Re: [I] [BUG] consumer or producer will create failed frequently when build PulsarClient with many unavailable broker nodes [pulsar]

["AuroraTwinkle (via GitHub)" <gi...@apache.org>]

AuroraTwinkle commented on issue #22934:
URL: https://github.com/apache/pulsar/issues/22934#issuecomment-2952283541

   > [@AuroraTwinkle](https://github.com/AuroraTwinkle) I see the referenced discussion [#22933](https://github.com/apache/pulsar/discussions/22933). That's a good problem description. I think it's a bug when the next available serviceUrl isn't attempted. That's the whole point of having multiple endpoints. However the solution in [#22935](https://github.com/apache/pulsar/pull/22935) has it's issues as I've commented on the PR. The general approach in the client should be that the connection is attempted and if it fails, it will pick the next endpoint.
   > 
   > I took a look at the code and it seems that `client.getCnxPool().getConnection(socketAddress)` calls should be using `client.getCnxPool().getConnection(serviceNameResolver)`.
   > 
   > `org.apache.pulsar.client.impl.ConnectionPool` would need a new `getConnection` method that accepts the `org.apache.pulsar.client.impl.ServiceNameResolver` instance. The ServiceNameResolver interface will need some changes. When there's a failure for one address, there should be a method that can be called by ConnectionPool to provide feedback that the connection failed for that address. This way the ServiceNameResolver can remove it from the main pool of addresses for a configurable period of time. If all main pool addresses are offline, the resolver implementation could start getting random addresses from the offline pool. There would also be a need to have a success feedback method so that ConnectionPool could call it each time a connection was successful. Would you like to implement such solution instead of [#22935](https://github.com/apache/pulsar/pull/22935)?
   
   @lhotari Hi, Lari :
   I have submitted a new PR to implement this solution. Would you please review it if you have time. Any suggestions are welcome.
   PIP-425: https://github.com/apache/pulsar/pull/24394
   Implemention: https://github.com/apache/pulsar/pull/24387


-- 
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: commits-unsubscribe@pulsar.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org