The Unidirectional Topic Operator (enabled by default in 0.39) seems to cause disruption when upgrading from 0.38 (with BTO enabled by default).
It seems to be reproducible with the following steps (4 times out of 4 tries):
Deploy Strimzi 0.38 and enable KafkaNodePools
Deploy the Kafka cluster with 3 Zoos, 2 node pol with 2 brokers each, CC enabled (I used this YAML -> this is my usual YAML, nothing was modified specifically for this test so not sure if any option there is causing this)
Once the Kafka cluster is running, deploy test clients:
Deploy Kafka Connect (I used this YAML -> again this is my usual YAML, nothing was modified specifically for this test so not sure if any option there is causing this; If you want to reuse it, you have to update the docker image and the push secret and use yours)
Once it is all running, upgrade to 0.39 and observe the upgrade
Expected behavior:
Everything upgrades fine, no custom resource transitions to any errors
Actual behavior:
Some KafkaTopic resources transition to error until they recover again few minutes later.
It is not clear if this has some impact on the topics in Kafka or just on the KafkaTopic resources. In one case, I saw some issues with a topic in the Connect cluster. But other times I reproduced it it seemed fine. SO not sure if that was a coincidence or not.
This is possibly a blocker for 0.39.0 release! At least unless we confirm that it is only a cosmetic issue in UTO without any negative impact. In that case it would be ugly, but perhaps not blocking.
The Unidirectional Topic Operator (enabled by default in 0.39) seems to cause disruption when upgrading from 0.38 (with BTO enabled by default).
It seems to be reproducible with the following steps (4 times out of 4 tries):
Expected behavior:
Actual behavior:
When observing the topics, I see this: