Very agree with to drop it. +1
--
发件人:Jeff Carter
发送时间:2018年9月29日(星期六) 10:18
收件人:dev
抄 送:chesnay ; Till Rohrmann ; user
主 题:Re: [DISCUSS] Dropping flink-storm?
+1 to drop it.
On Fri, Sep 28, 2018, 7:25 PM Hequn Cheng wrote:
>
I am using flink 1.5.2 in standalone mode. The machine is 8 cores with 32GB
memory. I set slots to 4 or 8. I set most jobs parallelism to 1. And each
machine has jobs less than the slot number.
The load average is too high as the picture shows below. Each line presents
load average 1 on one
Hi,
+1 to drop it. It seems that few people use it.
Best, Hequn
On Fri, Sep 28, 2018 at 10:22 PM Chesnay Schepler
wrote:
> I'm very much in favor of dropping it.
>
> Flink has been continually growing in terms of features, and IMO we've
> reached the point where we should cull some of the
Hi,
+1, I agree.
In addition, some users ask questions about the integration of Storm
compatibility mode with the newer Flink version on the mailing list.
It seems that they are not aware that some of Flink's new features are no
longer available in Storm compatibility mode.
This can be confusing
I am experiencing a rather odd error. We have a job running on a Flink
1.4.2 cluster with two Kafka input streams, one of the streams is processed
by an async function, and the output of the async function and the other
original stream are consumed by a CoProcessOperator, that intern emits
Scala
Hello Kostas,
Thank you very much for the details. Also thanks for that "feel free to
disagree" (however, I don't have any desire to disagree here :) thanks a
lot)
Regarding that mainStream.windowAll, did you mean that checkpointing of the
two branches (the main one and the monitoring one) will
Hi Bill,
I wrote those two medium posts you mentioned above. But clearly, the
techlab one is much better
I would suggest just "close the file when checkpointing" which is the
easiest way. If you use BucketingSink, you can modify the code to make it
work. Just replace the code from line 691 to 693
I'm trying to stream log messages (syslog fed into Kafak) into Parquet
files on HDFS via Flink. I'm able to read, parse, and construct objects for
my messages in Flink; however, writing to Parquet is tripping me up. I do
*not* need to have this be real-time; a delay of a few minutes, even up to
an
Yes.
Thanks for bringing this up Hequn! :-) I think Tuple would not be the best
container to use.
However, in search for alternative, shouldn't Collection / List be a more
suitable solution? Row seems to not fit in the context (as there can be
Rows with elements of different type).
I vaguely
Hi,
I haven't look into the code. If this is limited by Tuple, would it better
to implement it with Row?
Best, Hequn
On Fri, Sep 28, 2018 at 9:27 PM Rong Rong wrote:
> Hi Henry, Vino.
>
> I think IN operator was translated into either a RexSubQuery or a
> SqlStdOperatorTable.IN operator.
> I
Hi,
You can implement TopN on SQL/Table-api or write a datastream job with
ProcessFunction to solve the problem.
Best, Hequn
On Fri, Sep 28, 2018 at 9:38 AM 徐涛 wrote:
> Hi Hequn,
> If limit n is not supported in streaming, how to solve top n problem in
> stream scenario?
>
> Best
> Henry
>
>
I'm very much in favor of dropping it.
Flink has been continually growing in terms of features, and IMO we've
reached the point where we should cull some of the more obscure ones.
flink-storm, while interesting from a theoretical standpoint, offers too
little value.
Note that the bolt/spout
Hi, I am trying to move my job from one cluster to another cluster using
Savepoint. But It's failing while restoring on the new cluster.
In error, it's still trying to connect from some URL of old cluster. I have
checked all the properties and configuration. Is flink save the URL's while
Hi Henry, Vino.
I think IN operator was translated into either a RexSubQuery or a
SqlStdOperatorTable.IN operator.
I think Vino was referring to the first case.
For the second case (I think that's what you are facing here), they are
converted into tuples and the maximum we currently have in Flink
Hi everyone,
I would like to discuss how to proceed with Flink's storm compatibility
layer flink-strom.
While working on removing Flink's legacy mode, I noticed that some parts of
flink-storm rely on the legacy Flink client. In fact, at the moment
flink-storm does not work together with Flink's
What do you think about reverting this change (FLINK-8696), because it is
really hard to debug for users? A problem would be if people now rely on
the second argument being the hostname.
An alternative could be to filter out `cluster` and `local` if they should
appear as second argument. This
Hi Paul,
this looks to me like a Yarn setup problem. Could you check which value you
have set for dfs.namenode.delegation.token.max-lifetime? Per default it is
set to 7 days.
The reason why Flink needs to access HDFS is because the binaries and the
configuration files are stored there for an
Hi Chesnay,
I know it is an issue, And won't be fixed because of window merging feature
in case of session window.
But I am looking if someone has implemented aggregation function using
ProcessFunction and process() method instead of AggregationFunction and
aggregate() method.
I hope you got my
Please see: https://issues.apache.org/jira/browse/FLINK-10250
On 28.09.2018 11:27, vino yang wrote:
Hi Gaurav,
Yes, you are right. It is really not allowed to use RichFunction. I
will Ping Timo, he may give you a more professional answer.
Thanks, vino.
Gaurav Luthra
Hi Henry,
Maybe the number of elements in your IN clause is out of range? Its default
value is 20, you can modify it with this configuration item:
*withInSubQueryThreshold(XXX)*
This API comes from Calcite.
Thanks, vino.
徐涛 于2018年9月28日周五 下午4:23写道:
> Hi,
>
> When I am executing the
Hi Gaurav,
Yes, you are right. It is really not allowed to use RichFunction. I will
Ping Timo, he may give you a more professional answer.
Thanks, vino.
Gaurav Luthra 于2018年9月28日周五 下午4:27写道:
> Hi Vino,
>
> Kindly check below flink code.
>
> package
Hi Vino,
Kindly check below flink code.
package org.apache.flink.streaming.api.datastream.WindowedStream
@PublicEvolving
public SingleOutputStreamOperator
aggregate(AggregateFunction function) {
checkNotNull(function, "function");
if (*function instanceof RichFunction*) {
throw new
Hi,
When I am executing the following SQL in flink 1.6.1, some error throws out
saying that it has a support issue, but when I reduce the number of integers in
the “in” sentence, for example,
trackId in (124427150,71648998) , Flink does not complain anything, so I
wonder is there any
Hi Gaurav,
This is very strange, can you share your code and specific exceptions?
Under normal circumstances, it should not throw an exception.
Thanks, vino.
Gaurav Luthra 于2018年9月28日周五 下午3:27写道:
> Hi Vino,
>
> RichAggregateFunction can surely access the state. But the problem is, In
>
Hi,
I just downloaded flink-1.6.1-bin-scala_2.11.tgz from
https://flink.apache.org/downloads.html and noticed that it fails signature
verification with a
gpg: BAD signature from "Till Rohrmann (stsffap) "
message. The sha512 hash doesn't match either.
I switched to 1.6.0, which verifies OK.
Hi Gaurav,
Why do you think the RichAggregateFunction cannot access the State API?
RichAggregateFunction inherits from AbstractRichFunction (it provides a
RuntimeContext that allows you to access the state API).
Thanks, vino.
Gaurav Luthra 于2018年9月28日周五 下午1:38写道:
> Hi,
>
> As we are aware,
26 matches
Mail list logo