Hi Henry,
the DATE_FORMAT function is in a very bad state right now. I would
recommend to implement your own custom function right now.
This issue is tracked here:
https://issues.apache.org/jira/browse/FLINK-10032
Regards,
Timo
Am 31.10.18 um 07:44 schrieb 徐涛:
Hi Experts,
I found that DA
I've opened a ticket for this:
https://issues.apache.org/jira/browse/FLINK-10731
On Wed, Oct 31, 2018 at 7:41 AM Hequn Cheng wrote:
> Hi Flavio,
>
> You are right. Avg on dates is not supported. It requires numeric types.
> As a workaround, you can transform the datetime into a numeric type usin
Hi
Let's say we have a job which reads from a Flink kafka source
(FlinkKafkaConsumer011) and commits the offsets on each checkpoint.
When the job is started from an older savepoint, will it take the latest
offsets stored in Kafka for the consumer group or are the offsets taken from
the savepoin
Hi Bend,
The offsets would be restored from the savepoint. Flink would only fallback to
use the offsets on the brokers if there are no offset in its states.
Best,
Paul Lam
> 在 2018年10月31日,17:13,
> 写道:
>
> Hi
> Let’s say we have a job which reads from a Flink kafka source
> (FlinkKafkaCons
Hi Henry,
Thanks for your response. However we don't face this issue during normal
run as we have incremental checkpoints. Only when we try to take savepoint
(which tries to save entire state in one go), we face this problem.
Gagan
On Wed, Oct 31, 2018 at 11:41 AM 徐涛 wrote:
> Hi Gagan,
>
Hi Experts,
When I am running the following SQL in FLink 1.6.2, I got
org.apache.calcite.sql.parser.impl.ParseException
select
BUYER_ID,
AMOUNT,
concat(
from_unixtime(unix_timestamp(CREATE_TIME
Hi Experts,
When I am running the following SQL in FLink 1.6.2, I got
org.apache.calcite.sql.parser.impl.ParseException
select
BUYER_ID,
AMOUNT,
concat(
from_unixtime(unix_timestamp(CREATE_TIME
Hmm, so if there is this code that wants the $SCHEMA field, I wonder how it
ever worked. Did you try the newly rebuilt avro model against the older Flink
version (1.3.2)?
If possible, you could send the code and we could look into it? OR, could you
try running it in a debugger and set a breakpo
Hi Madan,
If your source has a parallelism > 1, then when the CSV file is split, only one
of the operators will get the split with the header row.
So in that case, how would you communicate the column name->index information
to the other operators?
If you force a parallelism of 1 for the sourc
Hi to all,
I'm using Flink 1.6.1 and I get different results when running the same
query on the same static dataset. There are times that I get a 'NaN' as
result of a select field-expression, while other times I get a valid
double. How is this possible?
This seems to happen only when I execute a co
Hi Flavio,
do you execute this query in a batch or stream execution environment?
In any case this sounds very strange to me. But is it guarateed that it
is not the fault of the connector?
Regars,
Timo
Am 31.10.18 um 14:54 schrieb Flavio Pompermaier:
Hi to all,
I'm using Flink 1.6.1 and I g
Hello!
We are migrating the the last 1.6 version and all the jobs seem to work fine,
but when we check individual jobs through the web interface we encounter the
issue that after clicking on a job, either it takes too long to load the
information of the job or it never loads at all.
Has anyone
I read a Parquet file from the filesystem. The input rows are always read
in the same way, but results are different.
My query is very big and maybe this affects somehow the query execution:
UM(CASE WHEN
isComplete(nome,sesso,cfPiva,cognome,immobili,deceduto,dataNascita,luogoNascita,impostePagate,
Hi Gagan
Savepoint would generally takes more time than usual incremental checkpoint,
you could try to increase checkpoint timeout time [1]
env.getCheckpointConfig().setCheckpointTimeout(90);
If you just want to resume from previous job without change the state-backend,
I think you coul
Adding more rows to the dataset lead to a deterministic error. My tests
says that the problem arise when adding the STDDEV_POP to the query..
Do you think it could be possible that there's a concurrency problem in its
implementation?
As far as I know STDDEV_POP is translated into basic aggregate functions
(SUM/AVG/COUNT). But if this error is reproducible in a little test
case, we should definitely track this in JIRA.
Am 31.10.18 um 16:43 schrieb Flavio Pompermaier:
Adding more rows to the dataset lead to a deterministic e
Hello!
We are migrating the the last 1.6.2 version and all the jobs seem to work fine,
but when we check individual jobs through the web interface we encounter the
issue that after clicking on a job, either it takes too long to load the
information of the job or it never loads at all.
Has anyo
Hi Pawel
First of all, I don't think the akka timeout exception has relationship with
checkpoint taking long time. And both RocksDBStateBackend and FsStateBackend
could have the async part of checkpoint, which would upload data to DFS in
general. That's why async part would take more time than
Hi All,
I have a GPS stream consumed by FlinkKafkaConsumer which contains a set of
GPSs of different users. I need to count a number of users per a specific
window of this stream.
Could anyone help me, a part of my code is below
// read data from Kafka
DataStream st
Hi,
So far I have added my keytab and principal in the flink-conf.yaml:
security.kerberos.login.keytab:
security.kerberos.login.principal:
But is there a way that I can add this to the "start script" -> run
yarn-cluster .
Thanks!
Hi Xuefu,
Thanks a lot for driving this big effort. I would suggest convert your
proposal and design doc into a google doc, and share it on the dev mailing
list for the community to review and comment with title like "[DISCUSS] ...
Hive integration design ..." . Once approved, we can document it
Do you mean have these two options as the command line options? If so,
AFAIK, I dont think it's supported now. What do you need it? Thanks.
On Wed, Oct 31, 2018 at 11:43 AM Marke Builder
wrote:
> Hi,
>
> So far I have added my keytab and principal in the flink-conf.yaml:
> security.kerberos.log
getting this error in my job manager too frequently. any help. Thanks!
java.util.concurrent.CompletionException: akka.pattern.AskTimeoutException:
Ask timed out on [Actor[akka://flink/user/jobmanager#1927353472]] after
[1 ms]. Sender[null] sent message of type
"org.apache.flink.runtime.message
Hi Shuiyi,
Good idea. Actually the PDF was converted from a google doc. Here is its link:
https://docs.google.com/document/d/1SkppRD_rE3uOKSN-LuZCqn4f7dz0zW5aa6T_hBZq5_o/edit?usp=sharing
Once we reach an agreement, I can convert it to a FLIP.
Thanks,
Xuefu
Hey Paul. Can you please point me to the code in Flink. Thanks!
--
Sent from: http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/
In running tests of flink jobs we are seeing some that yield really good
performance (2.5M records in minutes) and others that are struggleing to get
past 200k records processed. In the later case there are a large number of
keys, and each key gets state in the form of 3 value states. One hold
I have successfully managed to deploy a Flink cluster in Kubernetes without
JobManager high availability. Everything works great. The moment I enable
high availability, TaskManagers fail to contact the JobManager. My
configurations and logs are below. Can someone point me in the correct
dir
Hi marke,
My advice is not to keep your client connected to JM.
If you expect continuous output, you can sink it out.
In addition, it does not rule out that your JM load is too high, such as
the emergence of full GC and so on.
So, make sure your JM has enough resources to use and monitor it.
Than
Hi Rad,
You can take a look at the group window[1] of SQL. I think it may help you.
Best, Hequn
[1]
https://ci.apache.org/projects/flink/flink-docs-master/dev/table/sql.html#aggregations
On Thu, Nov 1, 2018 at 12:53 AM Rad Rad wrote:
> Hi All,
>
> I have a GPS stream consumed by FlinkKafkaCons
Hi Anil,
Sure. You could have a look at FlinkKafkaConsumerBase.java.
Best,
Paul Lam
> 在 2018年11月1日,03:37,Anil 写道:
>
> Hey Paul. Can you please point me to the code in Flink. Thanks!
>
>
>
> --
> Sent from:
> http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/
Hi Ken,
Yep correct.
Thank you.
On Wed, Oct 31, 2018 at 7:24 PM Ken Krugler
wrote:
> Hi Madan,
>
> If your source has a parallelism > 1, then when the CSV file is split,
> only one of the operators will get the split with the header row.
>
> So in that case, how would you communicate the colum
Thanks Yun for your inputs. Yes, increasing checkpoint helps and we are
able to save save points now. In our case we wanted to increase parallelism
so I believe savepoint is the only option as checkpoint doesn't support
code/parallelism changes.
Gagan
On Wed, Oct 31, 2018 at 8:46 PM Yun Tang wro
32 matches
Mail list logo