[Bug] Unnecessary calling of "removing consumer.pendindAcks" when cursor being reading more entries

[poorbarcode]

Search before reporting

• I searched in the issues and found nothing similar.

Read release policy

• I understand that unsupported versions don't get bug fixes. I will attempt to reproduce the issue on a supported version of Pulsar client and Pulsar broker.

Issue Description

PIP-379 depends on the consumer.pendingAcks , and the PR of PIP-379 added a new behaviour that clears consumer.pendingAcks that are smaller than md-position when calling dispatcher.readMoreEntries here. Which is very strange, it should mean that the consumer.pendingAcks does not work as expected. We’d better find the root cause of the incorrect consumer.pendingAcks , and remove the unnecessary mechanism that PIP-379 added.

Are you willing to submit a PR?

• I'm willing to submit a PR!

[lhotari]

PIP-379 depends on the consumer.pendingAcks , and the PR of PIP-379 added a new behaviour that clears consumer.pendingAcks that are smaller than md-position when calling dispatcher.readMoreEntries here. Which is very strange, it should mean that the consumer.pendingAcks does not work as expected. We’d better find the root cause of the incorrect consumer.pendingAcks , and remove the unnecessary mechanism that PIP-379 added.

It seems that #24736 addresses the problem that existed for PIP-379. However, for shared subscriptions, it is allowed that a different consumer acknowledges a message that was sent out to another consumer. I haven't checked whether the pending ack in the consumer that sent the message gets cleaned up in that case. If it isn't cleaned up, it would slowly leak memory.