Hi,
Increasing akka’s timeouts is rarely a solution for any problems - it either do
not help, or just mask the issue making it less visible. But yes, it is
possible to bump the limits:
https://ci.apache.org/projects/flink/flink-docs-release-1.3/setup/config.html#distributed-coordination-via-akk
Hi,
There are APIs to convert a dynamic table to retract stream. Is there any way
to construct a "retract" data stream and convert it into table? I want to read
the change log of relational database from kafka, "apply" the changes within
flink( by creating CRow DataStream), register/create a t
Hi Alieh,
Are you able to run the example WordCount application without losing
TaskManagers?
Greg
> On Jan 8, 2018, at 7:48 AM, Alieh Saeedi wrote:
>
> Hey all,
> I have an iterative algorithm implemented in Gelly. As long as I upgraded
> everything to flink-1.3.1 from 1.1.2, the runtime ha
Hello guys,
want to clarify for myself: since flink 1.4.0 allows to use hadoop-free
distribution and dynamic hadoop dependencies loading, I suppose that if to
download hadoop-free distribution, start cluster without any hadoop and
then load any job's jar which has some hadoop dependencies (i
used
Have a set of stateful operators that rely on keyed state. There is substantial
skew between keys (i.e. there will be 100 messages on keys A and B, and 10
messages each on keys C-J), and key selection assignment is dictated by the
needs of my application such that I can't choose keys in a way th
Which dependencies do I need to put in the lib directory? I have the
flink-shaded-hadoop2.jar and parquet-hadoop.jar marked as "provided" and
both are in the lib/ directory and I still get the error? I've also
set classloader.resolve-order: parent-first in my flink-conf.yaml.
val flinkDependencies
Hi!
For the exception in the "local file path" case, we already had a reason:
- You were having Hadoop code in your user code jar. Either not putting
the Hadoop dependency into your jar (rather in Flink's lib directory) or
setting classloading to "parent-first" should do the trick there.
For t
Hi Fabian,
Thanks a lot for the reply. Setting the classloader.resolve-order
configuration seems to have done the trick. For anybody else, having the
same problem as this, this is the config that I set:
/*classloader.resolve-order: parent-first
classloader.parent-first-patterns:
java.;org.apache.f
I'm running a flink-cluster (a mini one with just one node); but the
problem is that my TaskManager can't reach to my JobManager!
Here are logs from TaskManager
...
Trying to register at JobManager akka.tcp://flink@MY_PRIV_IP/user/jobmanager
(attempt 20, timeout: 30 seconds)
Trying to register at
Hi Christophe,
Flink exposes event-time and watermark not only in windows.
The easiest solution would be to use a ProcessFunction [1] which can access
the timestamp of a record.
I would apply a ProcessFunction on a keyed stream (keyBy(id) [2]) and store
the max timestamp per key in a ValueState [
Hi Aniket,
you could try to restore the previous behavior by configuring
classloader.resolve-order: parent-first
in the Flink configuration.
Best, Fabian
2018-01-08 23:09 GMT+01:00 ani.desh1512 :
> *Background:* We have a setup of Flink 1.3.2 along with a secure MAPR
> (v5.2.1) cluster (Flink
Hi,
Till (in CC) might be able to help with Akka related questions.
Best, Fabian
2018-01-08 6:46 GMT+01:00 Hao Sun :
> I am running Flink 1.3.2 in my local docker environment.
>
> I see this error, not sure how to find the root cause.
> I am confused by this error message, why JM is trying to c
Hi everyone,
Let's imagine I have a stream of events coming a bit like this:
{ id: "1", value: 1, timestamp: 1 }
{ id: "2", value: 2, timestamp: 1 }
{ id: "1", value: 4, timestamp: 3 }
{ id: "1", value: 5, timestamp: 2 }
{ id: "2", value: 5, timestamp: 3 }
...
As you can see with the non monoto
Hi Kyle,
I'm not sure I understand the problem. I assume you have one sink for each
Avro type (Kafka topic).
If you have multiple sinks, why is it not possible to configure each one
with the correct Avro schema?
Best, Fabian
2018-01-05 22:11 GMT+01:00 Kyle Hamlin :
> I implemented an Avro to Pa
Thanks Boris.
I've filed FLINK-8391 [1] to extend the documentation.
Best, Fabian
[1] https://issues.apache.org/jira/browse/FLINK-8391
2018-01-05 19:52 GMT+01:00 Boris Lublinsky :
> Thanks
> This was it.
> It would help to have this in documentation along with
> `flink-queryable-state-client`
>
Hi,
I've launched JobManager and TaskManager on DC/OS successfully.
Now I have two new issues:
1) All TaskManagers are scheduled on a single node.
- Is it intended to maximize data locality and minimize network communication
cost?
- Is there an option in Flink to adjust the behavior of JobManag
Hi,
you can implement that with a ProcessFunction [1].
The ProcessFunction would have to compute counts at some granularity
(either event-time or processing-time) of records that are processed by the
ProcessFunction in processElement().
I would do this with a MapState that has a timestamp as key
17 matches
Mail list logo