upgrade error from 20.1; "error ApiVersion: might be caused by broker version < 0.10" #7635
Replies: 2 comments · 5 replies
-
Some comments to your upgrade process:
Unfortunately, you do not share the actual config you have of the cluster, what is the current condition and what is running or not. SO it is hard to comment on the actual issue. I do not know librdkafka, so I'm not familiar with what it logs and how it works. The URL From the address, I guess you use Ingress, so my best guess would be that this error means that you are not reaching the broker. So the question is if the brokers are ready or not and what are they doing. |
Beta Was this translation helpful? Give feedback.
All reactions
-
Hi, In another word, my upgrade process happened as below: Installing new CRDs happened after each upgrade process. From 0.20 to 0.22, from 0.22 to 0.23 and from 0.23 to 0.31. The mentioned error happened exactly after first upgrade from 0.20 to 0.22 and exists after next upgrades as well. Regarding kafkacat or librdkafka, it worked before upgrade. anyway with other tools like docker container to consume or produce got error again while before upgrade it was fine. If there is any specific client you need to have it logs, please let me know to check and shared. Also "tls" is enabled in kafka cluster. In the following I've shared kafka-cluster config. And you are right, the address is ingress. it is reachable and mentioned port is open. There is no any error in kafka brokers log. Here you can find kafka-cluster config file. If any other config is required please let me know.
##kafka configmap
#kafka StrimziPodSet
|
Beta Was this translation helpful? Give feedback.
All reactions
-
Well, the cluster seems to work fine based on the statuses of all the resources. I do not know how would it relate to the upgrade, but my best guess is that the Ingress is not working. You should double check that it works for exaple using As I said, I do not know librdkafka -> the error message from the regular Kafka clients would be easier. But my guess is that where ever it connects, it is not the Kafka brokers because it does not understand the other side. Which would point to the Ingress. You could also try to run something from inside your Kubernetes cluster against the internal 9093 listener which would connect directly to the brokers and not through the Ingress controller. That might help to verify where the issue is. |
Beta Was this translation helpful? Give feedback.
All reactions
-
👍 1 -
❤️ 1
-
Hi
|
Beta Was this translation helpful? Give feedback.
All reactions
-
Well, I guess what matters most is if it now works for you. TBH, I also struggled a bit with the Ingress classes as the different Nginx Ingress Controller versions behave a bit differently. Some general notes to what you mentioned:
|
Beta Was this translation helpful? Give feedback.
All reactions
-
Well, it's nice that Ingress class can be configured in the listener configuration too. |
Beta Was this translation helpful? Give feedback.
All reactions
-
Do we Have any Details Doc or Video Step by stem Strimzi Upgrade from 0.20 to latest one. all this i m still confused where to start what to write. can yuo please help me with that. |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Hi,
Cluster version: 0.20.1
Deployed as Helm chart.
Situation: after upgrading to any version newer than current one we get following errors and could not connect from external client to cluster.
Error:
with kafkacat:
+connect.config
Steps for upgrade process from 0.20 to 0.22:

based on Strimzi docs:
#Download 0.22 Helm chart:
#Update values.yaml wih correct wathNamespace and DNSDomain:
#Run upgrade command:
#after this command operator, zookeeper, kafka, entityoperator are restarted with no error in logs.
#upgrade crds (step1)
#upgrade crds (step2)
#after these steps, all crds version changed to v1beta2 version.

at this steps we get error while before upgrade process it worked!
Anyway based on docs, we continue to upgrade to latest version but we got same error on version 0.31.1:
#first enable STRIMZI_FEATURE_GATES in kafka operator deployment file:
#Download 0.31.1 helm chart version:
#Update values.yaml wih correct wathNamespace and DNSDomain:
#update and add new crd(strimzipodset):
#at this step, kafka operator complains about incorrect kafka-cluster version.
#so we manually change following fields in kafka-cluster:
And again we get same error after all components restarted and up and running!
Beta Was this translation helpful? Give feedback.
All reactions