Thanks for your reply and work. Let me elaborate a bit:
1、Do you ensure that when snapshotting, all the messages produced to kafka get acked?
2、AFAIK kafka producer doesn't guarantee message order, does that impact the following processes.
1，Yes, We use synchronous write to submit logs, writing 1000 logs at a time.When we save the snapshot, we will first check whether the snapshot.logSeq has been submitted to Kafka.
2，Only one partition is used for each log topic, so the sequence can be guaranteed.
1，Yes, We use synchronous write to submit logs, writing 1000 logs at a time.When we save the snapshot, we will first check whether the snapshot.logSeq has been submitted to Kafka. 2，Only one partition is used for each log topic, so the sequence can be guaranteed.
Originally posted by @greensheng in https://github.com/gitbitex/gitbitex-spot/issues/11#issuecomment-545828540
我发现在orderBook中，seq要超过10000才回去保存一次快照，下次恢复的时候，快照很有可能和kafka对应不上，怎么办呢？