oneby-wang commented on issue #24901: URL: https://github.com/apache/pulsar/issues/24901#issuecomment-3465760927
I read the logs and found something strange: broker-1 still owns DataEvents topic and has at least one active producer after broker-2 take control over this topic. Broke-2 recovery read L165820 and retentioned it due to retention policy, then broker-1 writed to L165820 and fenced it, so `No such ledger exists on Metadata Server` exception happened. The key point is to find out why broker-1 still owns DataEvents topic and still has producer writing to L165820? Could you provide the following information? 1. Client logs. Client detailed exception logs, publish logs, reconnection logs. Which broker did the client connect to(before and after the exception)? Did the client success reconnect to one broker(which) after publish failure? 2. Load balance logs about DataEvents bundle. Is there any load balance happened on pulsar cluster? Did broker-1 unload DataEvents bundle? And did broker-2 load DataEvents bundle? Is there any bundle split operation happened? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
