rebase to 1.7 #15100

deads2k · 2017-07-07T19:17:57Z

Pull to rebase to 1.7.

builds
starts
test-cmd - https://ci.openshift.redhat.com/jenkins/job/test_pull_request_origin/3148/
test-integration - https://ci.openshift.redhat.com/jenkins/job/test_pull_request_origin/3148/
test-go - https://ci.openshift.redhat.com/jenkins/job/test_pull_request_origin/3155/
test-end-to-end - https://ci.openshift.redhat.com/jenkins/job/test_pull_request_origin/3148/
GCE
- test/extended/conformance.sh exited with code 0 after 01h 01m 20s
conformance - 4 failures remaining

Failing e2es

Extended.[k8s.io] Services should be able to create a functioning NodePort service - sjennings thought this wouldn't work without a cloud provider. - I've disabled for now, added to post rebase list.
Extended.[k8s.io] StatefulSet [k8s.io] Basic StatefulSet functionality [StatefulSetBasic] should adopt matching orphans and release non-matching pods - this passes, but cleanup fails, so it takes 10 minutes
Extended.[k8s.io] StatefulSet [k8s.io] Basic StatefulSet functionality [StatefulSetBasic] should not deadlock when a pod's predecessor fails - this passes, but cleanup fails, so it takes 10 minutes
Extended.[k8s.io] SchedulerPriorities [Serial] Pod should be schedule to node that don't match the PodAntiAffinity terms - I've disabled for now, added to post rebase list.

Flake

Extended.[k8s.io] SchedulerPriorities [Serial] Pod should perfer to scheduled to nodes pod can tolerate observed https://ci.openshift.redhat.com/jenkins/job/test_pull_request_origin_extended_conformance_gce/4329/
end-to-end-docker.sh is flaking a lot on Flake: github.com/openshift/origin/test/end-to-end/core.test/end-to-end/core.sh:389 #14897

deads2k · 2017-07-07T19:24:52Z

@rootfs @liggitt @sttts @derekwaynecarr @bparees @fabianofranz @juanvallejo @danwinship You all have commits with your names in them. Please review them for correctness. I will re-ping as new commits are added for particular people.

bparees · 2017-07-07T19:32:14Z

pkg/build/registry/buildconfiginstantiate/rest.go

@@ -241,7 +241,7 @@ func (h *binaryInstantiateHandler) handle(r io.Reader) (runtime.Object, error) {
 	if err != nil {
 		return nil, errors.NewInternalError(fmt.Errorf("unable to connect to node, could not retrieve TLS client config: %v", err))
 	}
-	upgrader := spdy.NewRoundTripper(tlsClientConfig)
+	upgrader := spdy.NewRoundTripper(tlsClientConfig, false)


what's the question/what does false represent?

what's the question/what does false represent?

FollowRedirects. You don't follow all of your upstream dependencies? I'm shocked! :)

all? any! :)

anyway i guess we wouldn't anticipate a redirect on this codepath, but i'm not sure there's a good reason not to follow a redirect if there is one?

anyway i guess we wouldn't anticipate a redirect on this codepath, but i'm not sure there's a good reason not to follow a redirect if there is one?

It requires peeking and guessing at what the content is. I think I'd default to off.

bparees · 2017-07-07T19:32:41Z

pkg/cmd/admin/migrate/volumesource/volumesource.go

-		if volume.VolumeSource.Metadata != nil {
-			return true
-		}
-	}


not sure why this is in the commit w/ my name?

not sure why this is in the commit w/ my name?

I wondered what happened to @enj's commit.

bparees · 2017-07-07T19:34:58Z

pkg/cmd/cli/cmd/newapp.go

@@ -502,6 +502,9 @@ func CompleteAppConfig(config *newcmd.AppConfig, f *clientcmd.Factory, c *cobra.
 	if config.ClientMapper == nil {
 		config.ClientMapper = resource.ClientMapperFunc(f.ClientForMapping)
 	}
+	if config.CategoryExpander == nil {
+		config.CategoryExpander = f.CategoryExpander()
+	}


not sure what this is but if it's needed for the resourcebuilder now, it seems reasonable enough to me.

danwinship · 2017-07-07T20:03:51Z

I think separate: DANW: router metrics changed/disabled! was supposed to be tagged @smarterclayton not me. (At least, he originally wrote the code that gets modified by that commit, and I know nothing about it.)

deads2k · 2017-07-07T20:08:49Z

I think separate: DANW: router metrics changed/disabled! was supposed to be tagged @smarterclayton not me. (At least, he originally wrote the code that gets modified by that commit, and I know nothing about it.)

Ok. I'll update for that then. I just chose you based on package name.

deads2k · 2017-07-07T20:20:30Z

Alright, I've gotten down to just #15104 and #15101 stopping progress on compiling the main process, blocking e2e, test-integration.sh, and test-cmd.sh.

[deads@deads-dev-01 origin]$ nice make 2>&1 | tee ../build.log
hack/build-go.sh  
++ Building go targets for linux/amd64: cmd/openshift cmd/oc cmd/kubefed pkg/sdn/plugin/sdn-cni-plugin vendor/github.com/containernetworking/cni/plugins/ipam/host-local vendor/github.com/containernetworking/cni/plugins/main/loopback
# github.com/openshift/origin/pkg/proxy/hybrid
pkg/proxy/hybrid/proxy.go:97: not enough arguments in call to p.unidlingProxy.OnServiceUpdate
	have ([]*"github.com/openshift/origin/vendor/k8s.io/kubernetes/pkg/api".Service)
	want (*"github.com/openshift/origin/vendor/k8s.io/kubernetes/pkg/api".Service, *"github.com/openshift/origin/vendor/k8s.io/kubernetes/pkg/api".Service)
# github.com/openshift/origin/pkg/sdn/plugin/sdn-cni-plugin
pkg/sdn/plugin/sdn-cni-plugin/openshift-sdn.go:91: invalid pointer type *types.Result for composite literal
pkg/sdn/plugin/sdn-cni-plugin/openshift-sdn.go:106: result.Print undefined (type *types.Result is pointer to interface, not interface)
# github.com/openshift/origin/pkg/proxy/unidler
pkg/proxy/unidler/unidlersocket.go:265: svcInfo.ServiceRef undefined (type *userspace.ServiceInfo has no field or method ServiceRef)
pkg/proxy/unidler/unidlersocket.go:345: svcInfo.ServiceRef undefined (type *userspace.ServiceInfo has no field or method ServiceRef)
# github.com/openshift/origin/pkg/sdn/plugin
pkg/sdn/plugin/pod_linux.go:174: result.IP4 undefined (type "github.com/openshift/origin/vendor/github.com/containernetworking/cni/pkg/types".Result has no field or method IP4)
pkg/sdn/plugin/pod_linux.go:178: cannot use result (type "github.com/openshift/origin/vendor/github.com/containernetworking/cni/pkg/types".Result) as type *"github.com/openshift/origin/vendor/github.com/containernetworking/cni/pkg/types".Result in return argument:
	*"github.com/openshift/origin/vendor/github.com/containernetworking/cni/pkg/types".Result is pointer to interface, not interface
pkg/sdn/plugin/pod_linux.go:242: ipamResult.IP4 undefined (type *"github.com/openshift/origin/vendor/github.com/containernetworking/cni/pkg/types".Result is pointer to interface, not interface)
pkg/sdn/plugin/pod_linux.go:276: contVeth.Attrs undefined (type "net".Interface has no field or method Attrs)
pkg/sdn/plugin/pod_linux.go:276: cannot assign netlink.Link to contVeth (type "net".Interface) in multiple assignment
pkg/sdn/plugin/pod_linux.go:282: ipamResult.IP4 undefined (type *"github.com/openshift/origin/vendor/github.com/containernetworking/cni/pkg/types".Result is pointer to interface, not interface)
pkg/sdn/plugin/pod_linux.go:283: cannot use ipamResult (type *"github.com/openshift/origin/vendor/github.com/containernetworking/cni/pkg/types".Result) as type *current.Result in argument to ipam.ConfigureIface
pkg/sdn/plugin/pod_linux.go:295: hostVeth.Attrs undefined (type "net".Interface has no field or method Attrs)
pkg/sdn/plugin/pod_linux.go:296: contVeth.Attrs undefined (type "net".Interface has no field or method Attrs)

danwinship · 2017-07-07T20:22:05Z

pkg/proxy/hybrid/proxy.go

-	mainServicesHandler  config.ServiceConfigHandler
+	unidlingLoadBalancer EndpointsConfigHandler
+	mainEndpointsHandler EndpointsConfigHandler
+	mainServicesHandler  ServiceConfigHandler


It looks like the upstream types were renamed to EndpointsHandler and ServiceHandler. This is going to need a bunch of work though because the interfaces have grown and HybridProxier needs to implement the new functions. I guess you already filed #15101 for this.

It looks like the upstream types were renamed to EndpointsHandler and ServiceHandler. This is going to need a bunch of work though because the interfaces have grown and HybridProxier needs to implement the new functions. I guess you already filed #15101 for this.

Yeah. Sounds like I may as well revert that commit and let a real fix come. The guts changed a lot.

The hybrid proxy is gonna suck. They split things up so you no longer get the full endpoint set, you get add/remove/change/etc events. And that makes it harder for the hybrid proxy.

liggitt · 2017-07-10T01:44:31Z

pkg/deploy/apis/apps/v1/conversion.go

-	if defaulting, found := s.DefaultingInterface(reflect.TypeOf(*in)); found {
-		defaulting.(func(*RollingDeploymentStrategyParams))(in)
-	}
+	SetDefaults_RollingDeploymentStrategyParams(in)


I didn't think any defaulting was supposed to happen in conversion now

liggitt 11 hours ago Member
I didn't think any defaulting was supposed to happen in conversion now

I'm uncertain about changing behavior. This is prexisting.

liggitt · 2017-07-10T01:49:06Z

pkg/cmd/server/api/v1/conversions.go

+	if obj.MasterClients.OpenShiftLoopbackClientConnectionOverrides.Burst <= 0 {
+		obj.MasterClients.OpenShiftLoopbackClientConnectionOverrides.Burst = 300
+	}
+	setDefault_ClientConnectionOverrides(obj.MasterClients.OpenShiftLoopbackClientConnectionOverrides)


should this function be renamed so it is found by the defaulter generation?

liggitt · 2017-07-10T01:50:27Z

pkg/auth/ldaputil/client.go

@@ -67,7 +66,7 @@ func (l *ldapClientConfig) Connect() (ldap.Client, error) {
 	// Ensure tlsConfig specifies the server we're connecting to
 	if tlsConfig != nil && !tlsConfig.InsecureSkipVerify && len(tlsConfig.ServerName) == 0 {
 		// Add to a copy of the tlsConfig to avoid mutating the original
-		c := utilnet.CloneTLSConfig(tlsConfig)
+		c := tlsConfig.Clone()


are you switching to go1.8 in this PR?

nm, I see the build update commit

sttts · 2017-07-10T09:18:45Z

pkg/authorization/registry/clusterpolicy/strategy.go

@@ -67,12 +67,12 @@ func (strategy) ValidateUpdate(ctx apirequest.Context, obj, old runtime.Object)
 }

 // GetAttrs returns labels and fields of a given object for filtering purposes
-func GetAttrs(obj runtime.Object) (labels.Set, fields.Set, error) {
+func GetAttrs(obj runtime.Object) (labels.Set, fields.Set, bool, error) {


pls document the bool.

pls document the bool.

My post-rebase list has us rethinking our lives after manually updating 50 or identical methods to match the kube ones (also done manually by clayton). Yay initializers

sttts · 2017-07-10T09:19:06Z

pkg/authorization/registry/clusterpolicybinding/strategy.go

@@ -83,12 +83,12 @@ func (strategy) ValidateUpdate(ctx apirequest.Context, obj, old runtime.Object)
 }

 // GetAttrs returns labels and fields of a given object for filtering purposes
-func GetAttrs(obj runtime.Object) (labels.Set, fields.Set, error) {
+func GetAttrs(obj runtime.Object) (labels.Set, fields.Set, bool, error) {


godoc for the bool

sttts · 2017-07-10T09:22:46Z

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

 	"github.com/golang/glog"

+	"github.com/emicklei/go-restful-swagger12"


nit: no new line above

sttts · 2017-07-10T09:40:39Z

pkg/cmd/server/start/admission.go

-	_ "github.com/openshift/origin/pkg/quota/admission/runonceduration"
-	_ "github.com/openshift/origin/pkg/scheduler/admission/podnodeconstraints"
-	_ "github.com/openshift/origin/pkg/security/admission"
-	_ "k8s.io/apiserver/pkg/admission/plugin/namespace/lifecycle"


Do we call NewAdmissionOptions to register the ns lifecycle plugin? There is also RegisterAllAdmissionPlugins in k8s.io/apiserver.

deads2k · 2017-07-10T20:00:15Z

Now with working compilation.

@danwinship @dcbw ptal at fix/followup: DANW: sdn types no longer match. I have removed the SDN to keep moving forward. Can you try to reconcile the old and new interfaces? It's more in-depth than my knowledge goes.

@DirectXMan12 ptal at separate: SROSS: DISABLE UNIDLER. To move past #15101, I've completely disabled the unidler. Same request for a commit that reconciles old and and new.

@bparees new commit for you to review: separate: BPAREES: directly depend on the scc admission plugin

bparees · 2017-07-10T20:27:41Z

pkg/cmd/server/origin/controller/build.go

+	sccAdmission := sccadmission.NewConstraint()
+	sccAdmission.SetSecurityInformers(ctx.SecurityInformers)
+	sccAdmission.SetInternalKubeClientSet(ctx.ClientBuilder.KubeInternalClientOrDie(bootstrappolicy.InfraBuildControllerServiceAccountName))
+	if err := sccAdmission.Validate(); err != nil {


looks like this resolves #14909 ?

wait nevermind.

looks like this resolves #14909 ?

No, this just makes the hack fail closer to compile time instead of a random failure at runtime.

ok back to my original statement. looks like it resolves #14909.

lgtm.

deads2k · 2017-07-10T20:31:43Z

Now with a running server! Also, don't tell @smarterclayton that I just had to bump that 10 second timeout because my all in one didn't start.....

juanvallejo · 2017-07-10T20:57:19Z

pkg/cmd/cli/cmd/set/buildhook.go

@@ -139,7 +139,7 @@ func (o *BuildHookOptions) Complete(f *clientcmd.Factory, cmd *cobra.Command, ar
 	o.Cmd = cmd

 	mapper, typer := f.Object()
-	o.Builder = resource.NewBuilder(mapper, typer, resource.ClientMapperFunc(f.ClientForMapping), kapi.Codecs.UniversalDecoder()).
+	o.Builder = resource.NewBuilder(mapper, f.CategoryExpander(), typer, resource.ClientMapperFunc(f.ClientForMapping), kapi.Codecs.UniversalDecoder()).


hm, I believe the NewBuilder func added to the factory_builder should be in kube 1.7.

Any reason why we are still using resource.NewBuilder?

Any reason why we are still using resource.NewBuilder?

It's a rebase and smaller changes are easier. Not all the methods have access to the factory.

@juanvallejo you're in the list here: #14647

you're in the list here: #14647

ack

juanvallejo · 2017-07-10T20:58:59Z

pkg/bootstrap/docker/openshift/import.go

@@ -25,7 +25,7 @@ func ImportObjects(f *clientcmd.Factory, ns, location string) error {
 		return err
 	}
 	glog.V(8).Infof("Importing data:\n%s\n", string(data))
-	r := resource.NewBuilder(mapper, typer, resource.ClientMapperFunc(f.ClientForMapping), f.Decoder(true)).
+	r := resource.NewBuilder(mapper, f.CategoryExpander(), typer, resource.ClientMapperFunc(f.ClientForMapping), f.Decoder(true)).


Same question as 7a9f25d#r126534856

danwinship · 2017-07-11T12:33:37Z

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

+		return nil, err
+	}
+	// Resolve cmd flags to add any user overrides
+	if err := cmdflags.Resolve(options.ProxyArguments, proxyOptions.AddFlags); len(err) > 0 {


Why did this get moved to the start of the function? The goal is that options.ProxyArguments is supposed to override the defaults that get set here. In particular, you're supposed to be able to add:

proxyArguments: proxy-mode: ["userspace"]

to node-config.yaml to override the default value of proxyconfig.Mode that gets set below.

Eventually we want to enter with the config, not the options. Not all the overrides being expressed will have args. I'll see if I can bring the remaining ones back into options.

danwinship · 2017-07-11T12:58:42Z

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

@@ -476,8 +476,6 @@ func (c *NodeConfig) RunProxy() {
 	endpointsConfig.RegisterEventHandler(endpointsHandler)
 	go endpointsConfig.Run(utilwait.NeverStop)

-	recorder.Eventf(c.ProxyConfig.NodeRef, kapi.EventTypeNormal, "Starting", "Starting kube-proxy.")


It doesn't look like anything would have ever modified c.ProxyConfig.NodeRef after it was created in the old code. So you could just move its creation to here and do

nodeRef := &kclientv1.ObjectReference{ Kind: "Node", Name: hostname, UID: ktypes.UID(hostname), Namespace: "", } recorder.Eventf(nodeRef, kapi.EventTypeNormal, "Starting", "Starting kube-proxy.")

(based on the current NodeRef definition in kubernetes/cmd/kube-proxy/app/server.go)

danwinship · 2017-07-11T15:11:11Z

@deads2k revert 7ab20cb fix/followup: DANW: sdn types no longer match and the pkg/sdn/plugin/proxy.go part of c61da76 separate: DANW: endpointsconfighandler changes and grab danwinship@3b64be7

openshift-bot · 2017-07-17T12:15:37Z

Evaluated for origin test up to fe8b4e7

sttts · 2017-07-17T12:18:50Z

pkg/cmd/cli/kubectl_compat_test.go

+	"plugin",
+	"resize",
+	"rollingupdate",
+	"run-container",


run-container is deprecated. why does it suddenly show up?

all of these are deprecated in fact.

post-rebase todo exists.

sttts · 2017-07-17T12:26:47Z

pkg/cmd/cli/kubectl_compat_test.go

@@ -59,7 +68,7 @@ kubectlLoop:
 			continue
 		}

-		t.Errorf("missing %q in oc", kubecmd.Name())
+		t.Errorf("missing %q,", kubecmd.Name())


nit: comma?

sttts · 2017-07-17T12:34:36Z

pkg/cmd/admin/prune/images.go

-		return fmt.Errorf("invalid --registry-url flag: %v", err)
-	}
+	// golang validation tighten and our code actually expects this to be scheme-less
+	// TODO figure out how to validate


added to post-rebase todos if we leave this in.

sttts · 2017-07-17T12:37:03Z

pkg/cmd/server/start/start_master.go

@@ -719,6 +719,13 @@ func startControllers(options configapi.MasterConfig, allocationController origi
 		return err
 	}

+	//  the service account passed for the recyclable volume plugins needs to exist.  We want to do this via the init function, but its a kube init function
+	// for the rebase, create that service account here
+	// TODO make this a lot cleaner


added to post-rebase todos. ugly, but ok for now.

sttts · 2017-07-17T12:57:43Z

pkg/cmd/util/clientcmd/factory_object_mapping.go

+	Expansions: map[string][]schema.GroupResource{
+		"all": legacyOpenshiftUserResources,
+	},
+}


Comparing with the upstream variant, lgtm

sttts · 2017-07-17T12:59:16Z

These commits look ok:

"separate: defer adding new commands until after the rebase", post-rebase todo exists
"disable registry url validation in prune", post-rebase todo exists
"squash to wiring: create SA needed for recycling", ugly, post-rebase todo exists
"squash to api server wiring", lgtm
"separate: client factory fitting", lgtm

sttts · 2017-07-17T13:24:29Z

pkg/cmd/server/kubernetes/master/controller/volumes.go

-	return true, nil
-}
-
-type PersistentVolumeAttachDetachControllerConfig struct {


this was dead code?

got replaced by the "normal" one from upstream: https://github.com/kubernetes/kubernetes/blob/master/cmd/kube-controller-manager/app/controllermanager.go#L332-L333

sttts · 2017-07-17T13:44:51Z

pkg/cmd/server/kubernetes/master/controller/volumes.go

 		persistentvolumecontroller.ControllerParameters{
 			KubeClient:                ctx.ClientBuilder.ClientOrDie("persistent-volume-binder"),
 			SyncPeriod:                ctx.Options.PVClaimBinderSyncPeriod.Duration,
-			AlphaProvisioner:          alphaProvisioner,


lgtm comparing with upstream's kubernetes/kubernetes#44090

sttts · 2017-07-17T13:45:58Z

"separate with controllers: storage controller changes". lgtm

openshift-bot · 2017-07-17T14:31:40Z

continuous-integration/openshift-jenkins/test FAILURE (https://ci.openshift.redhat.com/jenkins/job/test_pull_request_origin/3193/) (Base Commit: 22888b2) (PR Branch Commit: fe8b4e7)

deads2k · 2017-07-17T14:33:25Z

@stevekuznetsov can you help me get my e2e disablement working?

I want to disable these:

Extended.[k8s.io] StatefulSet [k8s.io] Basic StatefulSet functionality [StatefulSetBasic] should adopt matching orphans and release non-matching pods
Extended.[k8s.io] StatefulSet [k8s.io] Basic StatefulSet functionality [StatefulSetBasic] should not deadlock when a pod's predecessor fails

deads2k · 2017-07-17T15:38:16Z

"separate with controllers: storage controller changes". lgtm

Alright, I think that's fully reviewed. I've rebased and the pull against master is here: #15234

deads2k · 2017-07-17T15:49:02Z

Just the stateful set errors. fully reviewed. I'm moving #15234

update godeps.json to run bump checkers

35d81a4

deads2k mentioned this pull request Jul 7, 2017

Re-enable idler #15101

Closed

bparees reviewed Jul 7, 2017

View reviewed changes

deads2k mentioned this pull request Jul 7, 2017

sdn fails to compile on kube 1.7 #15104

Closed

danwinship reviewed Jul 7, 2017

View reviewed changes

liggitt reviewed Jul 10, 2017

View reviewed changes

sttts reviewed Jul 10, 2017

View reviewed changes

deads2k force-pushed the rebase-01-0705 branch from ec1bdb4 to d2316b0 Compare July 10, 2017 19:55

bparees reviewed Jul 10, 2017

View reviewed changes

juanvallejo reviewed Jul 10, 2017

View reviewed changes

danwinship reviewed Jul 11, 2017

View reviewed changes

juanvallejo approved these changes Jul 11, 2017

View reviewed changes

deads2k mentioned this pull request Jul 11, 2017

git url parsing unit test failure with golang-1.8 #14831

Closed

deads2k force-pushed the rebase-01-0705 branch from e6479ca to 394d28f Compare July 11, 2017 20:27

deads2k and others added 11 commits July 17, 2017 07:51

squash to admission change

4ce0724

STTTS: squash to api server wiring

b270ac5

MFOJTIK: squash to wiring: create SA needed for recycling

141f1b0

MFOJTIK: disable registry url validation in prune

d10d4c7

node default config value changes

86af1c6

Fix cross-platform compile of pod manager

fa59793

sdn: additional kube-1.7 rebase fixup

ddbc566

add FSStorageOS to SCC

1830fb9

FFRANZ: separate: defer adding new commands until after the rebase

264d870

disable projectrequest unit test because the fake client is weird

8c876d5

disable a few failing e2e tests

fe8b4e7

deads2k force-pushed the rebase-01-0705 branch from 4c36c4b to fe8b4e7 Compare July 17, 2017 12:09

sttts reviewed Jul 17, 2017

View reviewed changes

deads2k closed this Jul 17, 2017

deads2k mentioned this pull request Jul 17, 2017

rebase to kube 1.7 on master #15234

Merged

yastij mentioned this pull request Jul 18, 2017

Kubelet is not cleaning up pod volumes, leaving statefulset in pending state kubernetes/kubernetes#48716

Closed

deads2k deleted the rebase-01-0705 branch August 3, 2017 19:27

		"github.com/golang/glog"

		"github.com/emicklei/go-restful-swagger12"

rebase to 1.7 #15100

rebase to 1.7 #15100

Uh oh!

Conversation

deads2k commented Jul 7, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

deads2k commented Jul 7, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

danwinship commented Jul 7, 2017

Uh oh!

deads2k commented Jul 7, 2017

Uh oh!

deads2k commented Jul 7, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

deads2k commented Jul 10, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

deads2k commented Jul 10, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

deads2k commented Jul 7, 2017 •

edited

Loading

sttts Jul 17, 2017 •

edited

Loading