Pravega Operator API Resource Hook.

[co-jo]

Change log description

• Adds a manifest which defines a post-install/upgrade hook to make sure the pravegacluster resource exists.

Purpose of the change

• If one is unlucky with the timings when install the pravega-operator chart, directly followed by the pravega chart, the api-resource PravegaCluster may not exist by the time the pravega chart tries to create a PravegaCluster.

What the code does

• See above.

How to verify it

Run many install/uninstall cycles of the mentioned charts -- see that the pravega chart does not start before its resource dependency is registered.

[SrishT]

@co-jo although this is a great add-on for our charts, it isn't failing in the negative case as it should.
Like I gave an incorrect operator image to fail the deployment

$ kubectl get pods
NAME                                   READY   STATUS             RESTARTS   AGE
bookkeeper-bookie-0                    1/1     Running            0          7d12h
bookkeeper-bookie-1                    1/1     Running            0          7d12h
bookkeeper-bookie-2                    1/1     Running            0          7d12h
bookkeeper-operator-6b8f99d879-v4fh4   1/1     Running            0          7d12h
nfs-server-provisioner-1594131302-0    1/1     Running            0          8d
pravega-operator-5c5776fb84-njhd4      0/1     ImagePullBackOff   0          20m
zookeeper-0                            1/1     Running            0          7d12h
zookeeper-1                            1/1     Running            0          7d12h
zookeeper-2                            1/1     Running            0          7d12h
zookeeper-operator-7b45886df9-kkz56    1/1     Running            0          7d12h

And kubectl api-resources | grep pravegaclusters returns no response
But even then helm installation does not reflect this failure and shows that the resource is successfully deployed

$ helm install pravega-operator charts/pravega-operator/
NAME: pravega-operator
LAST DEPLOYED: Thu Jul 16 01:57:23 2020
NAMESPACE: default
STATUS: deployed
REVISION: 1
TEST SUITE: None

$ helm ls
NAME                             	NAMESPACE	REVISION	UPDATED                                	STATUS  	CHART                       	APP VERSION   
pravega-operator                 	default  	1       	2020-07-16 03:26:57.77338856 +0530 IST 	deployed	pravega-operator-0.5.1      	0.5.1      

Could u look into the negative test case for all 3 operators?

[co-jo]

@co-jo although this is a great add-on for our charts, it isn't failing in the negative case as it should.
Like I gave an incorrect operator image to fail the deployment

$ kubectl get pods
NAME                                   READY   STATUS             RESTARTS   AGE
bookkeeper-bookie-0                    1/1     Running            0          7d12h
bookkeeper-bookie-1                    1/1     Running            0          7d12h
bookkeeper-bookie-2                    1/1     Running            0          7d12h
bookkeeper-operator-6b8f99d879-v4fh4   1/1     Running            0          7d12h
nfs-server-provisioner-1594131302-0    1/1     Running            0          8d
pravega-operator-5c5776fb84-njhd4      0/1     ImagePullBackOff   0          20m
zookeeper-0                            1/1     Running            0          7d12h
zookeeper-1                            1/1     Running            0          7d12h
zookeeper-2                            1/1     Running            0          7d12h
zookeeper-operator-7b45886df9-kkz56    1/1     Running            0          7d12h

And kubectl api-resources | grep pravegaclusters returns no response
But even then helm installation does not reflect this failure and shows that the resource is successfully deployed

$ helm install pravega-operator charts/pravega-operator/
NAME: pravega-operator
LAST DEPLOYED: Thu Jul 16 01:57:23 2020
NAMESPACE: default
STATUS: deployed
REVISION: 1
TEST SUITE: None

$ helm ls
NAME                             	NAMESPACE	REVISION	UPDATED                                	STATUS  	CHART                       	APP VERSION   
pravega-operator                 	default  	1       	2020-07-16 03:26:57.77338856 +0530 IST 	deployed	pravega-operator-0.5.1      	0.5.1      

Could u look into the negative test case for all 3 operators?

Sure, thanks for the input!

Edit: I believe the result you mentioned is unrelated to my changes. I tested the situation you described using the v0.5.1-rc0 release and get the same results. I suppose I could work to fix that problem, but that would be a separate issue -- unrelated to the motivation for these 3 PRs.

colin@unix:~/tmp$ cd charts/pravega-operator && git status
HEAD detached at v0.5.1-rc0
nothing to commit, working tree clean
colin@unix:~/tmp/charts/pravega-operator$ cd ~/tmp
colin@unix:~/tmp$ helm install pravega-operator charts/pravega-operator/charts/pravega-operator \
>         --set webhookCert.crt=$(cat crt.txt)\
>         --set image.tag="1234"
NAME: pravega-operator
LAST DEPLOYED: Thu Jul 16 13:13:09 2020
NAMESPACE: default
STATUS: deployed
REVISION: 1
TEST SUITE: None
colin@unix:~/tmp$ helm list
NAME                    NAMESPACE       REVISION        UPDATED                                 STATUS          CHART                           APP VERSION
bookkeeper-cluster      default         1               2020-07-16 12:55:52.02599322 -0700 PDT  deployed        bookkeeper-0.7.0                0.7.0
bookkeeper-operator     default         1               2020-07-16 12:55:46.318653036 -0700 PDT deployed        bookkeeper-operator-0.1.2       0.1.2
nfs-client-provisioner  default         1               2020-07-16 12:53:27.885740176 -0700 PDT deployed        nfs-client-provisioner-1.2.8    3.1.0
pravega-operator        default         1               2020-07-16 13:13:09.816741198 -0700 PDT deployed        pravega-operator-0.5.1          0.5.1
zookeeper-cluster       default         1               2020-07-16 12:53:42.911246015 -0700 PDT deployed        zookeeper-0.1.0
zookeeper-operator      default         1               2020-07-16 12:53:35.437655779 -0700 PDT deployed        zookeeper-operator-0.1.0
colin@unix:~/tmp$ kubectl get pods
NAME                                      READY   STATUS         RESTARTS   AGE
bookkeeper-cluster-bookie-0               1/1     Running        0          17m
bookkeeper-cluster-bookie-1               1/1     Running        0          17m
bookkeeper-cluster-bookie-2               1/1     Running        0          17m
bookkeeper-operator-96596cdf-x9zzs        1/1     Running        0          17m
nfs-client-provisioner-795dcfb54d-csn2w   1/1     Running        0          20m
pravega-operator-6f9445b5c4-mq58f         0/1     ErrImagePull   0          20s
zookeeper-cluster-0                       1/1     Running        0          19m
zookeeper-cluster-1                       1/1     Running        0          19m
zookeeper-cluster-2                       1/1     Running        0          18m
zookeeper-operator-76bd88d64f-v7t2q       1/1     Running        0          19m

[co-jo]

Closed (and replaced with #427) due to complication with fixing the DCO check.