https://www.gooddata.com/ logo
Join Slack
Powered by
Can someone help me with <GoodData.CN> migration ...
# gooddata-cn
s
Can someone help me with GoodData.CN migration to AWS? We're using rancher and running into these problems with helm charts.
@Balamurali Ananthan
r
It appears you're installing pulsar chart 3.0.0 with default values, becuase it tries to create ClusterRole for Grafana. We suggest disabling optional pulsar components (like grafana). Refer to https://www.gooddata.com/developers/cloud-native/doc/2.2/deploy-and-install/cloud-native/helm-chart-installation/#use-customized-[…]syaml-for-pulsar for example values. Another issue is that gooddata-cn is tested with pulsar 2.9.x while helm chart apache/pulsar 3.0.0 installs a newer version of Apache Pulsar 2.10.2.
Note that older versions of apache/pulsar chart (<3.0.0) still use PodDisruptionBudget in policy/v1beta1 api_version and it will not work on kubernetetes 1.25+
it's possible to install this older chart on 1.25 at the expense of disabling PodDisruptionBudget by setting the following values:
Copy code
bookkeeper:
  pdb:
    usePolicy: false
zookeeper:
  pdb:
    usePolicy: false
broker:
  pdb:
    usePolicy: false
s
Thank you, will give this a try today
I've excluded grafana, etc
image.png
But getting some new errors
this is with pulsar 2.9.2... which seems like it should be the right version?
oh should I try pulsar 3.0?
Screen Shot 2023-02-24 at 12.51.59 PM.png
I tried pulsar 3.0 and got similar results
r
Hi Sheila, it seems you still have issues to set the right version of pulsar helm chart. Both previous screenshots show that you installed 3.0.0 in both attempts, not the 2.9.2 you were mentioning. As I said, 3.0.0 was not tested yet (it may work, or may not) and the overall structure of helm chart changed a lot, comparing to 2.9.x version. I don't know the way how Rancher manages its bundles, so I can't provide you with any insight. I recommend you to review bundle configuration one more time and check if the 2.9.2 (or 2.9.4) version is really installed. In case that your rancher's underlying kubernetes stack is based on k8s version 1.25 or higher, don't forget to disable pdb as mentioned above.
s
we are on 1.23
I have tried both 2.9.2 and 3.0.0 with similar results
Screen Shot 2023-02-27 at 9.20.25 AM.png
r
k8s 1.23 is OK for pulsar 2.9.2. The screenshot you sent comes from the default values.yaml that are part of apache pulsar helm chart. Did you applied customized values during installation, as specified in https://www.gooddata.com/developers/cloud-native/doc/2.2/deploy-and-install/cloud-native/helm-chart-installation/#use-customized-[…]syaml-for-pulsar ? In that case, things like grafana ClusterRole would not be installed in scope of pulsar chart deployment.
s
## Monitoring Components ## ## Control what components of the monitoring stack to deploy for the cluster monitoring: # monitoring - prometheus prometheus: false # monitoring - grafana grafana: false # monitoring - node_exporter node_exporter: false # alerting - alert-manager alert_manager: false
You can look at the YAML if you'd like
r
Well, that's a lot of data, I will check tomorrow more in detail. At the first glance, there are files that do not belong to helm chart (and are not a part of the official pulsar helm chart). Please delete your post as it contains confidental data.
s
trying a pulsar helm chart from their website
Screen Shot 2023-02-27 at 12.17.32 PM.png
I tried it with just the values in 
customized-values-pulsar.yaml
and still getting those errors
making some progress though, that's far fewer errors
r
I just deciphered your bundle file. There's a lot of confusion, I'm afraid. Your fleet.yaml looks refers to:
Copy code
helm:
  valuesFiles:
    - ./pulsar/values.yaml
This should point to 
customized-values-pulsar.yaml
instead - I didn't find this or similar file in your bundle. Also, the targetCustomization for application-uswe2-alpha cluster that points to values-alpha.yaml doesn't look good because the values-alpha.yaml file appears to contain chart default values except for monitoring components disabled.
Hi Sheila, I installed the Rancher locally and attached k3d cluster into it. I successfully installed Apache Pulsar using the configuration stored in this repo: https://gitlab.com/mouchar/fleets/-/tree/main/pulsar I discovered one issue with our custom pulsar values that needs to be adjusted on Rancher-managed clusters. Because Rancher doesn't like cluster-wide resources (like ClusterRole), the following setting needs to be added to customized-values-pulsar.yaml:
Copy code
rbac:
  limit_to_namespace: true
Now I have all resources up and running
s
So: rbac: enabled: false psp: false limit_to_namespace: true
image.png
I'm making progress but I don't know why those resources aren't showing
2023-03-01T235721,388+0000 [main-SendThread(gooddata-install-raas-install-charts-pulsar-zookeeper:2181)] WARN org.apache.zookeeper.ClientCnxn - Session 0x0 for sever gooddata-install-raas-install-charts-pulsar-zookeeper:2181, Closing socket connection. Attempting reconnect except it is a SessionExpiredException. java.lang.IllegalArgumentException: Unable to canonicalize address gooddata-install-raas-install-charts-pulsar-zookeeper:2181 because it's not resolvable at org.apache.zookeeper.SaslServerPrincipal.getServerPrincipal(SaslServerPrincipal.java:78) ~[org.apache.zookeeper-zookeeper-3.6.3.jar:3.6.3] at org.apache.zookeeper.SaslServerPrincipal.getServerPrincipal(SaslServerPrincipal.java:41) ~[org.apache.zookeeper-zookeeper-3.6.3.jar:3.6.3] at org.apache.zookeeper.ClientCnxn$SendThread.startConnect(ClientCnxn.java:1161) ~[org.apache.zookeeper-zookeeper-3.6.3.jar:3.6.3] at org.apache.zookeeper.ClientCnxn$SendThread.run(ClientCnxn.java:1210) [org.apache.zookeeper-zookeeper-3.6.3.jar:3.6.3] [9.272s][info ][safepoint ] Application time: 1.0000846 seconds
image.png
Open in Slack
PreviousNext
Powered by