[
https://issues.apache.org/jira/browse/KAFKA-14692?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17686029#comment-17686029
]
Mickael Maison commented on KAFKA-14692:
----------------------------------------
In my test cluster the errors persist.
After the controller has printed "Completed migration of metadata from
Zookeeper to KRaft", it started printing "TRACE Sending RPCs to brokers for
metadata delta." every 500ms.
Then I updated the config from one of the brokers:
- changed broker.id to node.id
- added process.roles=broker
- removed all zookeeper configs
Restarted that broker and it spams:
{code:java}
[2023-02-08 19:06:20,410] ERROR Exception while processing request from
192.168.1.11:9092-192.168.1.11:65127-216 (kafka.network.Processor)
org.apache.kafka.common.errors.InvalidRequestException: Received request api
key LEADER_AND_ISR which is not enabled
{code}
On the controller side, there are few disconnection messages as broker 0 is
restarted
{code:java}
[2023-02-08 19:05:00,454] INFO [Controller id=1000, targetBrokerId=0] Node 0
disconnected. (org.apache.kafka.clients.NetworkClient)
[2023-02-08 19:05:00,454] WARN [Controller id=1000, targetBrokerId=0]
Connection to node 0 (mmaison-mac.home/192.168.1.11:9092) could not be
established. Broker may not be available.
(org.apache.kafka.clients.NetworkClient)
[2023-02-08 19:05:00,473] INFO [Controller id=1000, targetBrokerId=0] Client
requested connection close from node 0 (org.apache.kafka.clients.NetworkClient)
{code}
and then it starts spamming
{code:java}
[2023-02-08 19:05:09,204] INFO [Controller id=1000, targetBrokerId=0] Cancelled
in-flight LEADER_AND_ISR request with correlation id 10 due to node 0 being
disconnected (elapsed time since creation: 1535ms, elapsed time since send:
1535ms, request timeout: 30000ms) (org.apache.kafka.clients.NetworkClient)
[2023-02-08 19:05:09,241] INFO [Controller id=1000, targetBrokerId=0] Client
requested connection close from node 0 (org.apache.kafka.clients.NetworkClient)
[2023-02-08 19:05:09,381] INFO [Controller id=1000, targetBrokerId=0] Node 0
disconnected. (org.apache.kafka.clients.NetworkClient)
{code}
> Issues in Zookeeper to KRaft migration docs
> -------------------------------------------
>
> Key: KAFKA-14692
> URL: https://issues.apache.org/jira/browse/KAFKA-14692
> Project: Kafka
> Issue Type: Improvement
> Components: docs
> Reporter: Mickael Maison
> Assignee: David Arthur
> Priority: Major
>
> Following [https://kafka.apache.org/documentation/#kraft_zk_migration]
> 1) It completely skips the facts that the storage for the new quorum should
> be formatted using the existing cluster id.
> 2) In Provisioning the KRaft controller quorum:
> {{controller.quorum.voters=1@localhost:9093}} should be
> {{controller.quorum.voters=3000@localhost:9093}} as node.id=3000
> 3) When migrating brokers, it states:
> {code:java}
> # Don't set the IBP, KRaft uses "metadata.version" feature flag
> # inter.broker.protocol.version=3.4
> # Keep the migration enabled
> zookeeper.metadata.migration.enable=true
> # Remove ZooKeeper client configuration
> # zookeeper.connect=localhost:2181
> {code}
> However if I do that, my brokers fails to restart and print:
> {code:java}
> org.apache.kafka.common.config.ConfigException: If using
> zookeeper.metadata.migration.enable in KRaft mode, zookeeper.connect must
> also be set.
> {code}
> 4) When disabling zookeeper.metadata.migration.enable or keeping
> zookeeper.connect to get past this step, when brokers restart they print a
> lot of error messages:
> {code:java}
> org.apache.kafka.common.errors.InvalidRequestException: Received request api
> key LEADER_AND_ISR which is not enabled
> [2023-02-08 12:06:25,776] ERROR Exception while processing request from
> 192.168.1.11:9092-192.168.1.11:57210-107 (kafka.network.Processor)
> {code}
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
