??Timestamp??Watermark
----
??:
"user-zh"
I also tried enable native memory tracking, via jcmd, here is the memory
breakdown: https://ibb.co/ssrZB4F
since job manager memory configuration for flink 1.10.2 only has
jobmanager.heap.size, and it only translates to heap settings, should I
also set -XX:MaxDirectMemorySize and
Hi Team,
for Flink 1.11 Graphite Metrics. I see the following Error in the log.
Any suggestions?
020-10-23 21:55:14,652 ERROR
org.apache.flink.runtime.metrics.ReporterSetup- Could
not instantiate metrics reporter grph. Metrics might not be
exposed/reported.
Changing down to maven 3.2 shows an error. It seems like I'm hitting
flaky tests. I hit one error and then a different error when running again.
I'm not blocked now. My diff was already merged and the related tests
pass. Neither of these failures look related to my diff.
<<< FAILURE! - in
Hi
At Trackunit We have been using Mesos for long time but have now moved to k8s.
Med venlig hilsen / Best regards
Lasse Nedergaard
> Den 23. okt. 2020 kl. 17.01 skrev Robert Metzger :
>
>
> Hey Piyush,
> thanks a lot for raising this concern. I believe we should keep Mesos in
> Flink
Hi Till,
please see the screenshot of heap dump: https://ibb.co/92Hzrpr
Thanks!
Eleanore
On Fri, Oct 23, 2020 at 9:25 AM Eleanore Jin wrote:
> Hi Till,
> Thanks a lot for the prompt response, please see below information.
>
> 1. how much memory assign to JM pod?
> 6g for container memory
Hi Till,
Thanks a lot for the prompt response, please see below information.
1. how much memory assign to JM pod?
6g for container memory limit, 5g for jobmanager.heap.size, I think this is
the only available jm memory configuration for flink 1.10.2
2. Have you tried with newer Flink versions?
I
Hi Till,
thanks a lot for the explanation. Im using Flink 1.10.2 with java 11.
Thanks!
Eleanore
On Fri, Oct 23, 2020 at 4:31 AM Till Rohrmann wrote:
> Hi Eleanore,
>
> if you want to tolerate JM restarts, then you have to enable HA. W/o HA, a
> JM restart is effectively a submission of a new
Hi all,
I plan to run flink jobs as docker containers in a AWS Elastic Container Service. I will have checkpointing enabled where state is stored in a s3 bucket. Each deployment will run in a per-job mode. Are there
any non-obvious downsides to running these jobs with a local
Hey Piyush,
thanks a lot for raising this concern. I believe we should keep Mesos in
Flink then in the foreseeable future.
Your offer to help is much appreciated. We'll let you know once there is
something.
On Fri, Oct 23, 2020 at 4:28 PM Piyush Narang wrote:
> Thanks Kostas. If there's items
Thanks Kostas. If there's items we can help with, I'm sure we'd be able to find
folks who would be excited to contribute / help in any way.
-- Piyush
On 10/23/20, 10:25 AM, "Kostas Kloudas" wrote:
Thanks Piyush for the message.
After this, I revoke my +1. I agree with the previous
Thanks Piyush for the message.
After this, I revoke my +1. I agree with the previous opinions that we
cannot drop code that is actively used by users, especially if it
something that deep in the stack as support for cluster management
framework.
Cheers,
Kostas
On Fri, Oct 23, 2020 at 4:15 PM
Hi folks,
We at Criteo are active users of the Flink on Mesos resource management
component. We are pretty heavy users of Mesos for scheduling workloads on our
edge datacenters and we do want to continue to be able to run some of our Flink
topologies (to compute machine learning short term
I'm learning document
partFlat Aggregate
My code is:
https://paste.ubuntu.com/p/HmB4q2WJSb/
Could you tell me how to registerTableAggregateFunction
Thanks for your help
I'm trying again running the tests, now I have four cores (previously five)
and 12 GB RAM (previously 8 GB). I'm still hit by the OOM killer.
The command I'm running is:
mvn -Dflink.forkCount=1 -Dflink.forkCountTestPackage=1 clean verify
[INFO] BUILD FAILURE
[INFO]
I'm learningGroupBy Window Aggregation from
documenthttps://ci.apache.org/projects/flink/flink-docs-stable/dev/table/tableApi.html
My code is:
https://paste.ubuntu.com/p/GQqR4cqdp6/
pojo is:
https://paste.ubuntu.com/p/CF4yttTGQ4/
I got
A group window expects a time attribute for grouping in a
Glad to hear that you solved your problem. Afaik Flink should not read the
fields of messages and call hashCode on them.
Cheers,
Till
On Fri, Oct 23, 2020 at 2:18 PM Radoslav Smilyanov <
radoslav.smilya...@smule.com> wrote:
> Hi Till,
>
> I found my problem. It was indeed related to a mutable
Thanks for starting this survey Robert! I second Konstantin and Xintong in
the sense that our Mesos user's opinions should matter most here. If our
community is no longer using the Mesos integration, then I would be +1 for
removing it in order to decrease the maintenance burden.
Cheers,
Till
On
+1 for adding a warning about the removal of Mesos support and I would
also propose to state explicitly in the warning the version that we
are planning to actually remove it (e.g. 1.13 or even 1.14 if we feel
it is too aggressive).
This will help as a reminder to users and devs about the upcoming
Hi Averell,
it looks as if the org.apache.avro.Schema$Field contains a field which is
an unmodifiable collection. The Kryo serializer will try to deserialize
this field by creating an unmodifiable collection and then trying to add
the elements into it. This will fail.
I would recommend using the
+1 for adding a warning in 1.12 about planning to remove Mesos support.
With my developer hat on, removing the Mesos support would
definitely reduce the maintaining overhead for the deployment and resource
management related components. On the other hand, the Flink on Mesos users'
voices
Hi Robert,
+1 to the plan you outlined. If we were to drop support in Flink 1.13+, we
would still support it in Flink 1.12- with bug fixes for some time so that
users have time to move on.
It would certainly be very interesting to hear from current Flink on Mesos
users, on how they see the
Hi all,
I wanted to discuss if it makes sense to remove support for Mesos in Flink.
It seems that nobody is actively maintaining that component (except for
necessary refactorings because of interfaces we are changing), and there
are almost no users reporting issues or asking for features.
The
Hi Rado,
it is hard to tell the reason w/o a bit more details. Could you share with
us the complete logs of the problematic run? Also the job you are
running and the types of the state you are storing in RocksDB and use as
events in your job are very important. In the linked SO question, the
Hi Eleanore,
if you want to tolerate JM restarts, then you have to enable HA. W/o HA, a
JM restart is effectively a submission of a new job.
In order to tell you more about the Task submission rejection by the
TaskExecutor, I would need to take a look at the logs of the JM and the
rejecting
Hi Navneeth,
sorry for the late reply. To me it looks as
if
/mnt/checkpoints/150dee2a70cecdd41b63a06b42a95649/chk-52/76363f89-d19f-44aa-aaf9-b33d89ec7c6c
has not been mounted to the EC2 machine you are using to run the job. Could
you try to log in onto the machine when the problem occurs and
Hi Eleanore,
how much memory did you assign to the JM pod? Maybe the limit is so high
that it takes a bit of time until GC is triggered. Have you tried whether
the same problem also occurs with newer Flink versions?
The difference between checkpoints enabled and disabled is that the JM
needs to
你的数据有没有时间属性?可以用时间来判断吧
| |
熊云昆
|
|
邮箱:xiongyun...@163.com
|
签名由 网易邮箱大师 定制
在2020年10月23日 14:21,air23 写道:
你好,
这边发现使用cdc读取mysql ,然后写入mysql会有乱序问题
在上游mysql update一条数据,connert=print是有一条delete 和一条insert的数据,
但是jdbc写入mysql时候 发现mysql有时候是正常的,但是有时候会没有,当把并行度改成1的时候是正常的。
这边怀疑是乱序了,先insert 再delete了。所以导致结果表
Hi Manas,
that is a good point. Feel free to open an issue for this. It is not the
first time that your question appeared on the mailing list.
Regards,
Timo
On 23.10.20 07:22, Manas Kale wrote:
Hi Timo,
I figured it out, thanks a lot for your help.
Are there any articles detailing the
Hi,yujianbo
只要任务结束,不管是cancel、failed、killed都会在history sever展示,
可以先去hdfs查看配置的目录是否存在任务相关的文件夹;也可以尝试重启一下history
server试试。麻烦问一下,你的任务使用什么api写的,以及版本、提交方式?
yujianbo wrote
> 大佬,我发现我配置完后就只能看到完成的任务在history
Hi,zhisheng
1.默认的刷新时间10s以及5s都测试过,真实体验是反应时间有点长,达到分钟级别,猜测这个参数的设置意义不大;
2.其实页面提供了Runing job
List,理论上是可以展示的,如果不能展示,web用的同一套前端代码的话,觉得有点鸡肋。使用起来,目前只能查看job的一些统计信息,如
ck相关,dag相关。
使用Filnk on yarn per job提交方式, 已经启动了yarn
JobHistoryServer,应该是不会产生影响的,除了sql其他api的任务监控正常。
Hi Hailong Wang,
非常感谢您的回答,很好地解决了我的疑问。
经过代码测试,NOW()等函数返回值确实会实时更新。
Best,
Longdexin
--
Sent from: http://apache-flink.147419.n8.nabble.com/
Hi Dan,
afaik Flink does not have a dedicated MongoDB connector (except for the
DataSet API which is rather old). Hence I believe that the 2nd option seems
to be more promising.
Cheers,
Till
On Thu, Oct 22, 2020 at 6:45 AM Dan Hill wrote:
> Has anyone connected these two?
>
> Looking through
Hi Suchithra,
thanks for doing this analysis. I think we should try to upgrade the
affected libraries. I have opened issues to do these changes [1, 2, 3, 4,
5]. In the future, it would be great if you could first reach out to
priv...@flink.apache.org so that we can fix these problems without
Hi Longdexin,
thanks for reaching out to the Flink community. I am pulling in Jark who
might be able to help you with this question.
Cheers,
Till
On Thu, Oct 22, 2020 at 2:56 PM Longdexin <274522...@qq.com> wrote:
> From my point of view, the value of NOW() function in SQL is certain by the
>
大佬,我发现我配置完后就只能看到完成的任务在history sever上面,失败的看不到。现在疑惑的是失败的能不能出现在history server
--
Sent from: http://apache-flink.147419.n8.nabble.com/
Hey there,
I noticed that when building and testing Flink itself, logging seems to be
non-existing or very quiet.
I had a look at the logging conf files (such
as flink-tests/src/test/resources/log4j2-test.properties) and the pattern
seems to be that the logging is turned off in tests. At least
Hi Josson,
Thanks for great investigation and coming back to use. Aljoscha, could you
help us here? It looks like you were involved in this original BEAM-3087
issue.
Best,
Piotrek
pt., 23 paź 2020 o 07:36 Josson Paul napisał(a):
> @Piotr Nowojski @Nico Kruber
>
> An update.
>
> I am able
Hi,
None of the operator is renamed or removed. Testing is carried out with exactly
same binary used with 1.9 and 1.11.2. Checkpoint saved in 1.9 is not being able
to retrieve in 1.11.2
From: Sivaprasanna
Sent: Friday, October 23, 2020 10:57 AM
To: Partha Mishra
Cc: user@flink.apache.org
hi ,all
我的问题解决了. 出现该问题的原因如下:
因为通过堡垒机端口转发, 所以需要在bootstrap.servers 写上所有 kafka borker即可
1. 修改 kafka 外网配置
>> broker1 配置:
>> listeners=PLAINTEXT://xxx-a-1:9092,EXTERNAL://:9797
>> advertised.listeners=PLAINTEXT://xxx-a-1:9092,EXTERNAL://:9797
>>
Hi,
We have a job of Flink 1.11.0 running on YARN that reached FAILED state cause
its jobmanager lost leadership
during a ZK full GC. But after the ZK connection was recovered, somehow the job
was reinitiated again
with no checkpoints found in ZK, and hence used an earlier savepoint to
Hi,
我的做法如我所说,是用反射将parser拿出来的,比较hack但是很简单而且很稳妥
代码差不多就是下面这个样子
Flink version: custom version base on 1.11.x
@PostConstruct
private void setup() throws NoSuchFieldException, IllegalAccessException {
final StreamTableEnvironmentImpl env = (StreamTableEnvironmentImpl)
你好,
这边发现使用cdc读取mysql ,然后写入mysql会有乱序问题
在上游mysql update一条数据,connert=print是有一条delete 和一条insert的数据,
但是jdbc写入mysql时候 发现mysql有时候是正常的,但是有时候会没有,当把并行度改成1的时候是正常的。
这边怀疑是乱序了,先insert 再delete了。所以导致结果表 没有这条数据,请问flink sql 或者flink cdc 怎么保证有序。
我看官方文档[1]应该是设置state.checkpoints.num-retained , 默认是1, 但是设置了没有效果, 官方说默认是1,
但是我发现好像是10 ,
同时我也设置了其他的属性,比如
execution.checkpointing.externalized-checkpoint-retention:
RETAIN_ON_CANCELLATION
是可行,所以我的设置应该没有什么问题
44 matches
Mail list logo