# At least once

* [Write to Kafka config](#write-to-kafka-config)
  * [In-Sync replicas](#in-sync-replicas)
  * [Flowchart](#flowchart)
* [Message loss](#message-loss)
  * [Minimize by acks config](#minimize-by-acks-config)
    * [Acks config == Qurom num](#acks-config--qurom-num)
  * [Minimize by disk flush](#minimize-by-disk-flush)
    * [Loss scenario](#loss-scenario)
    * [Config parameters](#config-parameters)

## Write to Kafka config

### In-Sync replicas

* Kafka dynamically maintains a set of in-sync replicas (ISR) that are caught-up to the leader. Only members of this set are eligible for election as leader. A write to a Kafka partition is not considered committed until all in-sync replicas have received the write.

### Flowchart

![](https://1010073591-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F-Mk8dv8Mfudl_6ziUzDf%2Fuploads%2Fgit-blob-6586980a228cd88bf947ad27757b6f3d895b3b6f%2FmessageQueue_backlog_acksConfig.png?alt=media)

## Message loss

### Minimize by acks config

* To minimize msg loss possibility, set acks=all and forbid the unclean election.

![](https://1010073591-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F-Mk8dv8Mfudl_6ziUzDf%2Fuploads%2Fgit-blob-cd80c16fbeb1566ea4334556f99f7c01e9d7bbce%2FmessageQueue_backlog_msgLoss.png?alt=media)

* If we tell the client a message is committed, and the leader fails, the new leader we elect must also have that message. This yields a tradeoff: if the leader waits for more followers to acknowledge a message before declaring it committed then there will be more potentially electable leaders.

#### Acks config == Qurom num

* If you want to avoid message loss, set acks.config == quorum number of Kafka slaves.

### Minimize by disk flush

#### Loss scenario

![](https://1010073591-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F-Mk8dv8Mfudl_6ziUzDf%2Fuploads%2Fgit-blob-5a7b6be563ec452016b9d5530bcf24508c7cd51f%2FmessageQueue_backlog_diskflush.png?alt=media)

#### Config parameters

* **By Number**: log.flush.interval.ms: Every certain time, flush page cache to disk.
* **By Time**: Two combined config:
  * log.flush.scheduler.interval.ms:
  * log.flush.interval.messages: Each time when there is this number of messages, flush page cache to disk.