[ 
https://issues.apache.org/jira/browse/NIFI-6488?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16940112#comment-16940112
 ] 

Seokwon Yang commented on NIFI-6488:
------------------------------------

We are seeing the same UI unresponsiveness or UI not available (with 
SocketTimeoutException).  In our flow, we heavily use the PutAzureStorage 
processor to upload flow files to azure clod blog storage, and we are seeing 
the throttling on the azure side when UI is not available.

We increase the timeout parameters and reduce the thread count as suggested in 
multiple blogs, but when we load data heavily on our cluster, we starts to see 
this issue again and again.

 

It would be nice if NIFI UI does not get affected by the data flow and and run 
in a separate JVM like nifi bootstrap process.    Any plan to do separate out 
UI from data flow execution engine?

 

> NiFi web interface often non-responsive 
> ----------------------------------------
>
>                 Key: NIFI-6488
>                 URL: https://issues.apache.org/jira/browse/NIFI-6488
>             Project: Apache NiFi
>          Issue Type: Bug
>    Affects Versions: 1.9.2
>         Environment: centos 7
>            Reporter: David Sargrad
>            Priority: Major
>         Attachments: image-2019-08-28-08-15-51-702.png, 
> image-2019-08-30-08-03-39-877.png
>
>
> I have been using NiFi for close to a year. Currently most of my experience 
> has been with building custom processors and prototyping a variety of flows. 
>  
> Often the NiFi web interface becomes non-responsive. Its typically during a 
> flow integration phase, where I leave flowfiles sitting in queues, sometimes 
> 3 or 4 queues may be maxed at 10,000 flow files. These are relatively small 
> flowfiles (perhaps 200 or 300 bytes). Yet, of course, the sum total of the 
> filled queues is in the tens of megabytes.
>  
> I am guessing that the JVM is nearing, or perhaps has exceeded, its maximum 
> allocated heap size. I'd think that NiFi protects against non-responsive UI's 
> by shedding load, presenting a warning, or doing something to preserve its 
> operational state. 
>  
> What is the best way for me to isolate this issue? I'd like to understand 
> exactly what drives NiFi to be non-responsive. If there is anything that I 
> can do to isolate this, and to help the NiFi team make this more robust, I'm 
> glad to dig.
>  
> Its been impossible for me to exactly characterize the conditions that lead 
> to a non-responsive NiFi UI. However I see this fairly frequently. 
>  
> I've updated the following picture. Since it did not seem to attach before. 
> I've also added a view of the memory and CPU consumption. In this particular 
> case I waited for about 30 seconds and eventually the NiFi user interface 
> came to life.
> !image-2019-08-28-08-15-51-702.png!



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to