yyj8 commented on PR #22133: URL: https://github.com/apache/pulsar/pull/22133#issuecomment-1972371848
> > @yyj8 Wouldn't this already be covered by the `ControlledClusterFailover` mechanism? https://pulsar.apache.org/api/client/3.2.x/org/apache/pulsar/client/api/ControlledClusterFailoverBuilder.html > > +1 > > @yyj8 Please check "PIP-121: Pulsar cluster level auto failover on client side", doc: #13315 and impl: #13316 > > Instead of adding yet another solution, could you simply use PIP-121? You should be able to provide your own custom logic to customize it. @lhotari I have read "PIP-121: Pulsar cluster level auto failover on client side", only the detection of whether the IP and port can be connected has been implemented. It is a very complex matter to determine whether a cluster can truly provide production and consumption services normally, and whether it has affected business processes. So, the goal of my PIP is to manually intervene in cluster state switching when the cluster part is unavailable and affects business processes, in order to achieve client to cluster connection switching. Because it is difficult to accurately determine whether this partially abnormal scenario has affected the business program through program automation, it is likely to cause misjudgment and ultimately affect the business program. Since initiating a request to obtain cluster health status is part of cluster probing, I placed the client's probing logic in the `probeAvailable `method of class `AutoClusterFailover.java` -- 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]
