Rebase 1.6.1 #13653

sttts · 2017-04-06T13:51:24Z

Bump kubernetes to 1.6.1 and master from Apr 4.

Replaces #13483
Fixes #13419
Fixes #13722
Fixes #13571
xref #12458

1.6 post-rebase tasks: #13732

TODOs:

TODOs for the next rebase to origin master:

Broken extended tests (run on local Fedora 25 as AWS CI is down)

https://docs.google.com/a/redhat.com/spreadsheets/d/1OXZtuzmU86jxkN2qe7YuOCFXNDPtg1oEVD-CHIZI6u0/edit?usp=sharing

liggitt · 2017-04-06T19:55:05Z

@sttts upstreams of note:
kubernetes/kubernetes#43871
kubernetes/kubernetes#44097

stevekuznetsov · 2017-04-10T14:15:32Z

hack/godep-save.sh

@@ -31,4 +31,4 @@ REQUIRED_BINS=(
  "./..."
 )

-"${GODEP}" save -t "${REQUIRED_BINS[@]}"
+GOPATH=$GOPATH:$GOPATH/src/k8s.io/kubernetes/staging "${GODEP}" save -t "${REQUIRED_BINS[@]}"


Are other tools going to need this $GOPATH addition?

Only godep. We create symlinks in vendor/ to the staging dirs. So other tools like go-build have no problems.

This symlink business might go away when we extend our tooling to handle upstream commits over multiple repos. But as that is not trivial, we try to postpone that.

stevekuznetsov · 2017-04-10T14:24:03Z

hack/godep-save.sh

+
+# godep fails to copy all package in staging because it gets confused with the symlinks.
+# Hence, we copy over manually until we have proper staging repo tooling.
+rsync -avx --include='*.go' --include='*/' --exclude='*' $GOPATH/src/k8s.io/kubernetes/staging/src/* vendor/k8s.io/kubernetes/staging/src/


Does this need a --delete in case changes later remove a file here?

We remove vendor/ before godep-save. So this is fine.

stevekuznetsov · 2017-04-10T14:25:02Z

hack/godep-save.sh

+  echo "s/$NEWREV/$OLDREV/" >> "$TMPGOPATH/undo.sed"
+}
+
+undo::forks::in::godep::json () {


We use :: to denote namespace-y things ... just use _ or - here.

Namespaces in bash... :')

Was about to write the same comment a few lines above. I totally agree with Steve here. I'd suggest the prefix to be godep::save, which would give you godep::save::undo_forks_in_godep_json here.

Let's be consistent with that and don't make our Bash guru said 😉

Done, using -.

sttts · 2017-04-11T08:53:43Z

[test]

soltysh · 2017-04-11T10:38:39Z

[test]

stevekuznetsov · 2017-04-11T12:00:20Z

hack/build-go.sh

@@ -7,6 +7,10 @@ source "$(dirname "${BASH_SOURCE}")/lib/init.sh"
 build_targets=("$@")
 platform="$(os::build::host_platform)"

+# Set build tags for these binaries
+readonly OS_GOFLAGS_TAGS="include_gcs include_oss"
+readonly OS_GOFLAGS_TAGS_$(os::build::platform_arch)="gssapi"


@smarterclayton why did these get put into build-cross only in current master?

The second line adds the build requirement that gssapi devel packages (used with CGo) are installed. Not sure we want that in general, maybe only for releases?

sttts · 2017-04-11T16:22:13Z

pkg/cmd/server/kubernetes/node/node_config_test.go

@@ -102,6 +100,7 @@ func TestKubeletDefaults(t *testing.T) {
 			RegistryPullQPS:                5.0,
 			ResolverConfig:                 kubetypes.ResolvConfDefault,
 			KubeletCgroups:                 "",
+			CgroupsPerQOS:                  true,


@derekwaynecarr @sjenning can you please review the new node config in the 1.6 rebase here?

We are seeing this on the CI AWS machines when running hack/end-to-end-docker.sh (in contrast, it's fine on a Fedora 25 machine):

I0411 12:32:33.862520 1647 kubelet.go:1757] skipping pod synchronization - [Failed to start ContainerManager failed to initialise top level QOS containers: root container /kubepods doesn't exist]

Do we miss any other settings?

@sttts What OS type/version and Docker version are the CI machines running?

# uname -a Linux ip-172-18-13-151.ec2.internal 3.10.0-327.22.2.el7.x86_64 #1 SMP Thu Jun 9 10:09:10 EDT 2016 x86_64 x86_64 x86_64 GNU/Linux

# rpm -q docker docker-1.12.6-14.el7.x86_64

# systemctl cat docker # /usr/lib/systemd/system/docker.service [Unit] Description=Docker Application Container Engine Documentation=http://docs.docker.com After=network.target Wants=docker-storage-setup.service Requires=rhel-push-plugin.socket Requires=docker-cleanup.timer [Service] Type=notify NotifyAccess=all EnvironmentFile=-/etc/sysconfig/docker EnvironmentFile=-/etc/sysconfig/docker-storage EnvironmentFile=-/etc/sysconfig/docker-network Environment=GOTRACEBACK=crash Environment=DOCKER_HTTP_HOST_COMPAT=1 Environment=PATH=/usr/libexec/docker:/usr/bin:/usr/sbin ExecStart=/usr/bin/dockerd-current \ --add-runtime docker-runc=/usr/libexec/docker/docker-runc-current \ --default-runtime=docker-runc \ --authorization-plugin=rhel-push-plugin \ --exec-opt native.cgroupdriver=systemd \ --userland-proxy-path=/usr/libexec/docker/docker-proxy-current \ $OPTIONS \ $DOCKER_STORAGE_OPTIONS \ $DOCKER_NETWORK_OPTIONS \ $ADD_REGISTRY \ $BLOCK_REGISTRY \ $INSECURE_REGISTRY ExecReload=/bin/kill -s HUP $MAINPID LimitNOFILE=1048576 LimitNPROC=1048576 LimitCORE=infinity TimeoutStartSec=10min Restart=on-abnormal MountFlags=slave [Install] WantedBy=multi-user.target

# docker info Containers: 0 Running: 0 Paused: 0 Stopped: 0 Images: 0 Server Version: 1.12.6 Storage Driver: devicemapper Pool Name: docker-docker--pool Pool Blocksize: 524.3 kB Base Device Size: 10.74 GB Backing Filesystem: xfs Data file: Metadata file: Data Space Used: 19.92 MB Data Space Total: 7.457 GB Data Space Available: 7.438 GB Metadata Space Used: 40.96 kB Metadata Space Total: 29.36 MB Metadata Space Available: 29.32 MB Thin Pool Minimum Free Space: 745.5 MB Udev Sync Supported: true Deferred Removal Enabled: true Deferred Deletion Enabled: false Deferred Deleted Device Count: 0 Library Version: 1.02.135-RHEL7 (2016-11-16) Logging Driver: journald Cgroup Driver: systemd Plugins: Volume: local Network: bridge null host overlay Authorization: rhel-push-plugin Swarm: inactive Runtimes: docker-runc runc Default Runtime: docker-runc Security Options: seccomp selinux Kernel Version: 3.10.0-327.22.2.el7.x86_64 Operating System: Red Hat Enterprise Linux Server 7.3 (Maipo) OSType: linux Architecture: x86_64 Number of Docker Hooks: 2 CPUs: 4 Total Memory: 15.26 GiB Name: ip-172-18-13-151.ec2.internal ID: 4GNM:G4KZ:FPMN:DQSK:ZWXO:WPQQ:TXXT:S432:5H3B:Z7PH:3WCO:NAUM Docker Root Dir: /var/lib/docker Debug Mode (client): false Debug Mode (server): false Registry: https://index.docker.io/v1/ WARNING: bridge-nf-call-iptables is disabled WARNING: bridge-nf-call-ip6tables is disabled Insecure Registries: ci.dev.openshift.redhat.com:5000 172.30.0.0/16 127.0.0.0/8 Registries: docker.io (secure)

stevekuznetsov · 2017-04-11T18:14:49Z

test/end-to-end/core.sh

@@ -129,7 +129,7 @@ os::cmd::expect_success 'oc env -n default dc/docker-registry REGISTRY_MIDDLEWAR
 os::cmd::expect_success 'oc rollout status dc/docker-registry'
 os::log::info "Restore configured to enable mirroring"

-registry_pod="$(oc get pod -n default -l deploymentconfig=docker-registry --show-all=false --template='{{(index .items 0).metadata.name}}')"
+registry_pod="$(oc get pod -n default -l deploymentconfig=docker-registry --template '{{range .items}}{{if not .metadata.deletionTimestamp}}{{.metadata.name}}{{end}}{{end}}')"


If there's more than one pod here it's going to smash the names together without spaces -- also, what has changed to lead to this happening?

Don't know yet. It's just taking ~10 seconds on my laptop for the old pod to go from Terminating to 100% deleted.

The rollout does not synchronize with the deletion. I think excluding pods which are in deletion state is correct.

We shouldn't have more than one active pod if our instance=1 deployment works. If not, the test breaks. We could add a space and select the first pod, but that would hide a problem.

So the behavior change here is the terminated --> deleted transition is much slower and this causes the get here to fail... agree that excluding terminated pods here is reasonable, as long as we're not plastering over a performance/behavior regression with this

deads2k · 2017-04-11T20:25:05Z

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

 	for _, fqKind := range fqKinds {
+		if unversioned && fqKind.Group == "" {


Why are you particular on the value of the group? Why isn't meta.k8s.io valid?

metav1.Status is only registered as unversioned into the core v1 group. Our build apigroup webhook storage returns Status in the New() method (

origin/pkg/util/rest/webhook.go

Line 56 in 34fd627

func (h *WebHook) New() runtime.Object {

). This code here checks on api installation time that the registered group matches to the storage being installed.

Hence, we have two choice: 1) we can register Status under the build apigroup (sttts@adc5280) or 2) make the api installer happy with unversioned kinds as well.

We chose (2) here to avoid side-effects from Status being registered under another apigroup.

Hence, we have two choice: 1) we can register Status under the build apigroup (sttts/origin@adc5280) or 2) make the api installer happy with unversioned kinds as well.

Option 1 seems more correct. I haven't fully considered the repercussions, but perhaps a followup issue is in order.

Does the serialized return include an apiVersion? Do we know what it is? meta.k8s.io or build.openshift.io seem good, but "" seems less good.

We haven't checked the returned version, but we should. Taking a look.

added a todo in the description above.

I start to think that the Status type is actually wrong, compare

origin/pkg/build/registry/buildconfig/webhook.go

Line 41 in 34fd627

func (w *WebHook) ServeHTTP(writer http.ResponseWriter, req *http.Request, ctx apirequest.Context, name, subpath string) error {

. We return Build if I read that code correctly. This also matches the integration tests:

origin/test/integration/webhook_test.go

Line 116 in 8262531

err = json.Unmarshal(body, returnedBuild)

@smarterclayton you wrote this ^^

Here is a PR: sttts#75

Merged sttts#75. It reverts this patch in APIInstaller. getResourceKind and replaces the Status type by Build in the webhook storage.

deads2k · 2017-04-11T20:26:24Z

pkg/security/scc/matcher.go

@@ -141,7 +141,7 @@ func resolveContainerSecurityContext(provider kscc.SecurityContextConstraintsPro
 	// since that is how the sc provider will eventually apply settings in the runtime.
 	// This results in an SC that is based on the Pod's PSC with the set fields from the container
 	// overriding pod level settings.
-	container.SecurityContext = sc.DetermineEffectiveSecurityContext(pod, container)
+	container.SecurityContext = sc.InternalDetermineEffectiveSecurityContext(pod, container)


I think this is worth a followup issue. Upstream is actively trying to remove this method, right?

added to #13732

Yeah I took the quick way out and remained on internal for now

We can get rid of InternalDetermineEffectiveSecurityContext using some conversion. I would leave it as it is until upstream really removes it. No need to rush here IMO.

deads2k · 2017-04-11T20:30:32Z

pkg/cmd/cli/cmd/export.go

+	var outputVersion schema.GroupVersion
+	outputVersionString := kcmdutil.GetFlagString(cmd, "output-version")
+	if len(outputVersionString) == 0 {
+		outputVersion = *clientConfig.GroupVersion


Isn't this going to be incorrect every time that you're getting resources from multiple groups. oc export builds,deployments. I really think we want to move away from choosing the --output-version like upstream and using use oc convert instead. We can make that one spot correct with a preferred version chain.

I have no suggestion that I think could make this work properly for the scenario I described.

I'm not sure what you're suggesting we do here?

deprecate output version, select the version you want via your input (oc export jobs.v1.batch,deployments.v1beta1.apps, etc)

can we make export output whatever version you requested?

soltysh · 2017-04-12T10:07:56Z

hack/godep-save.sh

+  echo "s/$NEWREV/$OLDREV/" >> "$TMPGOPATH/undo.sed"
+}
+
+undo::forks::in::godep::json () {


Was about to write the same comment a few lines above. I totally agree with Steve here. I'd suggest the prefix to be godep::save, which would give you godep::save::undo_forks_in_godep_json here.

soltysh · 2017-04-12T10:08:39Z

hack/godep-save.sh

+  echo "s/$NEWREV/$OLDREV/" >> "$TMPGOPATH/undo.sed"
+}
+
+undo::forks::in::godep::json () {


Let's be consistent with that and don't make our Bash guru said 😉

soltysh · 2017-04-12T10:19:18Z

Godeps/Godeps.json

@@ -3301,64 +3506,60 @@
 		},
 		{
 			"ImportPath": "google.golang.org/grpc",
-			"Comment": "v1.0.4",
-			"Rev": "777daa17ff9b5daef1cfdf915088a2ada3332bf0"
+			"Comment": "v1.0.2",


Why we are taking an older version of grpc?

Oh, I see there's another commit which vendors grpc inside coreos/etcd. Why we would bother with those multilevel Godeps instead of having one, which is verified. I think none of our tools that check Godeps looks into the nested one, and with the proliferation of those (I know at least one another doing this) we should be.

The generated code is not compatible. Upstream kube uses 1.0.2. etcd uses 1.0.4.

The options are vendor inside coreos/etcd or downgrade etcd to the same version kube is using.

Actually, looking at the current commitchecker implementation I think it does check all of Godeps.json.

soltysh · 2017-04-12T10:50:39Z

pkg/cmd/cli/describe/deployments.go

 		}

 		// We don't show the deployment history when running `oc rollback --dry-run`.
 		if d.config == nil && !isNotDeployed {
-			sorted := deploymentsHistory
+			var sorted []*v1.ReplicationController
+			// TODO(rebase-1.6): we should really convert the describer to use a versioned clientset


We should have another post-rebase issue wrt to that.

Added to #13732

soltysh · 2017-04-12T11:11:45Z

pkg/security/api/validation/validation_test.go

@@ -151,15 +151,15 @@ func TestValidatePodSecurityPolicyReview(t *testing.T) {
 				},
 			},
 		},
-		`spec.serviceAccountNames[0]: Invalid value: "my bad sa": must match the regex [a-z0-9]([-a-z0-9]*[a-z0-9])?(\.[a-z0-9]([-a-z0-9]*[a-z0-9])?)* (e.g. 'example.com')`: {
+		`spec.serviceAccountNames[0]: Invalid value: "my bad sa": a DNS-1123 subdomain must consist of lower case alphanumeric characters, '-' or '.', and must start and end with an alphanumeric character (e.g. 'example.com', regex used for validation is '[a-z0-9]([-a-z0-9]*[a-z0-9])?(\.[a-z0-9]([-a-z0-9]*[a-z0-9])?)*')`: {


You could squash all the changes fixing this test string into a single commit. At least fix-test: pkg/image/api/validation/validation_test.go and fix-test: pkg/api. I'd also squash fix-test: pkg/deploy/api/validation/validation_test.go error output with it as well.

Will do in the next rebase to origin master.

soltysh · 2017-04-12T11:40:10Z

pkg/cmd/server/bootstrappolicy/policy.go

@@ -155,7 +159,7 @@ func GetBootstrapClusterRoles() []authorizationapi.ClusterRole {
 					"replicationcontrollers/status", "resourcequotas", "resourcequotas/status", "securitycontextconstraints", "serviceaccounts", "services",
 					"services/status").RuleOrDie(),

-				authorizationapi.NewRule(read...).Groups(appsGroup).Resources("statefulsets", "statefulsets/status").RuleOrDie(),
+				authorizationapi.NewRule(read...).Groups(appsGroup).Resources("statefulsets", "statefulsets/status", "deployments", "deployments/scale", "deployments/status").RuleOrDie(),


You need similar changes in AdminRoleName, EditRoleName, ViewRoleName wrt do adding deployments to apps.

We should drop jobs and horizontalpodautoscalers from extensions, everywhere.

You need similar changes in AdminRoleName, EditRoleName, ViewRoleName wrt do adding deployments to apps.

Don't give /status access to the namespaced roles.

I'll fix jobs and hpas. I'm going to make cleaning up /status a post-rebase follow-up, as it's currently there for the namespaced roles.

deads2k · 2017-04-12T12:02:25Z

Commits that lgtm (I'll keep updating)

REVIEW: UPSTREAM: 00000: k8s.io/apiserver: accept unversioned core gr - with followup - Fixed.
REVIEW: use version that handles internal api objects to determine ef - with followup
REVIEW: adapt: wrap controller manager AddFlags with our controller list
REVIEW: adapt: pkg/cmd/cli/cmd/set printer changes
REVIEW: update: known kube apigroup versions in pkg/cmd/server/api
REVIEW: update: pkg/cmd/server/kubernetes/master TestAPIServerDefault
REVIEW: refactoring: merge storage factory code paths
REVIEW: adapt: convert proxy & dns to shared informers (we had to do this @now?)
REVIEW: update: cluster-reader permissions
REVIEW: fix: write versioned AdmissionConfiguration for admission plu - definitely leave this commit as its own thing.
SQUASH: REVIEW: initial 1.6 port of master setup
REVIEW: make sure we convert to v1 for List in PrintResourceInfos
REVIEW: initial 1.6 port of master setup - The storage setup concerns me. Fixed.

Commits that I don't like, but can't think of a small fix for

REVIEW: adapt: remove OutputVersion helper calls in pkg/cmd/cli/cmd - I think it fails on multiple groups

Commits with substantitive comments

REVIEW: update: pkg/cmd/server/kubernetes/master TestCMServerDefaults - get confirmation from @derekwaynecarr
REVIEW: update: etcd_storage_path_test.go, with unfinished etcd3 support - get review from @enj

deads2k · 2017-04-12T12:13:04Z

pkg/cmd/server/kubernetes/master.go

 		return groupVersionResources, nil
 	}
-	namespaceController := namespacecontroller.NewNamespaceController(kubeClient, clientPool, gvrFn, c.ControllerManager.NamespaceSyncPeriod.Duration, kapi.FinalizerKubernetes)
+	namespaceController := namespacecontroller.NewNamespaceController(kubeClient, clientPool, gvrFn, namespaceInformer, c.ControllerManager.NamespaceSyncPeriod.Duration, kapiv1.FinalizerKubernetes)


Why not c.Informers.KubernetesInformers().Core().V1().Namespaces()

Brain fart?

deads2k · 2017-04-12T12:16:12Z

pkg/cmd/server/origin/master_config.go

@@ -676,7 +681,7 @@ func newAuthenticator(config configapi.MasterConfig, restOptionsGetter restoptio
 		authenticators = append(authenticators, certauth)
 	}

-	resultingAuthenticator := union.NewFailOnError(authenticators...)
+	resultingAuthenticator := union.New(authenticators...)


This looks incorrect. We should still be NewFailOnError, right?

This is fixed in a follow-up from yesterday.

deads2k · 2017-04-12T12:16:29Z

pkg/cmd/server/origin/master_config.go

 	topLevelAuthenticators = append(topLevelAuthenticators, anonymous.NewAuthenticator())

-	return group.NewAuthenticatedGroupAdder(union.NewFailOnError(topLevelAuthenticators...)), nil
+	return group.NewAuthenticatedGroupAdder(union.New(topLevelAuthenticators...)), nil


NewFailOnError

Fixed as well: fb66c26

deads2k · 2017-04-12T12:22:24Z

pkg/cmd/server/kubernetes/master_config.go

+	etcdConfig.CertFile = masterConfig.EtcdClientInfo.ClientCert.CertFile
+	etcdConfig.CAFile = masterConfig.EtcdClientInfo.CA
+
+	storageFactory, err := kubeapiserver.NewStorageFactory(


As I recall it, this is the one that does the funny parsing of the options. I don't think that we want to respect the kube-arg for dealing with this. Seems like we should respect our config instead and use NewDefaultStorageFactory

Does this mean you want to drop everything server.APIEnablement.RuntimeConfig does?

Then we end up with this: sttts@f57955d

deads2k · 2017-04-12T12:23:31Z

pkg/cmd/server/kubernetes/master_config.go

+	if err != nil {
+		return nil, fmt.Errorf("error determining service IP ranges: %v", err)
+	}
+	if err := s.SecureServing.MaybeDefaultWithSelfSignedCerts(s.GenericServerRunOptions.AdvertiseAddress.String(), apiServerServiceIP); err != nil {


We shouldn't have to do this. By the time we're called, certs should be available.

deads2k · 2017-04-12T12:28:30Z

pkg/cmd/server/kubernetes/master_config.go

 		Informers:         informers,
 	}

 	return kmaster, nil
 }

+func ClientCARegistrationHook(options *configapi.MasterConfig) (*master.ClientCARegistrationHook, error) {


We don't get this for free?

No, hardcoded in the master. As long as we do not decompose or just plainly use master.BuildMasterConfig, we have to reimplement it: sttts@c12bd72

deads2k · 2017-04-12T12:35:54Z

pkg/cmd/server/kubernetes/master/master_config_test.go

-			ReconcilerSyncLoopPeriod:          metav1.Duration{Duration: 5 * time.Second},
+			ReconcilerSyncLoopPeriod:          metav1.Duration{Duration: 60 * time.Second},
+			Controllers:                       []string{"*"},
+			EnableTaintManager:                true,


@derekwaynecarr taints on or off?

deads2k · 2017-04-12T12:54:29Z

pkg/cmd/server/api/types.go

-	// Location is the path to a configuration file that contains the plugin's
-	// configuration
+	// Location is the path to a legacy configuration file that contains the plugin's
+	// configuration. DEPRECATED.


deprecated? Because kube now lets you split bits apart?

At least legacy in kube:

origin/vendor/k8s.io/kubernetes/staging/src/k8s.io/apiserver/pkg/admission/config.go

Line 108 in 9949be4

// convert the legacy format to the new admission control format

. Probably, here in the origin config neither "legacy" nor "DEPRECATED" makes much sense. Will remove it.

Done: sttts@4d3c856

…or in StatefulSet

sttts · 2017-04-27T13:54:29Z

[test]
[testextended][extended:all]

deads2k · 2017-04-27T14:02:22Z

[test]
[testextended][extended:all]

deads2k · 2017-04-27T14:34:32Z

[test]
[testextended][extended:all]

deads2k · 2017-04-27T14:37:44Z

[test]
[testextended][extended:all]

openshift-bot · 2017-04-27T14:37:55Z

Evaluated for origin testextended up to bd5ed9c

openshift-bot · 2017-04-27T14:38:05Z

continuous-integration/openshift-jenkins/testextended Running (https://ci.openshift.redhat.com/jenkins/job/test_pull_request_origin_extended/256/) (Base Commit: decce00) (Extended Tests: all)

openshift-bot · 2017-04-27T14:38:07Z

Evaluated for origin test up to bd5ed9c

openshift-bot · 2017-04-27T14:38:18Z

continuous-integration/openshift-jenkins/test Running (https://ci.openshift.redhat.com/jenkins/job/test_pull_request_origin/988/) (Base Commit: decce00)

deads2k · 2017-04-27T15:18:07Z

lgtm. I plan to merge on green.

soltysh · 2017-04-27T15:26:29Z

LGTM, 👍 on merge when green

0xmichalis · 2017-04-27T15:27:30Z

green on merge

…

On Thu, Apr 27, 2017 at 5:26 PM, Maciej Szulik ***@***.***> wrote: LGTM, 👍 on merge when green — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#13653 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ADuFfxAs35FgJq-dOWYCJ2c0rGt_gEwRks5r0LOpgaJpZM4M1oSL> .

deads2k · 2017-04-27T16:15:45Z

Two e2e flakes, known upstream. We should skip those for now.

One OpenShift flake, we should watch prevalence.

One bad unit test check. This needs to be fixed asap.

stevekuznetsov · 2017-04-27T16:21:49Z

Do we have issues for those three items?

sttts · 2017-04-27T16:28:11Z

@stevekuznetsov the unit test fix follows in a minute. The e2e flakes are listed in the follow-up task issue #13732.

sttts · 2017-04-27T16:37:14Z

I will collect all open comments from this PR tomorrow and add them to #13732 or create individual issues.

sttts mentioned this pull request Apr 6, 2017

WIP: Rebase 1.6.0 beta3 #13483

Closed

2 tasks

openshift-bot added the needs-rebase Indicates a PR cannot be merged because it has merge conflicts with HEAD. label Apr 6, 2017

soltysh mentioned this pull request Apr 10, 2017

Rename all controller.go to be easily distinguishable in logs #13699

Merged

stevekuznetsov reviewed Apr 10, 2017

View reviewed changes

sttts requested review from mfojtik, soltysh, liggitt, smarterclayton, pweil- and deads2k April 11, 2017 08:54

stevekuznetsov reviewed Apr 11, 2017

View reviewed changes

sttts commented Apr 11, 2017

View reviewed changes

stevekuznetsov reviewed Apr 11, 2017

View reviewed changes

deads2k reviewed Apr 11, 2017

View reviewed changes

soltysh reviewed Apr 12, 2017

View reviewed changes

sttts mentioned this pull request Apr 12, 2017

1.6 post-rebase tasks #13732

Closed

32 tasks

deads2k reviewed Apr 12, 2017

View reviewed changes

sttts added 2 commits April 27, 2017 15:52

squash: pods PATCH permission for rc controller

f4528e2

UPSTREAM: 44730: Check for terminating Pod prior to launching success…

dc20aea

…or in StatefulSet

sttts force-pushed the rebase-1.6.1 branch from 1d26af4 to a656a65 Compare April 27, 2017 13:53

Run make-update

0d6b11a

sttts force-pushed the rebase-1.6.1 branch from a656a65 to 0d6b11a Compare April 27, 2017 14:32

openshift-bot removed the needs-rebase Indicates a PR cannot be merged because it has merge conflicts with HEAD. label Apr 27, 2017

deads2k merged commit 3659383 into openshift:master Apr 27, 2017

liggitt mentioned this pull request Apr 28, 2017

Set version tags for client-go #13928

Merged

This was referenced Jul 5, 2017

Extended.[k8s.io] ConfigMap should be consumable from pods in volume [Conformance] [Volume] #14876

Closed

Downward API volume should provide podname as non-root with fsgroup [Feature:FSGroup] [Volume] #15003

Closed

		for _, fqKind := range fqKinds {
		if unversioned && fqKind.Group == "" {

Rebase 1.6.1 #13653

Rebase 1.6.1 #13653

Conversation

sttts commented Apr 6, 2017 • edited

TODOs:

TODOs for the next rebase to origin master:

Broken extended tests (run on local Fedora 25 as AWS CI is down)

liggitt commented Apr 6, 2017 • edited

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

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sttts commented Apr 11, 2017

soltysh commented Apr 11, 2017

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

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

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

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

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

deads2k commented Apr 12, 2017 • edited by sttts

Commits that lgtm (I'll keep updating)

Commits that I don't like, but can't think of a small fix for

Commits with substantitive comments

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

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

sttts commented Apr 6, 2017 •

edited

liggitt commented Apr 6, 2017 •

edited

deads2k commented Apr 12, 2017 •

edited by sttts