dhinesherode91 commented on issue #8429: URL: https://github.com/apache/pulsar/issues/8429#issuecomment-955159405
@congbobo184 The issue is not yet resolved. We are using Pulsar 2.7.1. Both broker and Pulsar client are using the same version. Our consumer is subscribed to a single topic only. Ours is an AVRO schema based topic. On debugging the code, I could arrive at one point, Once the consumer receives the first message it checks the schema version of the message and asks the broker for the schema details. At that time when the broker is serving heavy loads and is unable to process the request, then the client cannot get the schema and it can throw the above exception. Is it possible ? Also I would like to know what is the proper configuration for Pulsar production load ? Currently our Pulsar will receive 10K messages per minute and will grow to 100K in near future. We have deployed our Pulsar in AWS Kubernetes (m5.xlarge instances nodegroup) with, 5 brokers (each 1CPU, 3Gi Memory), 5 bookies (each 1CPU, 4Gi Memory with separate ledger and journal mounted as EBS), 5 zookeeper (each 500m CPU, 2Gi Memory) , 5 proxy instances (each 500m CPU, 2Gi Memory). With this above configuration, we are facing connection exception occasionally. Kindly advise. -- 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]
