Rebase 1.7.6 #16615

soltysh · 2017-09-29T16:59:15Z

soltysh · 2017-09-29T16:59:34Z

The only problem I have right now is:

F0929 16:56:13.334553    6360 openapi.go:38] Failed to register open api spec for root: Duplicate Operation ID createNamespacedDeploymen
tConfigRollback for path /oapi/v1/namespaces/{namespace}/deploymentconfigrollbacks and /oapi/v1/namespaces/{namespace}/deploymentconfigs/{name}/rollback.

deads2k · 2017-09-29T17:05:08Z

The only problem I have right now is:

F0929 16:56:13.334553 6360 openapi.go:38] Failed to register open api spec for root: Duplicate Operation ID createNamespacedDeploymen
tConfigRollback for path /oapi/v1/namespaces/{namespace}/deploymentconfigrollbacks and /oapi/v1/namespaces/{namespace}/deploymentconfigs/{name}/rollback.

That's just because your server isn't far enough to fail on something else :)

I think we had a patch somewhere renaming those.

stevekuznetsov · 2017-09-29T17:23:22Z

/unassign

soltysh · 2017-09-29T21:59:56Z

@deads2k found the problem, this was introduced in this fix kubernetes/kubernetes@5b7783f after I removed it it's working. We'll need to figure out if the fix is wrong or we messed something up. I haven't checked how this works in 1.8, though. Let's see how tests results are now.

soltysh · 2017-09-30T17:58:40Z

@liggitt remember when you've mentioned k8s 1.7 bumped azure driver, now I see this is biting us really hard, docker distribution's azure driver relies on some methods not present anymore in the newer version :/ @smarterclayton fyi. What should we do with it?

liggitt · 2017-09-30T18:02:06Z

This is why we have to get our dockerregistry built separately, so we can track kube levels of docker, etc. Diamond dependencies are going to become impossible

soltysh · 2017-09-30T19:07:06Z

Yeah, that's the long term goal. But thankfully I realized we have a vendor directory inside our docker/distribution, which just needs the old Azure dependency and it looks like it's working...

soltysh · 2017-10-02T10:48:31Z

This is almost green, except for TestNodeAuthorizer and TestScopedProjectAccess integration tests.

soltysh · 2017-10-02T12:09:18Z

So apparently rebasing the rebase of the master helped. This is ready for reviews.

mfojtik · 2017-10-02T12:11:48Z

vendor/k8s.io/kubernetes/staging/src/k8s.io/apiserver/pkg/endpoints/openapi/openapi.go

@@ -63,6 +63,10 @@ func GetOperationIDAndTags(r *restful.Route) (string, []string, error) {
 	op := r.Operation
 	path := r.Path
 	var tags []string
+	// TODO: This is hacky, figure out where this name conflict is created and fix it at the root.
+	if strings.HasPrefix(path, "/apis/extensions/v1beta1/namespaces/{namespace}/") && strings.HasSuffix(op, "ScaleScale") {


create issue pls, this is ultraugly

Created #16635.

mfojtik · 2017-10-02T12:13:05Z

Origin changes looks reasonable to me (except #16615 (comment) maybe ;-). Good job!
LGTM

deads2k · 2017-10-02T12:40:56Z

vendor/k8s.io/kubernetes/staging/src/k8s.io/apiserver/pkg/endpoints/installer.go

@@ -589,15 +589,6 @@ func (a *APIInstaller) registerResourceHandlers(path string, storage rest.Storag

 		routes := []*restful.RouteBuilder{}

-		// If there is a subresource, kind should be the parent's kind.
-		if hasSubresource {


I don't think we should do this. This is messing with the kinds reported via discovery, right?

deads2k · 2017-10-02T12:41:29Z

UPSTREAM: docker/distribution: <carry>: Azure dependencies smells like a bad bump. YOu accidentally wrote a directory should not have perhaps?

soltysh · 2017-10-02T12:42:32Z

UPSTREAM: docker/distribution: : Azure dependencies smells like a bad bump. YOu accidentally wrote a directory should not have perhaps?

Nope, @liggitt warned me that 1.7.x bumps Azure deps, which in-turn conflict with docker distrubution's requirements. I had to move the old Azure deps to distribution's vendor dir. It's intentional.

deads2k · 2017-10-04T23:05:49Z

/retest

soltysh · 2017-10-05T07:12:40Z

/retest

soltysh · 2017-10-05T07:13:50Z

LOL, and I thought that I'm the only one hitting:

/usr/lib/golang/pkg/tool/linux_amd64/link: flushing /tmp/go-link-610563765/go.o: write /tmp/go-link-610563765/go.o: cannot allocate memory

apparently our CI suffered that one as well.

soltysh · 2017-10-05T08:32:23Z

/retest

soltysh · 2017-10-05T08:33:16Z

Failed on:

[WARNING] `go test` had the following output to stderr:
# testmain
/usr/lib/golang/pkg/tool/linux_amd64/link: running gcc failed: exit status 1
/usr/bin/ld: final link failed: No space left on device
collect2: error: ld returned 1 exit status

go build testmain: /usr/lib/golang/pkg/tool/linux_amd64/link: signal: killed
# testmain
2017/10/05 07:26:04 reading input: read /go/src/github.com/openshift/origin/_output/local/go/pkg/linux_amd64_race/github.com/openshift/origin/vendor/k8s.io/apimachinery/pkg/util/yaml.a: bad address
# testmain
/usr/lib/golang/pkg/tool/linux_amd64/link: flushing /tmp/go-link-276328245/go.o: write /tmp/go-link-276328245/go.o: cannot allocate memory
# testmain
/usr/lib/golang/pkg/tool/linux_amd64/link: flushing /tmp/go-link-585832900/go.o: write /tmp/go-link-585832900/go.o: cannot allocate memory
# testmain
/usr/lib/golang/pkg/tool/linux_amd64/link: running gcc failed: exit status 1
/usr/bin/ld: final link failed: No space left on device
collect2: error: ld returned 1 exit status

0xmichalis · 2017-10-05T08:54:44Z

/retest

soltysh · 2017-10-05T10:14:07Z

/retest

soltysh · 2017-10-05T10:14:47Z

TestScopedProjectAccess was mentioned by @smarterclayton to be more flaky. I think we'll have to dig into it more.

soltysh · 2017-10-05T10:35:26Z

/retest

deads2k · 2017-10-05T11:51:21Z

/retest

0xmichalis · 2017-10-05T12:28:13Z

@deads2k add a retest-not-required label if you want it merged as quick as possible?

deads2k · 2017-10-05T12:34:00Z

@deads2k add a retest-not-required label if you want it merged as quick as possible?

That carries a risk of breakage, but I can live with that. Do we have an easily accessible view of the branch jobs to see what happens after this merges?

0xmichalis · 2017-10-05T12:34:47Z

https://deck-ci.svc.ci.openshift.org/?type=postsubmit

These are jobs that run on every commit in master.

deads2k · 2017-10-05T12:35:11Z

https://deck-ci.svc.ci.openshift.org/?type=postsubmit

These are jobs that run on every commit in master.

thanks.

0xmichalis · 2017-10-05T12:37:51Z

You likely want origin only: https://deck-ci.svc.ci.openshift.org/?repo=openshift%2Forigin&type=postsubmit

Btw, I verified that this PR is not conflicting with the current batch in the queue. So as soon as that job finishes, we should see this merged.

deads2k · 2017-10-05T12:57:55Z

@soltysh failed again

/retest

openshift-merge-robot · 2017-10-05T15:34:16Z

Automatic merge from submit-queue.

deads2k · 2017-10-05T19:27:22Z

@soltysh @ironcladlou I need SHAs to start the apimachinery, client-go, apiserver, and metrics release-1.7.6 branches from

smarterclayton · 2017-10-05T20:33:35Z

Godeps references commit "a08f5eeb6246134f4ae5443c0593d72fd057ea7c" which doesn't exist under a branch in openshift/kubernetes, which means git fetch can't grab it.

Can someone tag this to whatever branch it should be tagged to?

smarterclayton · 2017-10-05T21:00:17Z

Another hack/restore-godeps failure:

godep: error restoring dep (github.com/docker/distribution/vendor/google.golang.org/grpc/codes): Wanted to restore rev 48294d928ced5dd9b378f7fd7c6f5da3ff3f2c89, already restored rev 1e2bbed6e09c6c8047f52af965a6e301f346d04e for another package in the repo
godep: error restoring dep (github.com/docker/distribution/vendor/google.golang.org/grpc/credentials): Wanted to restore rev 48294d928ced5dd9b378f7fd7c6f5da3ff3f2c89, already restored rev 1e2bbed6e09c6c8047f52af965a6e301f346d04e for another package in the repo
godep: error restoring dep (github.com/docker/distribution/vendor/google.golang.org/grpc/grpclog): Wanted to restore rev 48294d928ced5dd9b378f7fd7c6f5da3ff3f2c89, already restored rev 1e2bbed6e09c6c8047f52af965a6e301f346d04e for another package in the repo
godep: error restoring dep (github.com/docker/distribution/vendor/google.golang.org/grpc/internal): Wanted to restore rev 48294d928ced5dd9b378f7fd7c6f5da3ff3f2c89, already restored rev 1e2bbed6e09c6c8047f52af965a6e301f346d04e for another package in the repo
godep: error restoring dep (github.com/docker/distribution/vendor/google.golang.org/grpc/metadata): Wanted to restore rev 48294d928ced5dd9b378f7fd7c6f5da3ff3f2c89, already restored rev 1e2bbed6e09c6c8047f52af965a6e301f346d04e for another package in the repo
godep: error restoring dep (github.com/docker/distribution/vendor/google.golang.org/grpc/naming): Wanted to restore rev 48294d928ced5dd9b378f7fd7c6f5da3ff3f2c89, already restored rev 1e2bbed6e09c6c8047f52af965a6e301f346d04e for another package in the repo
godep: error restoring dep (github.com/docker/distribution/vendor/google.golang.org/grpc/transport): Wanted to restore rev 48294d928ced5dd9b378f7fd7c6f5da3ff3f2c89, already restored rev 1e2bbed6e09c6c8047f52af965a6e301f346d04e for another package in the repo
godep: error restoring dep (github.com/docker/distribution/vendor/google.golang.org/api/gensupport): Wanted to restore rev 48294d928ced5dd9b378f7fd7c6f5da3ff3f2c89, already restored rev 1e2bbed6e09c6c8047f52af965a6e301f346d04e for another package in the repo
godep: error restoring dep (github.com/docker/distribution/vendor/google.golang.org/appengine/internal/base): Wanted to restore rev 48294d928ced5dd9b378f7fd7c6f5da3ff3f2c89, already restored rev 1e2bbed6e09c6c8047f52af965a6e301f346d04e for another package in the repo
godep: error restoring dep (github.com/docker/distribution/vendor/google.golang.org/appengine/internal/datastore): Wanted to restore rev 48294d928ced5dd9b378f7fd7c6f5da3ff3f2c89, already restored rev 1e2bbed6e09c6c8047f52af965a6e301f346d04e for another package in the repo

Godeps/Godeps.json has docker/distribution in there twice at two different Revs.

@derekwaynecarr

Automatic merge from submit-queue. UPSTREAM: google/cadvisor: 1766: adaptive longOp for du operation google/cadvisor#1766 xref https://bugzilla.redhat.com/show_bug.cgi?id=1498632 @derekwaynecarr @ravisantoshgudimetla hold pending #16615

ironcladlou · 2017-10-06T13:29:25Z

So this is weird....

The referenced commit merged to release-1.7.6 from my original rebase PR in openshift/kubernetes@a08f5ee: "soltysh merged 166 commits into openshift:release-1.7.6 from ironcladlou:rebase-1.7.6"

Notice that openshift/kubernetes@a08f5ee isn't in the release-1.7.6 branch now.

BUT, here's a duplicate of the commit with a different sha in the release-1.7.6 branch openshift/kubernetes@ad0dbed.

👀

soltysh · 2017-10-06T13:39:47Z

@deads2k and I were syncing our forks, either of us must have overwritten the commits. Looking at the commits it appears it was me, not sure why though, since I was working on top of freshly pulled branch...

openshift-ci-robot added do-not-merge/work-in-progress Indicates that a PR should not merge because it is a work in progress. size/XXL Denotes a PR that changes 1000+ lines, ignoring generated files. labels Sep 29, 2017

openshift-merge-robot assigned smarterclayton and stevekuznetsov Sep 29, 2017

openshift-merge-robot added the needs-api-review label Sep 29, 2017

openshift-ci-robot unassigned stevekuznetsov Sep 29, 2017

soltysh force-pushed the rebase-1.7.6 branch from ad10827 to 74c00ed Compare September 29, 2017 21:58

soltysh force-pushed the rebase-1.7.6 branch 2 times, most recently from 8dc237c to 9868522 Compare October 2, 2017 10:44

openshift-merge-robot removed the needs-api-review label Oct 2, 2017

soltysh changed the title ~~WIP: Rebase 1.7.6~~ Rebase 1.7.6 Oct 2, 2017

openshift-ci-robot removed the do-not-merge/work-in-progress Indicates that a PR should not merge because it is a work in progress. label Oct 2, 2017

soltysh force-pushed the rebase-1.7.6 branch from 9868522 to ad05448 Compare October 2, 2017 10:54

soltysh assigned liggitt and deads2k Oct 2, 2017

mfojtik reviewed Oct 2, 2017

View reviewed changes

soltysh mentioned this pull request Oct 2, 2017

Fix double paths in openapi #16635

Closed

deads2k reviewed Oct 2, 2017

View reviewed changes

0xmichalis removed the queue/multiple-rebases label Oct 5, 2017

deads2k added the retest-not-required label Oct 5, 2017

derekwaynecarr mentioned this pull request Oct 5, 2017

UPSTREAM: 53446: kubelet: add latency metrics to network plugin manager #16696

Merged

sjenning mentioned this pull request Oct 5, 2017

UPSTREAM: google/cadvisor: 1766: adaptive longOp for du operation #16697

Merged

openshift-merge-robot merged commit 050c9ea into openshift:master Oct 5, 2017

soltysh deleted the rebase-1.7.6 branch October 5, 2017 15:56

ironcladlou mentioned this pull request Oct 5, 2017

3.7 pick list #16361

Closed

52 tasks

Rebase 1.7.6 #16615

Rebase 1.7.6 #16615

Conversation

soltysh commented Sep 29, 2017

soltysh commented Sep 29, 2017

deads2k commented Sep 29, 2017

stevekuznetsov commented Sep 29, 2017

soltysh commented Sep 29, 2017

soltysh commented Sep 30, 2017

liggitt commented Sep 30, 2017

soltysh commented Sep 30, 2017

soltysh commented Oct 2, 2017

soltysh commented Oct 2, 2017

mfojtik Oct 2, 2017

Choose a reason for hiding this comment

soltysh Oct 2, 2017

Choose a reason for hiding this comment

deads2k Oct 2, 2017

Choose a reason for hiding this comment

mfojtik commented Oct 2, 2017 • edited Loading

deads2k Oct 2, 2017

Choose a reason for hiding this comment

deads2k commented Oct 2, 2017

soltysh commented Oct 2, 2017

deads2k commented Oct 4, 2017

soltysh commented Oct 5, 2017

soltysh commented Oct 5, 2017

soltysh commented Oct 5, 2017

soltysh commented Oct 5, 2017

0xmichalis commented Oct 5, 2017

soltysh commented Oct 5, 2017

soltysh commented Oct 5, 2017

soltysh commented Oct 5, 2017

deads2k commented Oct 5, 2017

0xmichalis commented Oct 5, 2017

deads2k commented Oct 5, 2017

0xmichalis commented Oct 5, 2017

deads2k commented Oct 5, 2017

0xmichalis commented Oct 5, 2017 • edited Loading

deads2k commented Oct 5, 2017

openshift-merge-robot commented Oct 5, 2017

deads2k commented Oct 5, 2017

smarterclayton commented Oct 5, 2017 • edited Loading

smarterclayton commented Oct 5, 2017

ironcladlou commented Oct 6, 2017 • edited Loading

soltysh commented Oct 6, 2017

mfojtik commented Oct 2, 2017 •

edited

Loading

0xmichalis commented Oct 5, 2017 •

edited

Loading

smarterclayton commented Oct 5, 2017 •

edited

Loading

ironcladlou commented Oct 6, 2017 •

edited

Loading