063,
> Geschäftsführer: Bo PENG, Wanzhou MENG, Lifang CHEN
>
> This e-mail and its attachments contain confidential information from
> HUAWEI, which is intended only for the person or entity whose address is
> listed above. Any use of the information contained herein in any way
> (inclu
Hi,
a NoSuchMethod indicates that you are using incompatible versions.
You should check that the versions of your job dependencies and the version
cluster you want to run the job on are the same.
Best, Fabian
2016-10-27 7:13 GMT+02:00 NagaSaiPradeep :
> Hi,
> I am
Hi Radu,
I might not have complete understood your problem, but if you do
val env = StreamExecutionEnvironment.getExecutionEnvironment
val tEnv = TableEnvironment.getTableEnvironment(env)
val ds = env.fromElements( (1, 1L, new Time(1,2,3)) )
val t = ds.toTable(tEnv, 'a, 'b, 'c)
val results = t
Hi Paul,
Flink pushes the results of operators (including GroupReduce) to the next
operator or sink as soon as they are computed. So what you are asking for
is actually happening.
However, before the GroupReduceFunction can be applied, the whole data is
sorted in order to group the data. This
Hi Yassine,
I thought I had fixed that bug a few weeks a ago, but apparently the fix
did not catch all cases.
Can you please reopen FLINK-2662 and post the program to reproduce the bug
there?
Thanks,
Fabian
[1] https://issues.apache.org/jira/browse/FLINK-2662
2016-10-25 12:33 GMT+02:00 Yassine
The window is evaluated when a watermark arrives that is behind the
window's end time.
For instance, give the window in your example there are windows that end at
1:00:00, 1:00:30, 1:01:00, 1:01:30, ... (every 30 seconds).
given the windows above, the window from 00:59:00 to 1:00:00 will be
Fold + Window first, which
> should get rid of my heap issues.
>
>
>
> Thanks!
>
>
>
> *From: *Fabian Hueske <fhue...@gmail.com>
> *Reply-To: *"user@flink.apache.org" <user@flink.apache.org>
> *Date: *Monday, October 24, 2016 at 2:27 PM
>
&g
ion), so I’m looking at a way to customize the
> EventTimeSessionWindow, or perhaps create a custom EventTrigger, to force a
> session to close after either X seconds of inactivity or Y seconds of
> duration (or perhaps after Z events).
>
>
>
>
>
>
>
> *From: *Fabian Hueske <f
Hi Robert,
it is certainly possible to feed the same DataStream into two (or more)
operators.
Both operators should then process the complete input stream.
What you describe is an unintended behavior.
Can you explain how you figure out that both window operators only receive
half of the events?
The translation is done in multiple stages.
1. Parsing (syntax check)
2. Validation (semantic check)
3. Query optimization (rule and cost based)
4. Generation of physical plan, incl. code generation (DataStream program)
The final translation happens in the DataStream nodes, e.g., DataStreamCalc
Hi Pedro,
The sql() method calls the Calcite parser in line 129.
Best, Fabian
2016-10-17 16:43 GMT+02:00 PedroMrChaves :
> Hello,
>
> I am pretty new to Apache Flink.
>
> I am trying to figure out how does Flink parses an Apache Calcite sql query
> to its own
is evaluated and late events are processed alone, i.e.,
in my example <12:09, G> would be processed without [A, B, C, D].
When the allowed lateness is passed, all window state is purged regardless
of the trigger.
Best, Fabian
2016-10-17 16:24 GMT+02:00 Fabian Hueske <fhue...@gmail.com>:
Hi Yassine,
the difference is the following:
1) The BoundedOutOfOrdernessTimestampExtractor is a built-in timestamp
extractor and watermark assigner.
A timestamp extractor tells Flink when an event happened, i.e., it extracts
a timestamp from the event. A watermark assigner tells Flink what the
oop integration for
> AvroParquetInputFormat and CqlBulkOutputFormat in Flink (although we won't
> be using CqlBulkOutputFormat any longer because it doesn't seem to be
> reliable).
>
> -Shannon
>
> From: Fabian Hueske <fhue...@gmail.com>
> Date: Friday, October 14, 2016
Hi Santlal,
I'm afraid I don't know what is going wrong here either.
Debugging and correctly configuring the Taps was one of the major obstacles
when implementing the connector.
Best, Fabian
2016-10-14 14:40 GMT+02:00 Aljoscha Krettek :
> +Fabian directly looping in Fabian
Hi everybody,
I would like to propose to deprecate the utility methods to read data with
Hadoop InputFormats from the (batch) ExecutionEnvironment.
The motivation for deprecating these methods is reduce Flink's dependency
on Hadoop but rather have Hadoop as an optional dependency for users that
apply() accepts a WindowFunction which is essentially the same as a
GroupReduceFunction, i.e., you have an iterator over all events in the
window.
If you only want to count, you should have a look at incremental window
aggregation with a ReduceFunction or FoldFunction [1].
Best, Fabian
[1]
Hi Ken,
FYI: we just received a pull request for FLIP-12 [1].
Best, Fabian
[1] https://github.com/apache/flink/pull/2629
2016-10-11 9:35 GMT+02:00 Fabian Hueske <fhue...@gmail.com>:
> Hi Ken,
>
> I think your solution should work.
> You need to make sure though, that y
Hi Pedro,
the DataStream program would like this:
val eventData: DataStream[?] = ???
val result = eventData
.filter("action = denied")
.keyBy("user", "ip")
.timeWindow(Time.hours(1))
.apply("window.end, user, ip, count(*)")
.filter("count > 5")
.map("windowEnd, user, ip")
Note,
Hi Pedro,
support for window aggregations in SQL and Table API is currently work in
progress.
We have a pull request for the Table API and will add this feature for the
next release.
For SQL we depend on Apache Calcite to include the TUMBLE keyword in its
parser and optimizer.
At the moment the
Hi Shannon,
I tried to reproduce the problem in a unit test without success.
My test configures a HadoopOutputFormat object, serializes and deserializes
it, cally open, and verifies that a configured String property is present
in the getRecordWriter() method.
Next I would try to reproduce the
er(",")
>> .includeFields("101")
>> .ignoreInvalidLines()
>> .types(String.class, String.class);
>> withReadCSV.writeAsText("C:\\Users\\yassine\\Desktop\\withreadcsv.txt",
>> FileSystem.WriteMode.OVERWRITE).setPar
returnRecord = null;
> do {
> try {
> returnRecord = super.nextRecord(record);
> } catch (IOException e) {
> e.printStackTrace();
> }
> } while (returnRecord == null && !reachedEnd())
Hi Ken,
I think your solution should work.
You need to make sure though, that you properly manage the state of your
function, i.e., memorize all records which have been received but haven't
be emitted yet.
Otherwise records might get lost in case of a failure.
Alternatively, you can implement
Hi,
you can do it like this:
1) you have to split each label record of the main dataset into separate
records:
(0,List(a, b, c, d, e, f, g)) -> (0, a), (0, b), (0, c), ..., (0, g)
(1,List(b, c, f, a, g)) -> (1, b), (1, c), ..., (1, g)
2) join word index dataset with splitted main dataset:
Hi Rashmi,
as Marton said, you do not need to start a local Flink instance
(start-lcoal.bat) if you want to run programs from your IDE.
Maybe running a local instance causes a conflict when starting an instance
from IDE.
Developing and running Flink programs on Windows should work, both from the
rto...@gmail.com>:
> Humm
>
> Your solution compile with out errors, but IncludedFields Isn't working:
> [image: Imágenes integradas 1]
>
> The output is incorrect:
> [image: Imágenes integradas 2]
>
> The correct result must be only 1º Column
> (a,1)
> (aa,1)
>
As the exception says the class
org.apache.flink.api.scala.io.jdbc.JDBCInputFormat does not exist.
You have to do:
import org.apache.flink.api.java.io.jdbc.JDBCInputFormat
There is no Scala implementation of this class but you can also use Java
classes in Scala.
2016-10-07 21:38 GMT+02:00
_OBJECT> using a private long value in
> the mapper that increments on every map call). It works, but by any chance
> is there a more succinct way to do it?
>
> On Thu, Oct 6, 2016 at 1:50 PM, Fabian Hueske <fhue...@gmail.com> wrote:
>
>> Maybe this can be done by
Maybe this can be done by assigning the same window id to each of the N
local windows, and do a
.keyBy(windowId)
.countWindow(N)
This should create a new global window for each window id and collect all N
windows.
Best, Fabian
2016-10-06 22:39 GMT+02:00 AJ Heller :
> The
> >
> > On Tue, Oct 4, 2016 at 11:56 PM, Hironori Ogibayashi <
> ogibaya...@gmail.com>
> > wrote:
> >>
> >> Thank you for the response.
> >> Regarding adding to the page, I will check with our PR department.
> >>
> >> Regards,
>
Hi Yassine,
AFAIK, there is no built-in way to ignore corrupted compressed files.
You could try to implement a FileInputFormat that wraps the CsvInputFormat
and forwards all calls to the wrapped CsvIF.
The wrapper would also catch and ignore the EOFException.
If you do that, you would not be
Hi Philipp,
If I got your requirements right you would like to:
1) load an initial hashmap via JDBC
2) update the hashmap from a stream
3) use the hashmap to enrich another stream.
You can use a CoFlatMap to do this:
stream1.connect(stream2).flatMap(new YourCoFlatMapFunction).
FYI: FLINK-4497 [1] requests Scala tuple and case class support for the
Cassandra sink and was opened about a month ago.
[1] https://issues.apache.org/jira/browse/FLINK-4497
2016-09-30 23:14 GMT+02:00 Stephan Ewen :
> How hard would it be to add case class support?
>
>
et but for what I got, this is slightly different from what I need.
>
> 2016-09-30 10:04 GMT+02:00 Fabian Hueske <fhue...@gmail.com>:
>
>> Hi Simone,
>>
>> I think I have a solution for your problem:
>>
>> val s: DataStream[(Long, Int, ts)] = ??? // (id, state,
Hi Simone,
I think I have a solution for your problem:
val s: DataStream[(Long, Int, ts)] = ??? // (id, state, time)
val stateChanges: DataStream[(Int, Int)] = s // (state, cntUpdate)
.keyBy(_._1) // key by id
.flatMap(new StateUpdater) // StateUpdater is a stateful FlatMapFunction.
It has
Great, thanks!
I gave you contributor permissions in JIRA. You can now also assign issues
to yourself if you decide to continue to contribute.
Best, Fabian
2016-09-29 16:48 GMT+02:00 jaxbihani :
> Hi Fabian
>
> My JIRA user is: jaxbihani
> I have created a pull request
Hi Ken,
you can certainly have partitioned sources and sinks. You can control the
parallelism by calling .setParallelism() method.
If you need a partitioned sink, you can call .keyBy() to hash partition.
I did not completely understand the requirements of your program. Can you
maybe provide
Hi Anchit,
Flink does not yet have a streaming sink connector for HBase. Some members
of the community are working on this though [1].
I think we resolved a similar issue for the Kafka connector recently [2].
Maybe the related commits contain some relevant code for your problem.
Best, Fabian
Hi Markus,
thanks for the stacktraces!
The client is indeed stuck in the optimizer. I have to look a bit more into
this.
Did you try to set JoinHints in your plan? That should reduce the plan
space that is enumerated and therefore reduce the optimization time (maybe
enough to run your application
Hi Yassine, can you share a stacktrace of the job when it got stuck?
Thanks, Fabian
2016-09-22 14:03 GMT+02:00 Yassine MARZOUGUI :
> The input splits are correctly assgined. I noticed that whenever the job
> is stuck, that is because the task *Combine (GroupReduce at
No, this is not possible unless you use an external service such as a
database.
The assigners might run on different machines and Flink does not provide
utilities for r/w shared state.
Best, Fabian
2016-09-15 20:17 GMT+02:00 Saiph Kappa :
> And is it possible to share
+1
I ran into that issue as well. Would be great to have that in the docs!
2016-09-09 11:49 GMT+02:00 Robert Metzger :
> Hi Steffen,
>
> I think it would be good to add it to the documentation.
> Would you like to open a pull request?
>
>
> Regards,
> Robert
>
>
> On Mon,
dence$3: scala.reflect.ClassTag[R])org.apache.flink.api.scala.
> DataSet[R]
> match expected type ?
>
> Thanks!
> Frank
>
>
> On Thu, Sep 8, 2016 at 6:56 PM, Fabian Hueske <fhue...@gmail.com> wrote:
>
>> Hi Frank,
>>
>> input should be of
Hi Frank,
input should be of DataSet[(BSONWritable, BSONWritable)], so a
Tuple2[BSONWritable, BSONWritable], right?
Something like this should work:
input.map( pair => pair._1.toString )
Pair is a Tuple2[BSONWritable, BSONWritable], and pair._1 accesses the key
of the pair.
Alternatively you
I would assign timestamps directly at the source.
Timestamps are not striped of by operators.
Reassigning timestamps somewhere in the middle of a job can cause very
unexpected results.
2016-09-08 9:32 GMT+02:00 Dong-iL, Kim :
> Thanks for replying. pushpendra.
> The
nst a 'bunch
> of keys' from DS2 and DS2 could shrink/expand in terms of the no., of
> keys will the key-value shard work in this case?
>
> On Wed, Sep 7, 2016 at 7:44 PM, Fabian Hueske <fhue...@gmail.com> wrote:
>
>> Operator state is always local in Flink. However
cross the cluster within this job1 ?
>
>
>
> On Wed, Sep 7, 2016 at 6:33 PM, Fabian Hueske <fhue...@gmail.com> wrote:
>
>> Is writing DataStream2 to a Kafka topic and reading it from the other job
>> an option?
>>
>> 2016-09-07 19:07 GMT+02:00
localize the cache within the cluster.
> Is there a way?
>
> Best Regards
> CVP
>
> On Wed, Sep 7, 2016 at 5:00 PM, Fabian Hueske <fhue...@gmail.com> wrote:
>
>> Hi,
>>
>> Flink does not provide shared state.
>> However, you can broadcast a str
Hi,
Flink does not provide shared state.
However, you can broadcast a stream to CoFlatMapFunction, such that each
operator has its own local copy of the state.
If that does not work for you because the state is too large and if it is
possible to partition the state (and both streams), you can
Thanks for the suggestion Vishnu!
Stackoverflow documentation looks great. I like the easy contribution and
versioning features.
However, I am a bit skeptical. IMO, Flink's primary documentation must be
hosted by Apache. Out-sourcing such an important aspect of a project to an
external service is
(x)) then
> evaluated relative to maxEventTime - lastWaterMarkTime. So if (maxEventTime
> - lastWaterMarkTime) > x * 1000 then the window is evaluated?
>
>
> Paul
> ------
> *From:* Fabian Hueske <fhue...@gmail.com>
> *Sent:* Thursday, September 1, 2016 1:
roblem , or there is no harm in storing the
> DTO ?
>
> I think the documentation should specify the point that the state will be
> maintained for user-defined operators to avoid confusion.
>
> Regards,
> Vinay Patil
>
> On Thu, Sep 1, 2016 at 1:12 PM, Fabian Hueske-2 [vi
Hi Paul,
BoundedOutOfOrdernessTimestampExtractor implements the
AssignerWithPeriodicWatermarks interface.
This means, Flink will ask the assigner in regular intervals (configurable
via StreamExecutionEnvironment.getConfig().setAutoWatermarkInterval()) for
the current watermark.
The watermark will
ame key on matchingAndNonMatching and
> flatmap to take care of rest logic.
>
> Or are you suggestion to use Co-FlatMapFunction after the outer-join
> operation (I mean after doing the window and
> getting matchingAndNonMatching stream )?
>
> Regards,
> Vinay Patil
>
> On
using RocksDB as state
>> backend since the state is not gone after checkpointing ?
>>
>> P.S I have kept the watermark behind by 1500 secs just to be safe on
>> handling late elements but to tackle edge case scenarios like the one
>> mentioned above we are having a
is that
> the previous windows will not purge, is that correct?
>
> final DataStream alertingMsgs = keyedStream
> .window(TumblingEventTimeWindows.of(Time.minutes(1)))
> .trigger(CountTrigger.of(1))
> .apply(new MyWindowProcessor())
Hi Paul,
This blog post [1] includes an example of an early trigger that should
pretty much do what you are looking for.
This one [2] explains the windowing mechanics of Flink (window assigner,
trigger, function, etc).
Hope this helps,
Fabian
[1]
Hi Flavio,
yes, Joda should not be excluded.
This will be fixed in Flink 1.1.2.
Cheers, Fabian
2016-08-29 11:00 GMT+02:00 Flavio Pompermaier :
> Hi to all,
> I've tried to upgrade from Flink 1.0.2 to 1.1.1 so I've copied the
> excludes of the maven shade plugin from the
Hi Markus,
you might be right, that a lot of time is spend in optimization.
The optimizer enumerates all alternatives and chooses the plan with the
least estimated cost. The degrees of freedom of the optimizer are rather
restricted (execution strategies and the used partitioning & sorting keys.
Thanks Ufuk and everybody who contributed to the release!
Cheers, Fabian
2016-08-08 20:41 GMT+02:00 Henry Saputra :
> Great work all. Great Thanks to Ufuk as RE :)
>
> On Monday, August 8, 2016, Stephan Ewen wrote:
>
> > Great work indeed, and big
Yes, that was my fault. I'm used to auto reply-all on my desktop machine,
but my phone just did a simple reply.
Sorry for the confusion,
Fabian
2016-06-29 19:24 GMT+02:00 Ovidiu-Cristian MARCU <
ovidiu-cristian.ma...@inria.fr>:
> Thank you, Aljoscha!
> I received a similar update from Fabian,
Hi Josh,
I assume that you build the SNAPSHOT version yourself. I had similar
version conflicts for Apache HttpCore with Flink SNAPSHOT versions on EMR.
The problem is cause by a changed behavior in Maven 3.3 and following
versions.
Due to these changes, the dependency shading is not working
1:04 Christophe Salperwyck
>> >> <christophe.salperw...@gmail.com> wrote:
>> >>>
>> >>> Thanks for the feedback and sorry that I can't try all this straight
>> >>> away.
>> >>>
>> >>> Is there a
epartitioning by key, which is unnecessary since
> I don't really care about keys.
>
> On 9 June 2016 at 22:00, Fabian Hueske <fhue...@gmail.com> wrote:
>
>> Hi Yukun,
>>
>> the problem is that the KeySelector is internally invoked multiple times.
>&g
<
christophe.salperw...@gmail.com>:
> Hi Fabian,
>
> Thanks for the help, I will try that. The backpressure was on the source
> (HBase).
>
> Christophe
>
> 2016-06-09 16:38 GMT+02:00 Fabian Hueske <fhue...@gmail.com>:
>
>> Hi Christophe,
>>
>>
Great, thank you!
2016-06-09 17:38 GMT+02:00 Elias Levy <fearsome.lucid...@gmail.com>:
>
> On Thu, Jun 9, 2016 at 5:16 AM, Fabian Hueske <fhue...@gmail.com> wrote:
>
>> thanks for your feedback. I think those are good observations and
>> suggestions to improve
We solved this problem yesterday at the Flink Hackathon.
The issue was that the source function was started with parallelism 4 and
each function read the whole file.
Cheers, Fabian
2016-06-06 16:53 GMT+02:00 Biplob Biswas :
> Hi,
>
> I tried streaming the source data 2
nces of map?
>
> 3) Question Cleared.
>
> 4) My question was can I use same ExecutionEnvironment for all flink
> programs in a module.
>
> 5) Question Cleared.
>
>
> Regards
> Ravikumar
>
>
>
> On 9 June 2016 at 17:58, Fabian Hueske <fhue...@gmail.com&
Hi Christophe,
where does the backpressure appear? In front of the sink operator or before
the window operator?
In any case, I think you can improve your WindowFunction if you convert
parts of it into a FoldFunction.
The FoldFunction would take care of the statistics
Hi,
you are computing a running aggregate, i.e., you're getting one output
record for each input record and the output record is the record with the
largest value observed so far.
If the record with the largest value is the first, the record is sent out
another time. This is what happened with
Hi Yukun,
the problem is that the KeySelector is internally invoked multiple times.
Hence it must be deterministic, i.e., it must extract the same key for the
same object if invoked multiple times.
The documentation is not discussing this aspect and should be extended.
Thanks for pointing out
Hi Ravikumar,
I'll try to answer your questions:
1) If you set the parallelism of a map function to 1, there will be only a
single instance of that function regardless whether it is execution locally
or remotely in a cluster.
2) Flink does also support aggregations, (reduce, groupReduce, combine,
Hi Elias,
thanks for your feedback. I think those are good observations and
suggestions to improve the Kafka producers.
The best place to discuss such improvements is the dev mailing list.
Would like to repost your mail there or open JIRAs where the discussion
about these changes can continue?
Hi Tarandeep,
the exception suggests that Flink tries to serialize RecordsFilterer as a
user function (this happens via Java Serialization).
I said suggests because the code that uses RecordsFilterer is not included.
To me it looks like RecordsFilterer should not be used as a user function.
It
Hi Elias,
yes, reduce, fold, and the aggregation functions (sum, min, max, minBy,
maxBy) on WindowedStream preform eager aggregation, i.e., the functions are
apply for each value that enters the window and the state of the window
will consist of a single value. In case you need access to the
No, that is not supported yet.
Beam provides a common API but the Flink runner translates programs against
batch sources into the DataSet API programs and Beam programs against
streaming source into DataStream programs.
It is not possible to mix both.
2016-05-26 10:00 GMT+02:00 Ashutosh Kumar
Hi Kirsti,
I'm not aware of anybody working on this issue.
Would you like to create a JIRA issue for it?
Best, Fabian
2016-05-23 16:56 GMT+02:00 KirstiLaurila :
> Is there any plans to implement this kind of feature (possibility to write
> to
> data specified
Actually, the program works correctly (according to the DataStream API)
Let me explain what happens:
1) You do not initialize the count variable, so it will be 0 (summing 0s
results in 0)
2) DataStreams are considered to be unbound (have an infinite size). KeyBy
does not group the records because
The problem seems to occur quite often.
Did you update your Flink version recently? If so, could you try to
downgrade and see if the problem disappears.
Is it otherwise possible that it is cause by faulty hardware?
2016-05-20 18:05 GMT+02:00 Flavio Pompermaier :
> This
I think that sentence is misleading and refers to the internals of Flink.
It should be removed, IMO.
You can only union two DataSets. If you want to union more, you have to do
it one by one.
Btw. union does not cause additional processing overhead.
Cheers, Fabian
2016-05-19 14:44 GMT+02:00
I think union is what you are looking for.
Note that all data sets must be of the same type.
2016-05-18 16:15 GMT+02:00 Ritesh Kumar Singh :
> Hi,
>
> How can I perform a reduce operation on a group of datasets using Flink?
> Let's say my map function gives out n
gt;>
>> On Tue, May 17, 2016 at 11:27 AM, Stephan Ewen <se...@apache.org> wrote:
>>
>>> Hi Naveen!
>>>
>>> I assume you are using Hadoop 2.7+? Then you should not see the
>>> ".valid-length" file.
>>>
>>> The fix you m
Hi Prateek,
the missing numbers are an artifact from how the stats are collected.
ATM, Flink does only collect these metrics for data which is sent over
connections *between* Flink operators.
Since sources and sinks connect to external systems (and not Flink
operators), the dash board does not
ansformations and rolling file sink
> pipeline.
>
>
>
> Thanks,
> Naveen
>
> From: Fabian Hueske <fhue...@gmail.com>
> Reply-To: "user@flink.apache.org" <user@flink.apache.org>
> Date: Friday, May 13, 2016 at 4:26 PM
>
> To: &qu
Hi,
Flink's exactly-once semantics do not mean that events are processed
exactly-once but that events will contribute exactly-once to the state of
an operator such as a counter.
Roughly, the mechanism works as follows:
- Flink peridically injects checkpoint markers into the data stream. This
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
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
port us in the development. We are waiting for them to start a
> discussion and as soon as we will have a more clear idea on how to proceed,
> we will validate it with the stuff you just said. Your confidence in
> Flink's operators gives up hope to achieve a clean solution.
>
> Than
Maybe the last example of this blog post is helpful [1].
Best, Fabian
[1]
https://www.mapr.com/blog/essential-guide-streaming-first-processing-apache-flink
2016-05-10 17:24 GMT+02:00 Srikanth :
> Hi,
>
> I read the following in Flink doc "We can explicitly specify a
if it is really required to implement a
> custom runtime operator but given the complexity of the integration of two
> distribute systems, we assumed that more control would allow more
> flexibility and possibilities to achieve an ideal solution.
>
>
>
>
>
> 2016-05-03
Hi Palle,
you can recursively read all files in a folder as explained in the
"Recursive Traversal of the Input Path Directory" section of the Data
Source documentation [1].
The easiest way to read line-wise JSON objects is to use
ExecutionEnvironment.readTextFile() which reads text files
apFunctionWrapper(new MetaMapFunction(meta)))
>
> What do you think? Is there the possibility to open a broadcasted Dataset
> as a Map instead of a List?
>
> Best,
> Flavio
>
>
> On Fri, May 6, 2016 at 12:06 PM, Fabian Hueske <fhue...@gmail.com> wrote:
>
>&
6 Actor System (instantiate into
> RichSinkFunction's open method)
>
> Am I wrong?
>
> Thanks again,
> Andrea
>
> 2016-05-06 13:47 GMT+02:00 Fabian Hueske <fhue...@gmail.com>:
>
>> Hi Andrea,
>>
>> you can use any OutputFormat to emit data fr
Hi Palle,
this sounds indeed like a good use case for Flink.
Depending on the complexity of the aggregated historical views, you can
implement a Flink DataStream program which builds the views on the fly,
i.e., you do not need to periodically trigger MR/Flink/Spark batch jobs to
compute the
Hi Andrea,
you can use any OutputFormat to emit data from a DataStream using the
writeUsingOutputFormat() method.
However, this method does not guarantee exactly-once processing. In case of
a failure, it might emit some records a second time. Hence the results will
be written at least once.
Hope
at it could be easy
> to define a link from operators to TableEnvironment and then to TableSource
> (using the lineage tag/source-id you said) and, finally to its metadata. I
> don't know whether this is specific only to us, I just wanted to share our
> needs and see if the table AP
Hi Max,
it is not possible to deactivate spilling to disk at the moment.
It might be possible to implement, but this would require a few more
changes to make it feasible.
For instance, we would need to add more fine-grained control about how
memory is distributed among operators.
This is
Sorry, I confused the mail threads. We're already on the user list :-)
Thanks for the suggestion.
2016-05-04 17:35 GMT+02:00 Fabian Hueske <fhue...@gmail.com>:
> I'm not so much familiar with the Kafka connector.
> Can you post your suggestion to the user or dev mailing list?
>
&
I'm not so much familiar with the Kafka connector.
Can you post your suggestion to the user or dev mailing list?
Thanks, Fabian
2016-05-04 16:53 GMT+02:00 Sendoh :
> Glad to see it's developing.
> Can I ask would the same feature (reconnect) be useful for Kafka
1201 - 1300 of 1535 matches
Mail list logo