flink-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stephan Ewen <se...@apache.org>
Subject Re: Read Kafka topic from the beginning
Date Thu, 03 Dec 2015 22:55:37 GMT
Hi Vladimir!

The Kafka Consumer can start from any offset internally (it does that for
example when recovering a failure).

Should be fairly straightforward to set that offset field initially from a
parameter. The FlinkKafkaConsumer is part of the user jars anyways. If you
want, you can give it a try to create a modified version that accepts that
parameter, and then package that instead of the standard one.

Greetings,
Stephan


On Thu, Dec 3, 2015 at 7:07 PM, Maximilian Michels <mxm@apache.org> wrote:

> Hi Vladimir,
>
> Did you pass the properties to the FlinkKafkaConsumer?
>
> Cheers,
> Max
>
> On Thu, Dec 3, 2015 at 7:06 PM, Vladimir Stoyak <vstoyak@yahoo.com> wrote:
> > Gave it a try, but does not seem to help. Is it working for you?
> >
> > Thanks
> >
> > Sent from my iPhone
> >
> >> On Dec 3, 2015, at 6:11 PM, Vladimir Stoyak <vstoyak@yahoo.com> wrote:
> >>
> >> As far as I know "auto.offset.reset" what to do if offset it not
> available or out of bound?
> >>
> >> Vladimir
> >>
> >>
> >> On Thursday, December 3, 2015 5:58 PM, Maximilian Michels <
> mxm@apache.org> wrote:
> >> Hi Vladimir,
> >>
> >> You may supply Kafka consumer properties when you create the
> FlinkKafkaConsumer.
> >>
> >> Properties props = new Properties();
> >>
> >> // start from largest offset - DEFAULT
> >> props.setProperty("auto.offset.reset", "largest");
> >> // start from smallest offset
> >> props.setProperty("auto.offset.reset", "smallest");
> >>
> >> I don't think it is possible to start from a specific offset. The
> >> offset is only unique per partition. You could modify the offsets in
> >> the Zookeeper state but you really have to know what you're doing
> >> then.
> >>
> >> Best regards,
> >> Max
> >>
> >>
> >>
> >>> On Thu, Dec 3, 2015 at 4:01 PM, Vladimir Stoyak <vstoyak@yahoo.com>
> wrote:
> >>> I see that Flink 0.10.1 now supports Keyed Schemas which allows us to
> rely on Kafka topics set to "compact" retention for data persistence.
> >>>
> >>> In our topology we wanted to set some topics with Log Compactions
> enabled and read topic from the beginning when the topology starts or
> component recovers. Does current Kafka Consumer implementation allow to
> read all messages in a topic from the beginning or from a specific offset.
> >>>
> >>> Thanks,
> >>> Vladimir
>

Mime
View raw message