I understand it now, even if we are able to attach custom partitioning. The
data shall still travel from stream nodes to broker on join topic, so
travel to network will still be there.
-Sameer.
On Tue, Dec 19, 2017 at 1:17 AM, Matthias J. Sax
wrote:
> > need to map the keys, modify them
> >> an
For the server log, is it possible to enable DEBUG logging ?
Thanks
On Mon, Dec 18, 2017 at 4:35 PM, HKT wrote:
> Thanks for reply.
>
> here is the client side log:
>
> 2017-12-19 08:26:08 [main] DEBUG o.a.k.c.p.i.TransactionManager -
> [Producer clientId=producer-1, transactionalId=hello] Tran
The Jira priority just increase by your report!
Of course, we are always happy about pull request :D
-Matthias
On 12/18/17 1:27 PM, Artur Mrozowski wrote:
> Yes, sounds like it. We run into problems at exactly same spot using BEAM
> as well, although in that case it resulted in data loss.
>
>
Thanks a lot!
On 12/18/17 12:46 PM, Dmitry Minkovsky wrote:
> You're welcome. Another one I found today
> https://docs.confluent.io/current/streams/developer-guide/dsl-api.html
>
>> groupedStream.windowedBy(TimeUnit.MINUTES.toMillis(5))
>
> should be
>
>> groupedStream.windowedBy(TimeWindows.of
Hi Maciek,
I expect that 1.0.1 will be released some time in January.
Ismael
On Mon, Dec 18, 2017 at 10:42 AM, Maciek Próchniak wrote:
> Hello,
>
> are there plans to release version 1.0.1?
>
> We are affected by https://issues.apache.org/jira/browse/KAFKA-6185 and
> cannot upgrade at the mome
Yes, sounds like it. We run into problems at exactly same spot using BEAM
as well, although in that case it resulted in data loss.
Thank you Matthias. Doesn't sound like it's going to be resolved any time
soon, does it?
/Artur
On Mon, Dec 18, 2017 at 8:11 PM, Matthias J. Sax
wrote:
> I think y
You're welcome. Another one I found today
https://docs.confluent.io/current/streams/developer-guide/dsl-api.html
> groupedStream.windowedBy(TimeUnit.MINUTES.toMillis(5))
should be
> groupedStream.windowedBy(TimeWindows.of(TimeUnit.MINUTES.toMillis(5)))
in two spots.
On Mon, Dec 18, 2017 at 2:2
> need to map the keys, modify them
>> and then do a join.
This will always trigger a rebalance. There is no API atm to tell KS
that partitioning is preserved.
Custom partitioner won't help for your case as far as I understand it.
-Matthias
On 12/17/17 9:48 PM, Sameer Kumar wrote:
> Actually,
Thanks for reporting this!
We will fix it.
-Matthias
On 12/17/17 7:05 PM, Philippe Derome wrote:
> I agree with Dmitry's first comment, it really looks like the paragraph he
> points to under "Table" was pasted without edit from the one previously
> that pertained to "KStream".
>
> On Sun, Dec
I think you are hitting: https://issues.apache.org/jira/browse/KAFKA-4609
-Matthias
On 12/18/17 1:52 AM, Artur Mrozowski wrote:
> Hi Bill,
>
> I am actually referring to duplicates as completely identical records. I
> can observe it when I convert result of left join between KTables to
> strea
It's better to use single producer.
closing() and recreating only result in unnecessary overhead and does
not give you any advantages.
-Matthias
On 12/18/17 3:12 AM, Shubham Dhoble wrote:
> I am accessing kafka through java code. The use case is, I want to publish
> stream of messages on two ka
I am accessing kafka through java code. The use case is, I want to publish
stream of messages on two kafka topic, which would be in a transaction.
My point of concern is would it be better to use one single global kafka
producer for all messages or invoking and deleting a new kafkaPoducer for
ever
Can you capture stack trace on the broker and pastebin it ?
Broker log may also provide some clue.
Thanks
On Mon, Dec 18, 2017 at 4:46 AM, HKT wrote:
> Hello,
>
> I was testing the transactional message on kafka.
> but I get a problem.
> the producer always blocking at second commitTransaction
Hi
I am wondering if there is a way to run the SSL and PLAINTEXT mode together
? I am running Kafka 10.2.1. We want our internal clients to use the
PLAINTEXT mode to write to certain topics, but any external clients should
use SSL to read messages on those topics. We also want to enforce ACLs.
To
Hello,
I was testing the transactional message on kafka.
but I get a problem.
the producer always blocking at second commitTransaction.
Here is my code:
Properties kafkaProps = new Properties();
kafkaProps.setProperty("bootstrap.servers", "localhost:9092");
kafkaProps.set
Hi Bill,
I am actually referring to duplicates as completely identical records. I
can observe it when I convert result of left join between KTables to
stream. The resulting stream will often contain identical messages.
For example we have
KTable left
{"claimcounter": 0, "claimreporttime": 55948.
Hello,
are there plans to release version 1.0.1?
We are affected by https://issues.apache.org/jira/browse/KAFKA-6185 and
cannot upgrade at the moment to 1.0.0 (on our UAT env we applied the
patch and problems are gone, but we are reluctant to do this on prod...),
we'd like to know if there a
17 matches
Mail list logo
