AloysZhang

@codelipenghui This is another problem. The root cause is the concurrent access to zookeeper. Before updating the `schemaLocator` on zk, it reads zk again. The purpose is to obtain the...

This problem is caused by the update loop. 1. Read exist schema entry 2. build new schemaEntry 3. create a new ledger 4. write schemaEnry to BK 5. builld new...

There are two places that can trigger `checkpoint`. 1. the scheduled tasks in `SyncThread.doCheckpoint` 2. the ledgerDir write-cache full and then flush The second way is the root cause of...

@codelipenghui PTAL, thanks

close this pull request due to https://github.com/apache/pulsar/pull/18701

pulsar 如果要实现分钟级别的数据对账，有什么比较好的实现方式吗