Question about flink message processing guarantee

[Jerry Peng]

Hello,

I have a question regarding link streaming.  I now if you enable
checkpointing you can have exactly once processing guarantee. If I do
not enable checkpointing what are the semantics of the processing? At
least once?

Best,

Jerry

[Márton Balassi]

Dear Jerry,

If you do not enable checkpointing you get the at most once processing guarantee (some might call that no guarantee at all). When you enable checkpointing you can choose between exactly and at least once semantics. The latter provides better latency.

Best,

Marton

On Tue, Dec 1, 2015 at 11:04 PM, Jerry Peng <[hidden email]> wrote:

Hello,

I have a question regarding link streaming.  I now if you enable
checkpointing you can have exactly once processing guarantee. If I do
not enable checkpointing what are the semantics of the processing? At
least once?

Best,

Jerry

[Till Rohrmann-2]

Just a small addition. Your sources have to be replayable to some extent. With replayable I mean that they can continue from some kind of offset. Otherwise the check pointing won't help you. The Kafka source supports that for example.

Cheers,
Till

On Dec 1, 2015 11:55 PM, "Márton Balassi" <[hidden email]> wrote:

Dear Jerry,

If you do not enable checkpointing you get the at most once processing guarantee (some might call that no guarantee at all). When you enable checkpointing you can choose between exactly and at least once semantics. The latter provides better latency.

Best,

Marton

On Tue, Dec 1, 2015 at 11:04 PM, Jerry Peng <[hidden email]> wrote:

Hello,

I have a question regarding link streaming.  I now if you enable
checkpointing you can have exactly once processing guarantee. If I do
not enable checkpointing what are the semantics of the processing? At
least once?

Best,

Jerry

[Stephan Ewen]

There is an overview of what guarantees what sources can give you: https://ci.apache.org/projects/flink/flink-docs-master/apis/fault_tolerance.html#fault-tolerance-guarantees-of-data-sources-and-sinks

On Wed, Dec 2, 2015 at 9:19 AM, Till Rohrmann <[hidden email]> wrote:

Just a small addition. Your sources have to be replayable to some extent. With replayable I mean that they can continue from some kind of offset. Otherwise the check pointing won't help you. The Kafka source supports that for example.

Cheers,
Till

On Dec 1, 2015 11:55 PM, "Márton Balassi" <[hidden email]> wrote:

Dear Jerry,

If you do not enable checkpointing you get the at most once processing guarantee (some might call that no guarantee at all). When you enable checkpointing you can choose between exactly and at least once semantics. The latter provides better latency.

Best,

Marton

On Tue, Dec 1, 2015 at 11:04 PM, Jerry Peng <[hidden email]> wrote:

Hello,

I have a question regarding link streaming.  I now if you enable
checkpointing you can have exactly once processing guarantee. If I do
not enable checkpointing what are the semantics of the processing? At
least once?

Best,

Jerry