Cross-site fails with 3 or more sites #2152
Conversation
test/e2e/xsite/xsite_test.go
Outdated
| version := kube.GetStatefulSet(infinispan.Name, infinispan.Namespace).ObjectMeta.ResourceVersion | ||
|
|
||
| time.Sleep(10 * time.Second) | ||
|
|
||
| kube.WaitForInfinispanPods(3, tutils.SinglePodTimeout, infinispan.Name, namespace) | ||
| kube.WaitForInfinispanCondition(infinispan.Name, namespace, ispnv1.ConditionWellFormed) | ||
|
|
||
| otherVersion := kube.GetStatefulSet(infinispan.Name, infinispan.Namespace).ObjectMeta.ResourceVersion | ||
| assert.Equal(t, version, otherVersion) |
There was a problem hiding this comment.
@ryanemerson is there a better way to test it?
There was a problem hiding this comment.
I think you should check the StatefulSet ObservedGeneration is 1 as the StatefulSet should never be modified once the initial cluster has been created. Take a look at spec_update_test for an idea of how to retrieve and check this.
There was a problem hiding this comment.
I'll try the generation... it is failing randomly locally 😢
test/e2e/xsite/xsite_test.go
Outdated
|
|
||
| defer kube.CleanNamespaceAndLogOnPanic(t, namespace) | ||
|
|
||
| infinispan := &ispnv1.Infinispan{ |
There was a problem hiding this comment.
We can reduce the verbosity a fair bit here by using tutils.DefaultSpec(t, testKube, func(i *ispnv1.Infinispan) { to define the CR.
There was a problem hiding this comment.
I had to set RUN_LOCAL_OPERATOR=false locally to make it work with that spec. Let's see if I need to update the GHA too.
There was a problem hiding this comment.
Added to the GHA too. It was failing without it.
There was a problem hiding this comment.
I think that's because with the DefaultSpec function and RUN_LOCAL_OPERATOR we're apply the webhook defaults. I'm surprised setting !RUN_LOCAL_OPERATOR works though, as that should prevent the Operator from being started so I wouldn't expect any reconciliation in the test 🤔
There was a problem hiding this comment.
Am I doing something wrong? The cross-site tests deploy the operator in the kind clusters:
There was a problem hiding this comment.
Nope, you're doing exactly the right thing ... I have just forgot that detail 😄
test/e2e/xsite/xsite_test.go
Outdated
| version := kube.GetStatefulSet(infinispan.Name, infinispan.Namespace).ObjectMeta.ResourceVersion | ||
|
|
||
| time.Sleep(10 * time.Second) | ||
|
|
||
| kube.WaitForInfinispanPods(3, tutils.SinglePodTimeout, infinispan.Name, namespace) | ||
| kube.WaitForInfinispanCondition(infinispan.Name, namespace, ispnv1.ConditionWellFormed) | ||
|
|
||
| otherVersion := kube.GetStatefulSet(infinispan.Name, infinispan.Namespace).ObjectMeta.ResourceVersion | ||
| assert.Equal(t, version, otherVersion) |
There was a problem hiding this comment.
I think you should check the StatefulSet ObservedGeneration is 1 as the StatefulSet should never be modified once the initial cluster has been created. Take a look at spec_update_test for an idea of how to retrieve and check this.
test/e2e/xsite/xsite_test.go
Outdated
| @@ -467,6 +468,87 @@ func TestSuspectAndHearbeatConfig(t *testing.T) { | |||
| expectHeartBeatConfiguration(t, testKubes["xsite2"], true, hbInterval, hbTimeout) | |||
| } | |||
|
|
|||
| func TestMultipleSites(t *testing.T) { | |||
There was a problem hiding this comment.
As we're not actually waiting for the xsite view to form, maybe we should link the issue as a comment and update the name to TestMultipleSiteDefinitions
Sort by site name when creating the configuration XML file to ensure a stable hashing Closes infinispan#1834
|
Thanks @pruivo |
Sort by site name when creating the configuration XML file to ensure a stable hashing
Closes #1834