Hi Fernando,
How much data are you trying to write? If you just use single messages for
testing, it could be that the default bulk settings are not working well.
If so, could you please adjust the following settings and report back?
public enum SinkOption {
BULK_FLUSH_MAX_ACTIONS,
Hi Norm,
the error message already points to the main issue: your property names are
not correct.
*Unsupported property keys: drivername update-mode password dburl username*
You should use the builder to properly configure the sink [1].
[1]
Hi David,
the obvious reason is that your state stored an enum value that is not
present anymore. It tries to deserialize the 512. entry in your enum that
is not available.
However, since it's highly unlikely that you actually have that many enum
values in the same enum class, we are actually
Hi Lei,
Currently, Flink SQL doesn't support to register a binlog format (i.e. just
define "order_id" and "order_no", but the json schema has other binlog
fields).
This is exactly what we want to support in FLIP-105 [1] and FLIP-95.
For now, if you want to consume such json data, you have to
If all the rest api could be viewed successfully, then the reason may be js
cache.
You could try to force a refresh(e.g. Cmd+Shft+R for Mac). It solved my
problem before.
Best,
Yang
pwestermann 于2020年3月4日周三 下午8:40写道:
> We recently upgraded from Flink 1.7 to Flink 1.9.2 and the history server
能不能先用doAs的方式来试一下,比如注册HiveCatalog的部分在UserGroupInformation.getLoginUser().doAs()里做,排查下是不是HiveMetaStoreClient没有用上你登录用户的信息。
另外你的hive版本是2.1.1么?从stacktrace上来看跟2.1.1的代码对不上,比如
HiveMetaStoreClient.java的第562行:
Hi Zhu Zhu and Abhinav,
I am able to verify the recovery from checkpoint based on your suggestions,
thanks a lot for the help!
Eleanore
On Wed, Mar 4, 2020 at 5:40 PM Bajaj, Abhinav
wrote:
> I implemented a custom function that throws up a runtime exception.
>
>
>
> You can extend from simpler
Great exploration. And thanks for your information.
I believe you have a deep understanding of Flink's internal mechanism.
Best,
Jingsong Lee
On Thu, Mar 5, 2020 at 12:09 PM faaron zheng wrote:
> I finally got through the runtimefilter in 1.10, the reason why it didn't
> call commit method is
I finally got through the runtimefilter in 1.10, the reason why it didn't
call commit method is in OperatorCodeGenerator. It should call endInput()
method correctly in generateOneInputStreamOperator. A complete process of
runtimefilter is: 1.Add insert and remove rule in batch rules. 2.Build side
I want to rigister a table from mysql binlog like this:
tEnv.sqlUpdate("CREATE TABLE order(\n"
+ "order_id BIGINT,\n"
+ "order_no VARCHAR,\n"
+ ") WITH (\n"
+ "'connector.type' = 'kafka',\n"
...
+ "'update-mode' = 'append',\n"
+ "
Hi Abhinav,
Do you mind sharing the complete 'jobmanager.log'?
org.apache.flink.runtime.jobmaster.slotpool.SlotPool - Cannot serve slot
> request, no ResourceManager connected.
>
Sometimes you see this log because the ResourceManager is not yet connect
when the slot request arrives the
I implemented a custom function that throws up a runtime exception.
You can extend from simpler MapFunction or more complicated
RichParallelSourceFunction depending on your use case.
You can add logic to throw a runtime exception on a certain condition in the
map or run method. .
Hi Eleanore,
You can change your application tasks to throw exceptions in a certain
frequency.
Alternatively, if the application has external dependencies (e.g. source),
you can trigger failures manually by manipulating the status of the
external service (e.g. shutdown the source service, or
Hi Eleanore,
You can change your application tasks to throw exceptions in a certain
frequency.
Alternatively, if the application has external dependencies (e.g. source),
you can trigger failures manually by manipulating the status of the
external service (e.g. shutdown the source service, or
Hey Kostas,
We’re a little bit off from a 1.10 update but I can certainly see if that
CompressWriterFactory might solve my use case for when we do.
If there is anything I can do to help document that feature, please let me
know.
Thanks!
Austin
On Wed, Mar 4, 2020 at 4:58 AM Kostas Kloudas
Hi,
I have a flink application and checkpoint is enabled, I am running locally
using miniCluster.
I just wonder if there is a way to simulate the failure, and verify that
flink job restarts from checkpoint?
Thanks a lot!
Eleanore
Hi,
I have a flink application and checkpoint is enabled, I am running locally
using miniCluster.
I just wonder if there is a way to simulate the failure, and verify that
flink job restarts from checkpoint?
Thanks a lot!
Eleanore
Hello,
I have this Exception in my datastream app and I can't find the root cause.
I consume data from Kafka and it fails when I try to get a value from my
MapState in RocksDB.
It was working in previous release of my app but I can't find the cause of this
error.
Same error with this change:
public class Teradata extends ConnectorDescriptor {
/**
* Constructs a {@link ConnectorDescriptor}.
*/
public Teradata() {
super("jdbc", 1, false);
}
@Override
protected Map toConnectorProperties() {
Map map = new
While I setup to reproduce the issue with debug logs, I would like to share
more information I noticed in INFO logs.
Below is the sequence of events/exceptions I notice during the time zookeeper
was disrupted.
I apologize in advance as they are a bit verbose.
* Zookeeper seems to be down
Thanks Xintong for pointing that out.
I will dig deeper and get back with my findings.
~ Abhinav Bajaj
From: Xintong Song
Date: Tuesday, March 3, 2020 at 7:36 PM
To: "Bajaj, Abhinav"
Cc: "user@flink.apache.org"
Subject: Re: JobMaster does not register with ResourceManager in high
I killed the task manager and job manager forcefully by the kill -9 command
and while recovering
I am checking the flag returned by the isRestored method in the
Intializestate function.
anyways I figured the issue and fixed it thanks for the support.
On Tue, Mar 3, 2020 at 7:24 PM Gary Yao
Hi Jacopo,
to prevent type erasure in Java, you need to create a sub-type that
contains only reified types.
Instead of using a generic type with bound variables in
stream.process(new MyKeyedBroadcastProcessFunction());
you can use
stream.process(new MyKeyedBroadcastProcessFunction()
Hi Yu,
are you using incremental checkpoints [1]? If so, then the smaller
checkpoints would be the deltas and the larger the complete state.
[1]
https://flink.apache.org/features/2018/01/30/incremental-checkpointing.html
On Wed, Mar 4, 2020 at 6:41 PM Yu Yang wrote:
> Hi all,
>
> We have a
Kostas and Gordon,
Thanks for the suggestions! I'm on RocksDB. We don't have that setting
configured so it should be at the default 1024b. This is the full "state.*"
section showing in the JobManager UI.
[image: Screen Shot 2020-03-04 at 9.56.20 AM.png]
Jacob
On Wed, Mar 4, 2020 at 2:45 AM
Hi all,
We have a flink job that does check-pointing per 10 minutes. We noticed
that for the check-points of this job, the _metadata file size can vary a
lot. In some checkpoint, we observe that _metadata file size was >900MB,
while in some other check-points of the same job, the _metadata file
Have you tried to just export Hadoop 3's classpath to `HADOOP_CLASSPATH`
and see if that works out of the box?
If the main use case is HDFS access, then there is a fair chance it might
just work, because Flink uses only a small subset of the Hadoop FS API
which is stable between 2.x and 3.x, as
Using Flink 1.10 and coding in Java 11, is it possible use to write to Teradata
in append mode? MySQL, PostgreSQL, and Derby are the only supported drivers
listed. Thanks.
https://ci.apache.org/projects/flink/flink-docs-stable/dev/table/connect.html#connectors
I created the ConnectorDescriptor
**
We apologize for cross-postings.
We appreciate your great help in forwarding this CFP to your
colleagues and friends.
Hi all,
Yes my problem is that I do not create the function inline but create a
function directly when creating the data stream job.
My code (which I cannot share) is exactly like your example, Yun, are you aware
if there is a way to prevent code erasure?
Kind regards,
Jacopo Gobbi
From:
Thank you guys. So I have no idea of why data is not being pushed to
Elasticsearch… ☹
My complete code is at
https://stackoverflow.com/questions/60512064/flink-is-not-adding-any-data-to-elasticsearch-but-no-errors
Btw, for some reason I still need to pass .documentType to the Elasticsearch
Hi!
Initially we were looking at 2) but 1) would be the best solution. I think
both are would be very valuable.
My only concern related to using the Schema Registry as a Catalog is the
interaction with other Catalogs in the system. Maybe you are using a Hive
catalog to track a bunch of tables,
Yes. From my perspective, deriving schema from schema registry is the most
important use case of FLINK-16420.
Some initial idea about this:
1) introduce a SchemaRegisteryCatalog to allow users run queries on
existing topics without manual table definition. see FLINK-12256
2) provide a connector
We recently upgraded from Flink 1.7 to Flink 1.9.2 and the history server UI
now seems to be broken. It doesn't load and always just displays a blank
screen.
The individual endpoints (e.g. /jobs/overview) still work.
Could this be an issue caused by the Angular update for the regular UI?
--
Hi,
Without knowing too much about flink serialization, I know that Flinks states
that it serializes POJOtypes much faster than even the fast Kryo for Java. I
further know that it supports schema evolution in the same way as avro.
In our project, we have a star architecture, where one flink
Hi Jark,
Thank you for the clarification this is exactly what I was looking for,
especially for the second part regarding schema registry integration.
This question came up as we were investigating how the schema registry
integration should look like :)
Cheers,
Gyula
On Wed, Mar 4, 2020 at
Hi Gyula,
That's a good point and is on the roadmap.
In 1.10, JSON and CSV format can derive format schema from table schema. So
you don't need to specify format schema in properties anymore if you are
using 1.10.
On the contrary, we are planning to derive table schema from format schema
if it
Hi Jacob,
Apart from what Klou already mentioned, one slightly possible reason:
If you are using the FsStateBackend, it is also possible that your state is
small enough to be considered to be stored inline within the metadata file.
That is governed by the "state.backend.fs.memory-threshold"
我的需求是2,现在我使用的是execEnv.createInput(inputFormat()),
我先去试试 env.addSource(new InputFormatSourceFunction(..)...)。
多谢~
原始邮件
发件人:"JingsongLee"< lzljs3620...@aliyun.com.INVALID ;
发件时间:2020/3/4 17:40
收件人:"user-zh"< user-zh@flink.apache.org ;
主题: Re: flink 1.8 内的StreamExecutionEnvironment 对于
Hi Weike,
You are right. It is not needed since Kafka 0.9+. We already have an issue
to make it optional.
See https://issues.apache.org/jira/browse/FLINK-16125. We are planning to
fix it in 1.10.1 too.
Best,
Jark
On Wed, 4 Mar 2020 at 18:23, Weike Dong wrote:
> Hi,
>
>
>
> Recently I have
Hi,
Recently I have found that in the Flink Kafka Connector, the parameter
"connector.properties.zookeeper.connect" is made mandatory for users.
Therefore without it, Flink would throw an exception saying "Caused by:
org.apache.flink.table.api.ValidationException: Could not find required
Hi All!
I am wondering if it would be possible to change the CREATE TABLE statement
so that it would also work without specifying any columns.
The format generally defines the available columns so maybe we could simply
use them as is if we want.
This would be very helpful when exploring
Hi Jacob,
Could you specify which StateBackend you are using?
The reason I am asking is that, from the documentation in [1]:
"Note that if you use the MemoryStateBackend, metadata and savepoint
state will be stored in the _metadata file. Since it is
self-contained, you may move the file and
Hi Austin,
I will have a look at your repo. In the meantime, given that [1] is already
merged in 1.10,
would upgrading to 1.10 and using the newly introduced
CompressWriterFactory be an option for you?
It is unfortunate that this feature was not documented.
Cheers,
Kostas
[1]
Hi,
你的需求是什么?下列哪种?
- 1.想用unbounded source,continuous的file source,监控文件夹,发送新文件,且需要支持多文件夹
- 2.只是想用bounded的input format,需要支持多文件
如果是1,现在仍然不支持。
如果是2,那你可以用env.addSource(new InputFormatSourceFunction(..)...)来支持多文件。
Best,
Jingsong Lee
--
Hi All,
If you have ever touched the docker topic in Flink, you
probably noticed that we have multiple places in docs and repos which
address its various concerns.
We have prepared a FLIP [1] to simplify the perception of docker topic in
Flink by users. It mostly advocates for an approach of
我在使用flink 1.8 自定义 FileInputFormat
的时候遇到了不兼容问题,初步看了源代码,不知道自己得出的结论是否正确,并想了解一下后续趋势和进展,麻烦各位大佬抽空指点一下,先行感谢~~
问题1: StreamExecutionEnvironment 为什么要做这样的限制?ContinuousFileMonitoringFunction
的作用是什么?
相关的代码描述如下
StreamExecutionEnvironment 对 FileInputFormat 对象有特殊的处理逻辑
if (inputFormat instanceof
47 matches
Mail list logo