Hi Ovidiu,
This is still subject to community discussion. We created a wiki page
to keep track of the essential features to be included in 1.0:
https://cwiki.apache.org/confluence/display/FLINK/1.0+Release
Otherwise, everything which is marked "Fix version 1.0.0" in JIRA is a
candidate to be
gt; On Dec 3, 2015, at 6:11 PM, Vladimir Stoyak <vsto...@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
Hi Niels,
Sorry for hear you experienced this exception. From a first glance, it
looks like a bug in Hadoop to me.
> "Not retrying because the invoked method is not idempotent, and unable to
> determine whether it was invoked"
That is nothing to worry about. This is Hadoop's internal retry
Hi Cory,
Thanks for reporting the issue. Scala should run independently of the
Java version. We are already using ASM version 5.0.4. However, some
code uses the ASM4 op codes which don't seem to be work with Java 8.
This needs to be fixed. I'm filing a JIRA.
Cheers,
Max
On Mon, Dec 7, 2015 at
For completeness, could you provide a stack trace of the error message?
On Mon, Dec 7, 2015 at 6:56 PM, Maximilian Michels <m...@apache.org> wrote:
> Hi Cory,
>
> Thanks for reporting the issue. Scala should run independently of the
> Java version. We are already using ASM vers
Hi Naveen,
I think you're not using the latest 1.0-SNAPSHOT. Did you build from
source? If so, you need to build again because the snapshot API has
been updated recently.
Best regards,
Max
On Thu, Dec 3, 2015 at 6:40 PM, Madhire, Naveen
wrote:
> Hi,
>
> I am
Hi Flavio,
I was working on this some time ago but it didn't make it in yet and
priorities shifted a bit. The pull request is here:
https://github.com/apache/flink/pull/640
The basic idea is to remove Flink's ResultPartition buffers in memory
lazily, i.e. keep them as long as enough memory is
Hi Cory,
The issue has been fixed in the master and the latest Maven snapshot.
https://issues.apache.org/jira/browse/FLINK-3143
Cheers,
Max
On Tue, Dec 8, 2015 at 12:35 PM, Maximilian Michels <m...@apache.org> wrote:
> Thanks for the stack trace, Cory. Looks like you were on the rig
I know this has been fixed already but, out of curiosity, could you
point me to the Kafka JIRA issue for this
bug? From the Flink issue it looks like this is a Zookeeper version mismatch.
On Tue, Dec 1, 2015 at 5:16 PM, Robert Metzger wrote:
> Hi Gyula,
>
> no, I didn't ;)
Thanks! I've linked the issue in JIRA.
On Tue, Dec 1, 2015 at 5:39 PM, Robert Metzger <rmetz...@apache.org> wrote:
> I think its this one https://issues.apache.org/jira/browse/KAFKA-824
>
> On Tue, Dec 1, 2015 at 5:37 PM, Maximilian Michels <m...@apache.org> wrote:
>&
Hi Madhukar,
Thanks for your question. When you instantiate the FlinkKafkaConsumer, you
supply a DeserializationSchema in the constructor. You simply create a
class which implements DeserializationSchema and contains the
KafkaAvroDecoder with the schema registry.
Like so:
public class
The docs have been updated.
On Thu, Nov 19, 2015 at 12:36 PM, Ufuk Celebi wrote:
> I’ve added a note about this to the docs and asked Max to trigger a new build
> of them.
>
> Regarding Aljoscha’s idea: I like it. It is essentially a shortcut for
> configuring the root path.
>
32)
>> at java.io.ObjectOutputStream.writeObject0(ObjectOutputStream.java:1178)
>> at
>> java.io.ObjectOutputStream.defaultWriteFields(ObjectOutputStream.java:1548)
>> at
>> java.io.ObjectOutputStream.writeSerialData(ObjectOutputStream.java:1509)
>> at
>>
Hi Jagat,
I think your issue here are not the JVM options. You are missing shell
environment variables during the container launch. Adding those to the
user's .bashrc or .profile should fix the problem.
Best regards,
Max
On Mon, Nov 23, 2015 at 10:14 PM, Jagat Singh
Hi Madhukar,
I'm not too familiar with Cloudfoundry but seems like you would have
to write a service integration. Ideally, a Hadoop YARN service is
already available in your Cloudfoundry environment.
You could then create an application which deploys a Flink job on YARN.
Best regards,
Max
On
run(SourceStreamTask.java:55)
> at
> org.apache.flink.streaming.runtime.tasks.StreamTask.invoke(StreamTask.java:218)
> at org.apache.flink.runtime.taskmanager.Task.run(Task.java:584)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at test.flink.MyAvro
Hi Christophe,
A fold function has two inputs: The state and a record to update the
state with. So you can update the SummaryStatistics (state) with each
Put (input).
Cheers,
Max
On Mon, Jun 13, 2016 at 11:04 AM, Christophe Salperwyck
wrote:
> Thanks for the
Is there an issue or a fix for proper use of the ClojureCleaner in
CoGroup.where()?
On Fri, Jun 10, 2016 at 8:07 AM, Aljoscha Krettek wrote:
> Hi,
> yes, I was talking about a Flink bug. I forgot to mention the work-around
> that Stephan mentioned.
>
> On Thu, 9 Jun 2016 at
Hi Theofilos,
Flink doesn't send the local client output to the Yarn cluster. I
think this will only change once we move the entire execution of the
Job to the cluster framework. All output of the actual Flink job
should be within the JobManager or TaskManager logs.
There is something wrong with
Hi Ravikumar,
In short: No, you can't use closures to maintain a global state. If
you want to keep an always global state, you'll have to use
parallelism 1 or an external data store to keep that global state.
Is it possible to break up your global state into a set of local
states which can be
Hi Josh,
I'm not a RocksDB expert but the workaround you described should work.
Just bear in mind that accessing RocksDB concurrently with a Flink job
can result in an inconsistent state. Make sure to perform atomic
updates and clear the RocksDB cache for the item.
Cheers,
Max
On Mon, Jun 13,
.
>
> If you are interested in this we can work together on adding proper support
> for TTL (time-to-live) to the Flink state abstraction.
>
> Cheers,
> Aljoscha
>
> On Mon, 13 Jun 2016 at 12:21 Maximilian Michels <m...@apache.org> wrote:
>>
>> Hi Josh,
>&g
Hi Arnaud,
One issue per thread please. That makes things a lot easier for us :)
Something positive first: We are reworking the resuming of existing
Flink Yarn applications. It'll be much easier to resume a cluster
using simply the Yarn ID or re-discoering the Yarn session using the
properties
1),3)
> 2> ((25,11),4)
> 2> ((46,44),2
> .."
>
> However, the yarn aggregated log contains only the jobmanager output. Is
> this expected or could it indicate a problem with my hadoop logging
> configuration not picking up taskmanager logs?
>
> Cheers,
> T
n running a Spark job for example on the same setup, the yarn aggregated
> log contains all the information printed out by the application.
>
> Cheers,
> Theofilos
>
>
> On 6/15/2016 10:14 AM, Maximilian Michels wrote:
>
> Please use the `yarn logs -applicationId ` to retrieve
n) cluster cluster to the to-be-created cluster.
Will be fixed in 1.1 and probably backported to 1.0.4.
On Wed, Jun 15, 2016 at 6:05 PM, Maximilian Michels <m...@apache.org> wrote:
> Hi Arnaud,
>
> One issue per thread please. That makes things a lot easier for us :)
>
&g
Thanks!
On Mon, Jun 13, 2016 at 12:34 PM, Christophe Salperwyck
wrote:
> Hi,
> I vote on this issue and I agree this would be nice to have.
>
> Thx!
> Christophe
>
> 2016-06-13 12:26 GMT+02:00 Aljoscha Krettek :
>>
>> Hi,
>> I'm afraid this
The "-yst" or "-yarnstreaming" parameter doesn't have an effect
anymore because the streaming mode has been removed. I filed an issue
some weeks ago: https://issues.apache.org/jira/browse/FLINK-3890
On Wed, May 25, 2016 at 10:27 PM, Aljoscha Krettek wrote:
> Hi Prateek,
>
Small addition: The Flink Runner translates into the DataSet or
DataStream API depending on the "streaming" flag of the
PipelineOptions. The default mode is batch. Ultimately, this flag we
be removed and replaced with an automated decision depending on the
sources used.
On Thu, May 26, 2016 at
Hi Robertson,
You need to supply a TypeInformation for the data read from the InputFormat.
val dataset = env.createInput(input, new TupleTypeInfo(Tuple1.class,
BasicTypeInfo.STRING_TYPE_INFO))
should do the trick.
Cheers,
Max
On Tue, May 31, 2016 at 1:13 PM, Robertson Williams
Hi Martin,
No worries. Thanks for letting us know!
Cheers,
Max
On Mon, May 30, 2016 at 9:17 AM, Martin Junghanns
wrote:
> Hi again,
>
> I had a bug in my logic. It works as expected (which is perfect).
>
> So maybe for others:
>
> Problem:
> - execute
> I'm using a keyby but would like to store the state.
>
> Thus what's the way to go?
>
> How do I have to handle the state in option 2).
>
> Could you give an example?
>
> Thanks
> --Simon
>
>> On 01 Jun 2016, at 15:55, Maximilian Michels <m...@apache.org>
+1 for a more helpful error message.
@Jacob Would you mind opening a JIRA issue at
https://issues.apache.org/jira/browse/FLINK?
On Thu, Jun 23, 2016 at 11:31 AM, Chesnay Schepler wrote:
> We should adjust the error message to contain the keyed stream thingy.
>
>
> On
Hi,
Flink prints the Yarn application id during deployment of the cluster.
You can then query the logs from Yarn using the `yarn logs
-applicationId ` command.
Please have a look at
https://hadoop.apache.org/docs/stable/hadoop-yarn/hadoop-yarn-site/YarnCommands.html#logs
Cheers,
Max
On Thu,
Hi Sendoh,
At the time the article was created, Elasticsearch 2.0 was only in the
making and by the time of publishing it had just been released. That's
why we used version 1.7.3. There is currently no 2.X version of the
Flink adapter but that will change very soon. There is an issue and a
The documentation layout changed in the master. Then new URL:
https://ci.apache.org/projects/flink/flink-docs-master/apis/streaming/savepoints.html
On Thu, Jan 14, 2016 at 2:21 PM, Niels Basjes wrote:
> Yes, that is exactly the type of solution I was looking for.
>
> I'll dive
Hi Welly,
There is no fixed timeline yet but we plan to make progress in terms
of authentication and encryption after the 1.0.0 release.
Cheers,
Max
On Wed, Jan 13, 2016 at 8:34 AM, Welly Tambunan wrote:
> Hi Stephan,
>
> Thanks a lot for the explanation.
>
> Is there any
Hi Andrea,
The number of network buffers should be sufficient. Actually, assuming
you have 16 task slots on each of the 25 nodes, it should be enough to
have 16^2 * 25 * 4 = 14400 network buffers.
See
https://ci.apache.org/projects/flink/flink-docs-master/setup/config.html#background
So we
Hi Anton,
I would suggest you simply put your moving average code in a
MapFunction where you can keep track of the current average using a
class field.
Cheers,
Max
On Fri, Jun 24, 2016 at 10:05 PM, Anton wrote:
> Hello
>
> I'm currently trying to learn Flink. And so far am
Hi Saliya,
Thanks for your question. Flink's type analyzer couldn't extract the
type information. You may implement the ResultTypeQueryable interface
in your custom source. That way you can manually specify the correct
type. If that doesn't help you, could you please share more of the
stack
Hi Nick,
SinkFunction just implements user-defined functions on incoming
elements. OutputFormat offers more lifecycle methods. Thus it is a
more powerful interface. The OutputFormat originally comes from the
batch API, whereas the SinkFunction originates from streaming. Those
were more separate
Hi Flavio,
You're right, it is not documented. However, there is a message that
explains that bc is required when you start the cluster. On many
clusters it is preinstalled. Please create a JIRA if you like.
Cheers,
Max
On Fri, Feb 5, 2016 at 5:28 PM, Flavio Pompermaier
Hi Pieter,
Which version of Flink are you using? It appears you've created a
Flink YARN cluster but you can't reach the JobManager afterwards.
Cheers,
Max
On Sat, Feb 6, 2016 at 1:42 PM, Pieter Hameete wrote:
> Hi Robert,
>
> unfortunately there are no signs of what is
2.11(//apache.mirrors.spacedump.net/flink/flink-0.10.1/flink-0.10.1-bin-hadoop27-scala_2.11.tgz),
> Is there a different version of client lib for scala 2.11?
>
>
> Best,
>
> Andrew
>
> On 08 Feb 2016, at 11:30, Maximilian Michels <m...@apache.org> wrote:
>
> Hi Andr
ormat. Having
> two separate class hierarchies is not helpful, hence the adapter. Much of
> this code already exists in the implementation of FileSinkFunction, so the
> project already supports it in a limited way.
>
> On Mon, Feb 8, 2016 at 4:16 AM, Maximilian Michels <m...@apache.org>
Hi David,
Just had a check as well. Can't find a 2.10 Jar in the lib folder.
Cheers,
Max
On Wed, Feb 10, 2016 at 6:17 AM, Chiwan Park wrote:
> Hi David,
>
> I just downloaded the "flink-1.0-SNAPSHOT-bin-hadoop2_2.11.tgz” but there is
> no jar compiled with Scala 2.10.
Hi Flavio,
To address your points:
1) It runs. That's fine.
2) It doesn't work to run a Java 8 compiled Flink job with Java 7
Flink cluster if you use Java 8 non-backwards-compatible features in
your job.
3) I compile Flink daily with Java 8. Also, we have Travis CI tests
which uses OpenJDK and
rget in the maven java compiler.
> If you change them to 1.8 flink-core doesn't compile anymore.
>
> On Thu, Feb 4, 2016 at 4:23 PM, Maximilian Michels <m...@apache.org> wrote:
>>
>> Hi Flavio,
>>
>> To address your points:
>>
>> 1) It runs. That's f
Hi Zhang,
Please have a look here for the 1.0.0-rc2:
Binaries: http://people.apache.org/~rmetzger/flink-1.0.0-rc2/
Maven repository:
https://repository.apache.org/content/repositories/orgapacheflink-1064
Cheers,
Max
On Sat, Feb 27, 2016 at 4:00 AM, #ZHANG SHUHAO#
:
> Hi Max,
> why do I need to register them? My job runs without problem also without
> that.
> The only problem with my POJOs was that I had to implement equals and hash
> correctly, Flink didn't enforce me to do it but then results were wrong :(
>
>
>
> On Wed, Feb 17, 20
Hi Stefano,
The Job should stop temporarily but then be resumed by the new
JobManager. Have you increased the number of execution retries? AFAIK,
it is set to 0 by default. This will not re-run the job, even in HA
mode. You can enable it on the StreamExecutionEnvironment.
Otherwise, you have
tests (I'm not expert enough to not read the docs ;)) but it doesn't
> mention some specific requirement regarding the execution retries, I'll
> check it out, thank!
>
> On Mon, Feb 15, 2016 at 12:51 PM, Maximilian Michels <m...@apache.org> wrote:
>>
>> Hi Stefano,
Hi Flavio,
Stephan was referring to
env.registerType(ExtendedClass1.class);
env.registerType(ExtendedClass2.class);
Cheers,
Max
On Wed, Feb 10, 2016 at 12:48 PM, Flavio Pompermaier
wrote:
> What do you mean exactly..? Probably I'm missing something here..remember
> that
HI Biplob,
Could you please supply some sample code? Otherwise it is tough to
debug this problem.
Cheers,
Max
On Tue, Feb 16, 2016 at 2:46 PM, Biplob Biswas wrote:
> Hi,
>
> No, we don't start a flink job inside another job, although the job creation
> was done in a
Hey Ankur,
If the output after cancelling the job is correct, I assume the
changes haven't been flushed to disk before. For further
investigation, could you share some code with us?
Cheers,
Max
On Wed, Feb 24, 2016 at 8:54 PM, Ankur Sharma
wrote:
> Hey,
>
> I am
0 AM, Thomas Lamirault
> <thomas.lamira...@ericsson.com> wrote:
>>
>> Hi Max,
>>
>> I will try these workaround.
>> Thanks
>>
>> Thomas
>>
>>
>> De : Maximilian Michels [m...@apache.org]
>
Thanks for noticing, Janardhan. Fixed for the next release.
On Fri, Mar 11, 2016 at 6:38 AM, janardhan shetty
wrote:
> Thanks Balaji.
>
> This needs to be updated in the Job.java file of quickstart application.
> I am using 1.0 version
>
> On Thu, Mar 10, 2016 at 9:23 PM,
Maximilian,
>
> I just downloaded the version from your google drive and used that to run my
> test topology that accesses HBase.
> I deliberately started it twice to double the chance to run into this
> situation.
>
> I'll keep you posted.
>
> Niels
>
>
> On
Hi Andrea,
The dashboard is available in both cases. It only shows the job
manager logs. For the task manager, you will have to use the Yarn
commands.
Cheers,
Max
On Wed, Mar 9, 2016 at 12:47 PM, Andrea Sella
wrote:
> Hi,
> I am experimenting the integration between
Great to have this out!
@Radu: You may use this LinkedIn post:
https://www.linkedin.com/groups/7414853/7414853-6113008761373290497
@Igor: Having a one month window should work fine. The CEP library
only keeps track of the current state of the events which enables you
to process large amounts of
Hi Prez,
It appears Spring's Classloader is not set up correctly.
Unfortunately, I'm not familiar with the way Springboot works.
You added flink-connector-kafka-0.9_2.10 but also mentioned you're
using Scala 2.11. That is bound to cause troubles :)
Cheers,
Max
On Thu, Mar 3, 2016 at 8:02 PM,
ry deployed job run as
> the user that ran the start-cluster.sh script (same behavior as running a
> YARN session)? Or users can kinit on each node and then submit jobs that
> will be individually run with their credentials?
>
> Thanks again.
>
> On Wed, Mar 16, 2016 at 10:30 A
Hi Stefano,
You have probably seen
https://ci.apache.org/projects/flink/flink-docs-release-1.0/setup/config.html#kerberos
?
Currently, all nodes need to be authenticated with the Kerberos before
Flink is started (not just the JobManager). Could it be that the
start-cluster.sh script actually is
Hi Gwenhael,
That is not possible right now. As a workaround, you could have three
DataSets that are constructed by reading recursively from each directory
and unify these later. Alternatively, moving/linking the directories in a
different location would also work.
I agree that it would be nice
Sorry. Wrong mailing list...
On Mon, Mar 21, 2016 at 11:47 AM, Maximilian Michels <m...@apache.org> wrote:
> FYI: The Runner registration has been fixed. The Flink runner
> explicitly registers as of [1]. Also, the SDK tries to look up the
> PipelineRunner class in case it has not
On Sat, Mar 19, 2016 at 6:43 PM, Maximilian Michels <m...@apache.org> wrote:
> Great to see such a lively discussion here.
>
> I think we'll support sinks through the Write interface (like in
> batched execution) and also have a dedicated wrapper for the Flink
> sinks. Th
Hi Stefano,
That is currently a limitation of the Kerberos implementation. The
Kerberos authentication is performed only once the Flink cluster is
brought up. The Yarn session is then tight to a particular user's
ticket. Note, that you need at least Hadoop version 2.6.1 or higher to
run
n!
>>
>> On Mon, Mar 7, 2016 at 2:38 PM, Maximilian Michels <m...@apache.org> wrote:
>>>
>>> Hi Stefano,
>>>
>>> Essentially the Yarn Session is not much different from a per-job Yarn
>>> cluster. In either case, a Flink cluster
Hi Balaji,
You forgot to execute your Flink program using
env.execute();
Cheers,
Max
On Wed, Mar 2, 2016 at 1:36 PM, Balaji Rajagopalan
wrote:
> def main(args: Array[String]): Unit = {
>
>
>
> val env: StreamExecutionEnvironment =
>
Hi Ufuk,
Thanks for updating the page. The "latest documentation" points to the
page itself and not the documentation. I've fixed that and added the
slides from Big Data Warsaw.
Cheers,
Max
On Mon, Apr 4, 2016 at 12:09 PM, Ufuk Celebi wrote:
> @Paris: Just added it. Thanks for
Made a few suggestions. Reads well, Till!
On Mon, Apr 4, 2016 at 10:10 AM, Ufuk Celebi wrote:
> Same here.
>
> +1 to publish
>
> On Mon, Apr 4, 2016 at 10:04 AM, Aljoscha Krettek wrote:
>> Hi,
>> I like it. Very dense and very focused on the example but I
Hi Patcharee,
What do you mean by "nothing happened"? There is no output? Did you
check the logs?
Cheers,
Max
On Thu, Apr 28, 2016 at 12:10 PM, patcharee wrote:
> Hi,
>
> I am testing the streaming wiki example -
>
I have to warn you that the Storm SpoutWrapper and the TwitterSource
are currently the only stoppable sources. However, we could make more
stoppable, e.g. the KafkaConsumer.
On Tue, Apr 19, 2016 at 12:38 AM, Robert Schmidtke
wrote:
> I'm on 0.10.2 which seems to be still
ped on me is the allocation of memory for jni libs. I
> do use a native library in my application, which is likely the culprit. I
> need to account for its memory footprint when doing my memory calculations.
>
> Thanks,
> Timur
>
>
> On Mon, Apr 25, 2016 at 10:28 AM, Maximil
Hi Kaniska,
I've replied to your mail on the Beam user mailing list.
Cheers,
Max
On Wed, Apr 27, 2016 at 4:56 PM, kaniska Mandal
wrote:
> I am facing some issues while reading / writing Avro data.
>
> Attached here the corresponding files and avro-generated pojo.
>
>
Hi Timur,
Which version of Flink are you using? Could you share the entire logs?
Thanks,
Max
On Mon, Apr 25, 2016 at 12:05 PM, Robert Metzger wrote:
> Hi Timur,
>
> The reason why we only allocate 570mb for the heap is because you are
> allocating most of the memory as off
ID = jobRes.getJobID();
> } catch (ProgramInvocationException ex) {
> Logger.getLogger(YarnRunner.class.getName()).log(Level.SEVERE, null,
> }
>
>
> Thanks,
> Theofilos
>
>
> On 2016-04-22 16:05, Maximilian Michels wrote:
>
> Hi Theofilos,
>
> Assuming
Hi Simon,
You'll have to write the property file to disk first to load it using
the ParameterTool.fromPropertiesFile method.
For example:
// copy config from Java resource to a file
File configOnDisk = new File("/path/to/config.properties");
ers
> Simon
>
>
> On 23 May 2016, at 16:30, Stefano Baghino <stefano.bagh...@radicalbit.io>
> wrote:
>
> Are you using Maven to package your project? I believe the resources
> plugin[1] can suit your needs.
>
> [1]:
> http://maven.apache.org/plugins/maven-resource
What error do you get when you don't register the Kryo serializer?
On Mon, May 23, 2016 at 11:57 AM, Flavio Pompermaier
wrote:
> With this last settings I was able to terminate the job the second time I
> retried to run it, without restarting the cluster..
> If I don't
Hi Josh,
1) Use a RichFunction which has an `open()` method to load data (e.g. from
a database) at runtime before the processing starts.
2) No that's fine. If you want your Rest API Sink to interplay with
checkpointing (for fault-tolerance), this is a bit tricky though depending
on the
void open(Configuration parameters) throws Exception {
> parameters.getInteger("myInt", -1);
> // .. do
>
>
>
> Cheers
> Simon
>
> On 23 May 2016, at 14:01, Maximilian Michels <m...@apache.org> wrote:
>
> Hi Simon,
>
> A
Hi David,
I'm afraid Flink logs all exceptions. You'll find the exceptions in the
/log directory.
Cheers,
Max
On Mon, May 23, 2016 at 6:18 PM, David Kim
wrote:
> Hello!
>
> Just wanted to check up on this. :)
>
> I grepped around for `log.error` and it *seems*
This should be resolved according to Apache Infra.
On Tue, May 17, 2016 at 11:28 PM, Henry Saputra wrote:
> Looks like it has been resolved, Could you try it again?
>
> On Tue, May 17, 2016 at 7:02 AM, Stefano Baghino
> wrote:
>>
>> I
Hi Flavio,
These error messages are quite odd. Looks like an off by one error in the
serializer/deserializer. Must be somehow related to the Kryo serialization
stack because it doesn't seem to occur with Flink's serialization system.
Does the job run fine if you don't register the custom Kryo
Hi Simon,
As Aljoscha said, the best way is to supply the configuration as class
fields. Alternatively, if you overload the open(..) method, it should
also show up in the Properties/Configuration tab on the Web interface.
Cheers,
Max
On Mon, May 23, 2016 at 11:43 AM, simon peyer
gt; job to be running constantly.
>
> Josh
>
> On Mon, May 23, 2016 at 5:56 PM, Maximilian Michels <m...@apache.org> wrote:
>>
>> Hi Josh,
>>
>> 1) Use a RichFunction which has an `open()` method to load data (e.g. from a
>> database) at runtime before t
Hi Ken,
When you're running Yarn, the Flink configuration is created once and
shared among all nodes (JobManager and TaskManagers). Please have a
look at the JobManager tab on the web interface. It shows you the
configuration.
Cheers,
Max
On Fri, Apr 29, 2016 at 3:18 PM, Ken Krugler
Hi Theofilos,
I'm not sure whether I understand correctly what you are trying to do.
I'm assuming you don't want to use the command-line client.
You can setup the Yarn cluster in your code manually using the
FlinkYarnClient class. The deploy() method will give you a
FlinkYarnCluster which you
Hi Stefano,
Thanks for reporting. I wasn't able to reproduce the problem. I ran
./bin/yarn-session.sh -n 1 -s 2 -jm 2048 -tm 2048 on a Yarn cluster
and it created a Flink cluster with a JobManager and a TaskManager
with two task slots. By the way, if you omit the "-s 2" flag, then the
default is
Hi!
In the latest master and in the upcoming 1.1, all files in the lib
folder will be shipped to the Yarn cluster and added to the class
path. In Flink version <= 1.0.x no files will be added to the ship
files by default (only the flink-dist*.jar will be shipped).
Regardless of the version, if
Hi Konstantin,
If you come from traditional on-premise installations it may seem
counter-intuitive to start a Flink cluster for each job. However, in
today's cluster world it is not a problem to request containers on
demand and spawn a new Flink cluster for each job. Per job clusters
are
Hi Robert,
>Unfortunately, during the Flink map and reduce phases the objects change
>their hash codes and become inconsistent with the keys of the original hashmap
If objects change their hash code values, then this means they are not
equal anymore. If this is not desired then the
Hi Vishnu Viswanath,
The keyed elements are spread across the 50 task slots (assuming you
have a parallelism of 50) using hash partitioning on the keys. Each
task slot runs one or multiple operators (depending on the slot
sharing options). One of them is a WindowOperator which will decide
when to
Hi Jack,
Considering the type of job you're running, you shouldn't run out of
memory. Could it be that the events are quite large strings? It could
be that the TextOutputFormat doesn't write to disk fast enough and
accumulates memory. Actually, it doesn't perform regular flushing
which could be
Hi Sameer,
That depends on the time characteristic you have chosen. If you have
set it to event time [1] then it will use event time, otherwise the
default is to use processing time.
When using event time, the element's timestamp is used to assign it to
the specified time windows in the
Hi Sameer,
If you use Event Time you should make sure to assign Watermarks and
Timestamps at the source. As you already observed, Flink may get stuck
otherwise because it waits for Watermarks to progress in time.
There is no timeout for windows. However, you can implement that logic
in your
Hi David,
You're starting two executions at the same time (in different
threads). Here's why:
Execution No 1
DataStreamUtils.collect(..) starts a Thread which executes your job
and collects stream elements. It runs asynchronously. The collect(..)
method returns after starting the thread.
You're clearing the "handState" on "GameEndHistory". I'm assuming this
event comes in before "CommCardHistory" where you check the state.
On Fri, Aug 12, 2016 at 6:59 AM, Dong-iL, Kim wrote:
> in my code, is the config of ExecutionEnv alright?
>
>
>> On Aug 11, 2016, at 8:47
Hi Philippe,
There is no particular reason other than hash partitioning is a
sensible default for most users. It seems like this is rarely an
issue. When the number of keys is close to the parallelism, having
idle partitions is usually not a problem due to low data volume. I see
that it could be
101 - 200 of 280 matches
Mail list logo