Hi guys.
Thanks for helping out.
We downgraded to HBase 0.98 and resolved some classpath issues and then
it worked.
/Palle
- Original meddelelse -
> Fra: Stephan Ewen
> Til: user@flink.apache.org
> Dato: Ons, 11. maj 2016 17:19
> Emne: Re: HBase write problem
>
> Just to narrow down
I think I found a workaround. Instead of reading Avro files as
GenericRecords, if I read them as specific records and then use a map to
convert (typecast) them as GenericRecord, the problem goes away.
I ran some tests and so far this workaround seems to be working in my local
setup.
-Tarandeep
O
The Flink PMC is pleased to announce the availability of Flink 1.0.3.
The official release announcement:
http://flink.apache.org/news/2016/05/11/release-1.0.3.html
Release binaries:
http://apache.openmirror.de/flink/flink-1.0.3/
Please update your Maven dependencies to the new 1.0.3 version and
Hi,
we have stream job with quite large state (few GB), we're using
FSStateBackend and we're storing checkpoints in hdfs.
What we observe is that v. often old checkpoints are not discarded
properly. In hadoop logs I can see:
2016-05-10 12:21:06,559 INFO BlockStateChange: BLOCK* addToInvalidat
Hi all,
If I have a Graph g: Graph g
and I would like to normalize all vertex values by the absolute max of all
vertex values -> what API function would I choose?
Thanks in advance!
Lydia
Great :)
On Thu, May 12, 2016 at 10:01 AM, Palle wrote:
> Hi guys.
>
> Thanks for helping out.
>
> We downgraded to HBase 0.98 and resolved some classpath issues and then it
> worked.
>
> /Palle
>
> - Original meddelelse -
>
> *Fra:* Stephan Ewen
> *Til:* user@flink.apache.org
> *Dato:*
No I am using 0.8.0.2 kafka. I did some experiments with changing the
parallelism from 4 to 16 now the lag has reduced to 20 min from 2 hours,
the cpu utilization (load avg) has gone up from 20-30 % to 50-60 % , so
parallelism does seem to play a role in reducing the processing lag in
flink as I e
Ok, I tested it and it works on the same example. :)
2016-05-11 12:25 GMT+02:00 Vasiliki Kalavri :
> Hi Simone,
>
> Fabian has pushed a fix for the streaming TableSources that removed the
> Calcite Stream rules [1].
> The reported error does not appear anymore with the current master. Could
> you
Hi,
I am running this following sample code to understand how iteration and
broadcast works in streaming context.
final StreamExecutionEnvironment env =
StreamExecutionEnvironment.getExecutionEnvironment();
env.setParallelism(4);
long i = 5;
Data
Hi Martin,
You can use a FoldFunction and a WindowFunction to process the same!
window. The FoldFunction is eagerly applied, so the window state is only
one element. When the window is closed, the aggregated element is given to
the WindowFunction where you can add start and end time. The iterator
Hi Tarandeep,
the AvroInputFormat was recently extended to support GenericRecords. [1]
You could also try to run the latest SNAPSHOT version and see if it works
for you.
Cheers, Fabian
[1] https://issues.apache.org/jira/browse/FLINK-3691
2016-05-12 10:05 GMT+02:00 Tarandeep Singh :
> I think I
Yes, this should work.
On Tue, 10 May 2016 at 19:01 Srikanth wrote:
> Yes, will work.
> I was trying another route of having a "finalize & purge trigger" that will
>i) onElement - Register for event time watermark but not alter nested
> trigger's TriggerResult
> ii) OnEventTime - Always pu
Good to know :)
On 12 May 2016 at 11:16, Simone Robutti
wrote:
> Ok, I tested it and it works on the same example. :)
>
> 2016-05-11 12:25 GMT+02:00 Vasiliki Kalavri :
>
>> Hi Simone,
>>
>> Fabian has pushed a fix for the streaming TableSources that removed the
>> Calcite Stream rules [1].
>> Th
I cannot follow completely. TwoInputStreamOperators defines two methods
to process watermarks for each stream.
So you can sync both stream within your outer join operator you plan to
implement.
-Matthias
On 05/11/2016 05:00 PM, Alexander Gryzlov wrote:
> Hello,
>
> We're implementing a streamin
Hmm, probably I don't really get how Flink's execution model works. As far
as I understand, the preferred way to throttle down stream consumption is
to simply have an operator with a conditional Thread.sleep() inside.
Wouldn't calling sleep() in either of TwoInputStreamOperator's
processWatermarkN(
That is correct. But there is no reason to throttle an input stream.
If you implements an Outer-Join you will have two in-memory buffers
holding the record of each stream of your "time window". Each time you
receive a watermark, you can remove all "expired" records from the
buffer of the other str
Yes, this is generally a viable design, and is actually something we
started off with.
The problem in our case is, however, that either of the streams can
occasionally (due to external producer's issues) get stuck for an arbitrary
period of time, up to several hours. Buffering the other one during
I see. But even if you would have an operator (A,B)->(A,B), it would not
be possible to block A if B does not deliver any data, because of
Flink's internal design.
You will need to use an custom solution: something like to a map (one
for each steam) that use an side-communication channel (ie, exte
Hey Maciek,
thanks for reporting this. Having files linger around looks like a bug to me.
The idea behind having the recursive flag set to false in the
AbstractFileStateHandle.discardState() call is that the
FileStateHandle is actually just a single file and not a directory.
The second call tryin
Thanks for the help. I use a Fold and a WindowFunction in conjunction now
and it works fine. Though I wish there would be a less complicated way to
do this.
cheers Martin
On Thu, May 12, 2016 at 11:59 AM, Fabian Hueske wrote:
> Hi Martin,
>
> You can use a FoldFunction and a WindowFunction to p
Hi Lydia,
there is no dedicated Gelly API method that performs normalization. If you
know the max value, then a mapVertices() would suffice. Otherwise, you can
get the Dataset of vertices with getVertices() and apply any kind of
operation supported by the Dataset API on it.
Best,
-Vasia.
On May 1
The issue is here: https://issues.apache.org/jira/browse/FLINK-3902
(My "explanation" before dosn't make sense actually and I don't see a
reason why this should be related to having many state handles.)
On Thu, May 12, 2016 at 3:54 PM, Ufuk Celebi wrote:
> Hey Maciek,
>
> thanks for reporting th
Hi to all,
running a job that writes parquet-thrift files I had this exception (in a
Task Manager):
io.netty.channel.nio.NioEventLoop - Unexpected
exception in the selector loop.
java.lang.OutOfMemoryError: Java heap space
2016-05-12 18:49:11,302 WARN
org.jboss.netty.ch
Hi
How can i measure throughput and latency of my application in flink 1.0.2
?
Regards
Prateek
--
View this message in context:
http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/How-to-measure-Flink-performance-tp6741p6863.html
Sent from the Apache Flink User Mailing List
Hi Prateek,
regarding throughput, what about simply filling the input Kafka topic
with some (a lot) of messages and monitor (e.g.
http://quantifind.github.io/KafkaOffsetMonitor/) how quickly Flink can
work the lag off. The messages should be representative of your use
case, of course.
Latency is
Hi,
I have a question regarding the Aggregators of a Scatter-Gather Iteration.
Is it possible to have a global aggregator that is accessible in
VertexUpdateFunction() and MessagingFunction() at the same time?
Thanks in advance,
Lydia
Hello,
I was reading about Flink's checkpoint and wanted to check if I correctly
understood the usage of barriers for exactly once processing.
1) Operator does alignment by buffering records coming after a barrier
until it receives barrier from all upstream operators instances.
2) Barrier is alw
thanks,
I'll try to reproduce it in some test by myself...
maciek
On 12/05/2016 18:39, Ufuk Celebi wrote:
The issue is here: https://issues.apache.org/jira/browse/FLINK-3902
(My "explanation" before dosn't make sense actually and I don't see a
reason why this should be related to having many s
Hi Gordon,
Thanks for advice - it's work perfect but only in elasticsearch case.
This pom version works for elasticsearch 2.2.1.
org.apache.flink
flink-connector-elasticsearch2_${scala.version}
1.1-SNAPSHOT
jar
false
${project.build.directory}/classes
org/apache/flink/**
If I can throw in my 2 cents, I agree with what Elias says. Without that
feature (not partitioning already partitioned Kafka data), Flink is in bad
position for common simpler processing, that don't involve shuffling at
all, for example simple readKafka-enrich-writeKafka . The systems like the
new
Hi,
I was recently experimenting with savepoints and various situations in
which they succeed or fail. I expected this example to fail:
https://gist.github.com/AndrewWhitaker/fa46db04066ea673fe0eda232f0a5ce1
Basically, the first program runs with a count window. The second program
is identical e
"Flink can't successfully restore a checkpoint" should be "Flink can't
successfully restore a savepoint".
On Thu, May 12, 2016 at 3:44 PM, Andrew Whitaker <
andrew.whita...@braintreepayments.com> wrote:
> Hi,
>
> I was recently experimenting with savepoints and various situations in
> which they
On Thu, May 12, 2016 at 10:44 PM, Andrew Whitaker
wrote:
> From what I've observed, most of the time when Flink can't successfully
> restore a checkpoint it throws an exception saying as much. I was expecting
> to see that behavior here. Could someone explain why this "works" (as in,
> flink accep
...btw I found this (in folder:
"flink/flink-streaming-connectors/flink-connector-elasticsearch2.pom.xml")
:
2.2.1
I change it to 2.3.2 version and of course rebuild with that command "mvn
clean install -DskipTests"
...but nothing is changed.
2016-05-12 22:39 GMT+02:00 rafal green :
> Sorr
Hello all,
Let's say I want to hold some state value derived during one
transformation, and then use that same state value in a subsequent
transformation? For example:
myStream
.keyBy(fieldID) // Some field ID, may be 0
.map(new MyStatefulMapper())
.map(new MySubsequentMapper())
Now, I defi
I don't think the valuestate defined in one map function is accessible in
other map function this is my understanding, also you need to be aware
there will be instance of map function created for each of your tuple in
your stream, I had a similar use case where I had to pass in some state
from one
Hi Lydia,
registered aggregators through the ScatterGatherConfiguration are
accessible both in the VertexUpdateFunction and in the MessageFunction.
Cheers,
-Vasia.
On 12 May 2016 at 20:08, Lydia Ickler wrote:
> Hi,
>
> I have a question regarding the Aggregators of a Scatter-Gather Iteration.
Hi Prateek,
https://github.com/dataArtisans/yahoo-streaming-benchmark/blob/master/flink-benchmarks/src/main/java/flink/benchmark/utils/ThroughputLogger.java
https://github.com/dataArtisans/yahoo-streaming-benchmark/blob/master/flink-benchmarks/src/main/java/flink/benchmark/utils/AnalyzeTool.java
Hi Vasia,
yes, but only independently within each Function or not?
If I set the aggregator in VertexUpdateFunction then the newly set value is not
visible in the MessageFunction.
Or am I doing something wrong? I would like to have a shared aggregator to
normalize vertices.
> Am 13.05.2016 um
39 matches
Mail list logo