[jira] [Comment Edited] (GEARPUMP-364) Heartbeat interval grow too large

["Pan Xue (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/GEARPUMP-364?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16299475#comment-16299475 ] 

Pan Xue edited comment on GEARPUMP-364 at 12/21/17 3:16 AM:
------------------------------------------------------------

When I call the CEP library (esper) method in the onNext method of the task to process the data, esper has a queue and multiple threads inside it to process the data. Apparently the gearpump will have a significant back pressure. The way I call esper is Asynchronous, onNext method does not block, and then the data back pressure is fantastic, see the processor on the page Average Message Receive Latency value is large

[^aaa.png]



was (Author: xxh2000):
When I call the CEP library (esper) method in the onNext method of the task to process the data, esper has a queue and multiple threads inside it to process the data. Apparently the gearpump will have a significant back pressure. The way I call esper is Asynchronous, onNext method does not block, and then the data back pressure is fantastic, see the processor on the page Average Message Receive Latency value is large



> Heartbeat interval grow too large 
> ----------------------------------
>
>                 Key: GEARPUMP-364
>                 URL: https://issues.apache.org/jira/browse/GEARPUMP-364
>             Project: Apache Gearpump
>          Issue Type: Bug
>    Affects Versions: 0.8.4
>            Reporter: Pan Xue
>         Attachments: aaa.png, gear.png
>
>
> I use the thread pool on the downstream onNext, which has 10 threads, when the program is running for some time, the log will appear heartbeat interval is growing too large: 6575 millis, and the amount of upstream data back pressure, from tens of thousands per Seconds down to a few thousand per second



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)