+1 It makes sense to me. I also encountered TooManyRequests in topic lookup when there’re a lot of topics.
It should be retriable instead of a simple error response to client. Thanks, Yunze > 2021年8月9日 下午10:11，Ivan Kelly <iv...@apache.org> 写道： > > Hi folks, > > I've created a PIP to do some rework on lookup timeouts and retries. > We've had major client incidents recently due to a client with many > many producers, which triggered a herding effect on broker restarts. > This PIP aims to alleviate some of the issues we saw. > > In summary, we want to retry (with backoff and jitter) on timeout and > on TooManyRequests. Kicking the error back to the client just results > in the clients restarting and trying again. > > Please take a look. > > https://github.com/apache/pulsar/wiki/PIP-91:-Separate-lookup-timeout-from-operation-timeout > > Cheers, > Ivan
