

[sig-arch] events should not repeat pathologically
{  1 events happened too frequently  event happened 27 times, something is wrong: ns/openshift-console-operator pod/console-operator-55ff985f74-zbf64 node/ip-10-0-155-208.ec2.internal - reason/ProbeError Readiness probe error: Get "https://10.130.0.35:8443/readyz": dial tcp 10.130.0.35:8443: connect: connection refused body:  }

The closest match for this failure that I found is BZ#1990419 [sig-arch] events should not repeat pathologically - event happened more than 20 times, something is wrong: ns/openshift-console pod/console-XXX node/XXX - reason/Unhealthy Readiness probe failed, which was closed due to insufficient data. Let's see if the same failure re-occurs.

/test e2e-aws-upgrade

[openshift-ci]

@gcs278: all tests passed!

Full PR test history. Your PR dashboard.

Details

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. I understand the commands that are listed here.

[openshift-ci-robot]

@gcs278: All pull requests linked via external trackers have merged:

• openshift/coredns#76

Jira Issue OCPBUGS-1731 has been moved to the MODIFIED state.

Details

In response to this:

Rebased to CoreDNS 1.10.0 so that we use the Kubernetes 1.25 packages.

I made this PR using the same process as #73, #52, #60, and #70:

1. git remote update
2. git checkout v1.10.0
3. git checkout -b merge-1.10.0-tmp
4. git checkout origin/master
5. echo 'merge coredns/coredns v1.10.0' | git commit-tree 'merge-1.10.0-tmp^{tree}' -p HEAD -p merge-1.10.0-tmp -F -
ead3b6507d7bd2fc3e67b79d8ff0dff296396bac
6. git branch merge-1.10.0 ead3b6507d7bd2fc3e67b79d8ff0dff296396bac
7. git checkout merge-1.10.0
8. git rev-list --reverse --no-merges 555dd549ee36ba7f626cf8c53707c74af318ec09..origin/master | git cherry-pick --stdin
   Note: 555dd549ee36ba7f626cf8c53707c74af318ec09 is the merge commit from the last merge Bug 2090827: Merge 1.9.2 #73
9. My merge commit from the above process was commit ead3b65.

Before the rebase, we had the following commits:

% git --no-pager log --oneline --no-merges 555dd549ee36ba7f626cf8c53707c74af318ec09..origin/master
ad24eacf6 Updating coredns images to be consistent with ART Reconciling with https://github.com/openshift/ocp-build-data/tree/2e5eba0108299087402c119ba0dc979a0a6c0bd1/images/coredns.yml
349db7a33 UPSTREAM: <carry>: openshift: Revert "Revert "add wildcard warnings (#5030)" (#5167)"
461cbbf60 UPSTREAM: <carry>: openshift: Revert "remove wildcard query functionality (#5019)"
7a41d2b87 UPSTREAM: <carry>: openshift: Disable dependabot
1d2d4f767 UPSTREAM: <carry>: openshift: Vendor required modules
3dc197d15 UPSTREAM: <carry>: openshift: .gitignore: Properly ignore coredns binary
9176cd5bf UPSTREAM: <carry>: openshift: Make coredns exclusion more specific
488cc73b1 UPSTREAM: <carry>: openshift: Add a `make test` target
b648be195 UPSTREAM: <carry>: openshift: Add product build config
1e6995352 UPSTREAM: <carry>: openshift: Add OWNERS
% 

I edited the "Vendor required modules" commit by running go mod vendor and go mod tidy and ammend commit 1d2d4f767 UPSTREAM: <carry>: openshift: Vendor required modules.

I squashed/merged the following commit into the existing "Add product build config" commit:

ad24eacf6 Updating coredns images to be consistent with ART Reconciling with https://github.com/openshift/ocp-build-data/tree/2e5eba0108299087402c119ba0dc979a0a6c0bd1/images/coredns.ym

Git merge conflicts for 461cbbf60 UPSTREAM: <carry>: openshift: Revert "remove wildcard query functionality (#5019)":
plugin/kubernetes/kubernetes.go

• Line 630: matchPortAndProtocol function comment update, but matchPortAndProtocol function has been removed so I ignored the updates (see PR Delta)

plugin/kubernetes/external.go

• Line 52: changes to variables declaration conflicted because they added endpoint as a string. I merged by taking upstream variable declarations, but added port = "*" and protocol = "*" assignment below for reversion (see PR Delta)
• Line 114 & 134: Refactoring moved calls to matchPortAndProtocol and added it in an additional location for resolving headless services (see PR delta)
  • I'm assuming we want wildcard support for headless and that wildcard support is remove via matchPortAndProtocol, but we don't have that function anyways so just using the match function.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.