kyma-project / addons Goto Github PK

Description

TL;DR;
Use rafter-upload-service for creating AssetGroup in testing testing-addon in Kyma.

Before 1.9 Kyma release, we have a bug in release candidate branch with external brokers (aws, azure, gcp). We have wrong port pointed to rafter-upload-service inside deployment configuration of helm-broker, and provisioning of broker always failed, because data (files) required to create ClusterAssetGroup for some broker's ServiceClasses were sent to not existing port.

Reasons

For testing all functionalities of Rafter in documentation creation step.

Description
Redis addon will be not in our main index.yaml anymore. We need to move it to index-testing.yaml and refactor tests to enable redis for them.

AC

• move addon from one index to another
• list all tests that use redis addon then keep them green by activation index-testing.yaml in test suite
• find all references where redis bundle is used, especially in examples repository and mention about repoURL activation
• release new bundles version, mainly update version 0.6.0 which will be used with Kyma 1.2

Description

It would be good for a test to extend our testing bundle with documentation under the docs folder. It's explained in this comment: kyma-project/kyma#3804 (review)

Reasons

We or the guys working on the console might have a case in the future to test how documentation is displayed on the console also our testing bundle with full plan assumes it supports all features so it also should have documentation.

Description
The deployment of broker failed to start

Expected result

Actual result

Steps to reproduce

1. Create a shoot cluster in Gardener on AWS (k8s version 1.18)
2. Create a kyma from app hub. The version of kyma is 1.15.1
3. Create a namespace app. Prepare credentials for AWS Service Broker addon. Create the broker with default plan.

Troubleshooting
The broker pod in app namespace failed to start with errors below:

5da4bded6d42e067cd77e84427567" network for pod "hb-aws-se-defaul-f21fb585-60fd-47ce-ba8e-f019d720329f-aws-944cp": networkPlugin cni failed to set up pod "hb-aws-se-defaul-f21fb585-60fd-47ce-ba8e-f019d720329f-aws-944cp_app" network: error with field Metadata.Labels (value) = 'hb-aws-se-defaul-f21fb585-60fd-47ce-ba8e-f019d720329f-aws-servi-service' (must be no more than 63 characters)
  Warning  FailedCreatePodSandBox  5m45s                  kubelet, ip-10-250-15-20.eu-central-1.compute.internal  Failed to create pod sandbox: rpc error: code = Unknown desc = failed to set up sandbox container "986539df24a336a1614b6014af6942801b0dabf510065b7f718a39c07bd220db" network for pod "hb-aws-se-defaul-f21fb585-60fd-47ce-ba8e-f019d720329f-aws-944cp": networkPlugin cni failed to set up pod "hb-aws-se-defaul-f21fb585-60fd-47ce-ba8e-f019d720329f-aws-944cp_app" network: error with field Metadata.Labels (value) = 'hb-aws-se-defaul-f21fb585-60fd-47ce-ba8e-f019d720329f-aws-servi-service' (must be no more than 63 characters)
  Warning  FailedCreatePodSandBox  5m41s (x4 over 5m44s)  kubelet, ip-10-250-15-20.eu-central-1.compute.internal  (combined from similar events): Failed to create pod sandbox: rpc error: code = Unknown desc = failed to set up sandbox container "ea2054347882e75727a3cbb8abba624be1fee4f2483772dd6316302150ce7937" network for pod "hb-aws-se-defaul-f21fb585-60fd-47ce-ba8e-f019d720329f-aws-944cp": networkPlugin cni failed to set up pod "hb-aws-se-defaul-f21fb585-60fd-47ce-ba8e-f019d720329f-aws-944cp_app" network: error with field Metadata.Labels (value) = 'hb-aws-se-defaul-f21fb585-60fd-47ce-ba8e-f019d720329f-aws-servi-service' (must be no more than 63 characters)
  Normal   SandboxChanged          48s (x272 over 5m53s)  kubelet, ip-10-250-15-20.eu-central-1.compute.internal  Pod sandbox changed, it will be killed and re-created.

It seems the deployment name is too long. Exampe: hb-aws-se-defaul-f21fb585-60fd-47ce-ba8e-f019d720329f-aws-servi

FIRST start with checking all release functions, like add latest with content two times, delete release etc. Then start implementation.

AC:

• stop generating bundles content into github-pages, leave the latest version of the github-pages content cause we already placed that URL in Kyma Helm Broker (we will remove that in a three months)
• every change in master will generate latest release (never used in Kyma itself, just for purpose of testing), see:
• we will release versions, and those ones will be updated one by one in Kyma Helm Broker (manually)
  -- first version 0.1.0 and use in kyma HB
• every release version add a new release-0.X branch (will be used for hotfixes)
  --first RB release-0.1
• release is a flat structure and consist of files:
  • index.yaml (stable-versions)
  • index-showcases.yaml
  • redis-0.0.3.tgz
  • bundle-x.tgx
  • ...
• in master branch, we have just one /bundles directory with all bundles #16
• index.yaml defines a new repository and contains selected bundles, Helm Broker points to that yaml file

Description

lambda-rc-0.0.1 is the only bundle that does not have any README.md document. Add a missing README.md.

Attachments

Description

Go through all the bundles documentation and improve it so that it's clear and consistent in all repo.

Reasons

Better user experience.

Description

The /bundles/docs/getting-started.md file contained references to anchors in /kyma/blob/master/docs/helm-broker/docs/05-01-helm-broker.md which have been removed. The links that used the removed anchors should be re-evaluated and adjusted.

Anchors:

• #configuring-the-helm-broker-externally
• #configure-repository-urls-in-the-runtime

Description

When I'm trying to install the Azure Service Broker (v1.14) using Helm 3 on a Kyma cluster version (1.14) Helm times out with this error:

client.go:530: [debug] azure-service-broker-docu-checker: Jobs active: 1, jobs failed: 0, jobs succeeded: 0
Error: failed post-install: timed out waiting for the condition
helm.go:94: [debug] failed post-install: timed out waiting for the condition

The logs of the job pod azure-service-broker-docu-checker-z5h97:

│ ----                                                                                                                                                                                                                                                                                     │
│ Get list of not ready AssetGroup:                                                                                                                                                                                                                                                        │
│ Got 39 not ready Asset Group                                                                                                                                                                                                                                                             │
│ ----                                                                                                                                                                                                                                                                                     │
│ Get list of not ready AssetGroup:                                                                                                                                                                                                                                                        │
│ Got 39 not ready Asset Group
...

This is the template, adjust https://github.com/kyma-project/community/tree/master/guidelines/internal-guidelines

Thanks to it you will also get templates for issues ;)

Make sure you have proper setting in PR flow:

• for mandatory review requirement,
• ci status requirement
• cla sign up requirement

AC

• Add short document about addon-template in https://github.com/kyma-project/addons/tree/master/docs and point to our testing addon that has all functionality from addon spec used.

• do not repeat kyma
• communicate with kyma organization new repo, it must be added to triage boards to include it in the triage process

Description
We need to make the bundle checker more robust.

Issues which need to be addressed:

• Implement checker for finding ID and NAMES duplications in bundles
  • check globally in /bundles directory
  • additional - check how SC behaves when ID is diff but name is same
• HowTo download checker and targz tools in the correct versions? For now, we are using go get and installing tools from newest master instead of fixed version. Additionally, if someone already has this tools locally then the will not be upgraded.

Description
In my Macbook, by following steps in https://github.com/kyma-project/addons/blob/master/addons/aws-service-broker-0.0.1/docs/overview.md, I'm stuck with step 6. It reports:

An error occurred (NoSuchEntity) when calling the CreateAccessKey operation: The user with name XXX cannot be found.

Expected result
Credential can be created in step 6.

Actual result

Steps to reproduce

Troubleshooting
USERNAME is reserved in MacOS.

Description

I am trying to provision AWS Service Broker on local installation of Kyma 1.3.0. It provisions successfully (I can see the Service Instance has status Running) but there are no services visible. In the logs I can see following messages:
kubectl -n aaa logs hb-aws-service-broker-default-e79702b8-a6da-11e9-9054-aws-wp42w awssb

E0715 09:31:55.381119       1 util.go:490] ResourceNotFoundException: Requested resource not found
	status code: 400, request id: S1IF8H1EF1TJ57AFBR3US8TL4VVV4KQNSO5AEMVJF66Q9ASUAAJG
I0715 09:41:48.981732       1 awsbroker.go:174] Listing objects bucket: awsservicebroker region: us-east-1 prefix: templates/latest
I0715 09:41:49.551549       1 awsbroker.go:193] Found 18 objects
I0715 09:41:49.551603       1 awsbroker.go:168] Updating listings cache with [{/athena false} {/auroramysql false} {/aurorapostgresql false} {/documentdb false} {/dynamodb false} {/elasticache false} {/elasticsearch false} {/emr false} {/kinesis false} {/kms false} {/lex false} {/polly false} {/rdsmariadb false} {/rdsmssql false} {/rdsmysql false} {/rdsoracle false} {/rdspostgresql false} {/redshift false} {/rekognition false} {/route53 false} {/s3 false} {/sns false} {/sqs false} {/translate false}]
E0715 09:41:49.551624       1 awsbroker.go:139] not found
E0715 09:41:49.551628       1 awsbroker.go:139] not found
E0715 09:41:49.551631       1 awsbroker.go:139] not found
E0715 09:41:49.551633       1 awsbroker.go:139] not found

Description

Provide a new GCP Service Broker installation at runtime. Old one was deprecated by google.

Reasons

Allow Kyma users to integrate GCP Service Broker with Kyma in an easy way with well documented steps described in a documentation of the service class.

Acceptance Criteria

• make an addon for GCP Service Broker
  • keep all naming conventions, like description etc synced with already existing Azure and AWS addons
  • check if addon requires some input parameters (for example Azure Broker has such one)
  • addon itself implementation and doc
• create service class documentation for above addon
  • add documentation per every GCP class, use reference to GCP documentation if available
• make sure all input parameters are displayed correctly in Service Catalog UI
• release addons repository and use new version in Kyma
• keep addon GCP image/chart version and docs with the same version
• change addon id (name must stay). I mean do not use previous one.
• install broker in kyma, delete and install should nicely work
• if seceret not provided, should fail asap

Attachments

• Documentation, see: https://github.com/GoogleCloudPlatform/gcp-service-broker
• Installation on Kubernetes, see: https://github.com/GoogleCloudPlatform/gcp-service-broker/tree/develop/deployments/helm/gcp-service-broker

Now you are temporarily using TravisCI because it was incubating project, you need to switch to the same CI rest of Kyma is using.

It doesn't make sense to move not to old CI as we are now replacing it with Prow https://github.com/kyma-project/community/tree/master/sig-and-wg/wg-prow

Description
The gcp-service-broker with the "default" plan is using a bundled mysql. The sub-chart provided as zip file is based on the outdated Deployment apiGroup and with that a deployment fails on k8s 1.16.

Please update the chart to the latest apiGroup.
I think inflating the zip file in the source code will also be beneficial as a search for outdated terms will have matches.

Expected result

Actual result

Steps to reproduce
Instantiate the broker with the "default" plan on kyma.10 with k8s 1.16 underneath

Troubleshooting

The checker installation fails:

Installing bundle checker...
# github.com/coreos/etcd/clientv3
../../coreos/etcd/clientv3/auth.go:116:72: cannot use auth.callOpts (type []"github.com/coreos/etcd/vendor/google.golang.org/grpc".CallOption) as type []"go.etcd.io/etcd/vendor/google.golang.org/grpc".CallOption in argument to auth.remote.AuthEnable
../../coreos/etcd/clientv3/auth.go:121:74: cannot use auth.callOpts (type []"github.com/coreos/etcd/vendor/google.golang.org/grpc".CallOption) as type []"go.etcd.io/etcd/vendor/google.golang.org/grpc".CallOption in argument to auth.remote.AuthDisable
../../coreos/etcd/clientv3/auth.go:126:100: cannot use auth.callOpts (type []"github.com/coreos/etcd/vendor/google.golang.org/grpc".CallOption) as type []"go.etcd.io/etcd/vendor/google.golang.org/grpc".CallOption in argument to auth.remote.UserAdd
../../coreos/etcd/clientv3/auth.go:131:86: cannot use auth.callOpts (type []"github.com/coreos/etcd/vendor/google.golang.org/grpc".CallOption) as type []"go.etcd.io/etcd/vendor/google.golang.org/grpc".CallOption in argument to auth.remote.UserDelete
../../coreos/etcd/clientv3/auth.go:136:122: cannot use auth.callOpts (type []"github.com/coreos/etcd/vendor/google.golang.org/grpc".CallOption) as type []"go.etcd.io/etcd/vendor/google.golang.org/grpc".CallOption in argument to auth.remote.UserChangePassword
../../coreos/etcd/clientv3/auth.go:141:104: cannot use auth.callOpts (type []"github.com/coreos/etcd/vendor/google.golang.org/grpc".CallOption) as type []"go.etcd.io/etcd/vendor/google.golang.org/grpc".CallOption in argument to auth.remote.UserGrantRole
../../coreos/etcd/clientv3/auth.go:146:80: cannot use auth.callOpts (type []"github.com/coreos/etcd/vendor/google.golang.org/grpc".CallOption) as type []"go.etcd.io/etcd/vendor/google.golang.org/grpc".CallOption in argument to auth.remote.UserGet
../../coreos/etcd/clientv3/auth.go:151:72: cannot use auth.callOpts (type []"github.com/coreos/etcd/vendor/google.golang.org/grpc".CallOption) as type []"go.etcd.io/etcd/vendor/google.golang.org/grpc".CallOption in argument to auth.remote.UserList
../../coreos/etcd/clientv3/auth.go:156:106: cannot use auth.callOpts (type []"github.com/coreos/etcd/vendor/google.golang.org/grpc".CallOption) as type []"go.etcd.io/etcd/vendor/google.golang.org/grpc".CallOption in argument to auth.remote.UserRevokeRole
../../coreos/etcd/clientv3/auth.go:161:80: cannot use auth.callOpts (type []"github.com/coreos/etcd/vendor/google.golang.org/grpc".CallOption) as type []"go.etcd.io/etcd/vendor/google.golang.org/grpc".CallOption in argument to auth.remote.RoleAdd
../../coreos/etcd/clientv3/auth.go:161:80: too many errors
Cannot install checker```

Description

Provide AWS Service Broker installation at runtime.

Reasons

Allow Kyma users to integrate AWS Service Broker with Kyma in an easy way with well documented steps described in a documentation of the service class.

Acceptance Criteria

• Make a bundle for AWS Service Broker
  • keep all naming conventions, like description etc synced with already existing Azure and GCP bundles
  • check if bundle requires some input parameters (for example Azure Broker has such one)
  • bundle itself implementation and doc
• Create service class documentation for above bundle
  • add documentation per every AWS class, use reference to AWS documentation
• Make sure all input parameters are displayed correctly in Service Catalog UI
• release bundles repository and use new version in Kyma
• Keep bundle AWS image/chart version and docs with the same version

Attachments

• Documentation, see: https://github.com/awslabs/aws-servicebroker/tree/master/docs
• Installation on Kubernetes, see: https://github.com/awslabs/aws-servicebroker/blob/master/docs/getting-started-k8s.md

It must have all the needed labels to properly manage the project https://github.com/kyma-project/community/labels

Description

I create a fresh kyma cluster from scratch, based on version 0.6.1.

I opened the service catalog in a pre-defined namespace.

Expected result

The local redis service is shown. The redis icon is shown right next to the service.

Actual result

The icons is missing.

The image is embedded from an external URL, which seems to return 200, but does not return any svg data.

Steps to reproduce

• Setup a fresh kyma cluster, 0.6.1 based on minikube
• Access the service catalog

Alternative:

• Try to access https://azure.microsoft.com/svghandler/redis-cache/

Troubleshooting

I searched the repo for the non-working URL. It's probably taken from:

https://github.com/kyma-project/bundles/blob/latest/bundles/redis-0.0.3/meta.yaml#12