Zhendong Liu
Zhendong Liu
@RongtongJin how about naming it as "controllerBrokerId", means it is an id for controller mode. Just like the dledgerSelfId, controllerBrokerId may have some relation to brokerId. when the broker is...
is there any impact on the "end to end" latency? Maybe a performance test report is needed.
@RongtongJin The truncation may also need to be done when the broker role changes.
@crazywen The original reason is that the HAClient.READ_MAX_BUFFER_SIZE muse be equal or larger than the haTransferBatchSize. Otherwise, the slave HAClient cannot read all the data to byteBufferRead as it has...
@crazywen It is not a good way to fix the BUFFER SIZE. You could introduce another config(for example haReceiveBatchSize) to control the HAClient read buffer size, which is now fixed...
@crazywen Usually, the config of each node should keep the same in the same cluster(or at least for pair of master-slave which has the same broker name). If so, just...
@lwclover I have read all the code. This solution is simple and useful in some cases. And the same time, there are two potential problems. Firstly, in the public cloud...
@lwclover This solution is enough for an inner network inside an organization. It may not be enough for common use cases. And this issue is so important that it is...
@lwclover Currently, we need to make the feature simple enough and prove it along the time. In the client and broker, we just add a general rpcHook and do not...
@lwclover It seems ok. BTW, some trivial code polishments could be done to make the code clear.