nifi-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Mark Payne <marka...@hotmail.com>
Subject Re: 3 of 7 Expand all Print all In new window SiteToSiteProvenanceReportingTask reports no data when event type filter is defined
Date Fri, 06 Oct 2017 18:54:57 GMT
Fred, Drew,

Just to close the loop here: I was able to recreate and understand the issue. It turns out
that if a batch of events are read
(default batch size of 1,000) and none of the events read match the filter, the reporting
task would stop advancing through
the events and would continually just read those same 1,000 events, none of which would match.
I submitted a PR to address
this.

Thanks
-Mark

On Oct 6, 2017, at 3:43 AM, Fred S <fs9387287@gmail.com<mailto:fs9387287@gmail.com>>
wrote:

Thanks for looking into the issue and sort of helping me rule out that I'm way of doing something
wrong here.

I've created an issue in JIRA: https://issues.apache.org/jira/browse/NIFI-4468

Thanks,
-Fred

On Thu, Oct 5, 2017 at 5:27 PM, Andrew Lim <andrewlim.apache@gmail.com<mailto:andrewlim.apache@gmail.com>>
wrote:
I did some more testing and found that if I restart the NiFi instance that is running the
reporting task, I start getting provenance events again in my other instance.

If I stop the reporting task, add the DOWNLOAD filter, restart the task and then generate
only download events, those download events do show up in my other instance.  But if I generate
different types of provenance events (RETRIEVE, DOWNLOAD, etc.), the latest download events
are not showing up again.

-Drew


> On Oct 5, 2017, at 11:02 AM, Andrew Lim <andrewlim.apache@gmail.com<mailto:andrewlim.apache@gmail.com>>
wrote:
>
> I had a SiteToSiteProvenanceReportingTask running in 1.3.0, so I gave this a try myself.
>
> I think I’m able to reproduce the issue that Fred is seeing.  But I’m not sure if
the issue is only related to adding an Event Type filter to the reporting task.
>
> With no filter, I generated some DOWNLOAD events by downloading some FlowFiles.  I did
see these come into my NiFi instance that handles the events.  Everything was working as expected.
>
> But then I stopped the reporting task, added the DOWNLOAD filter and then restarted the
task.  Then I downloaded some FlowFiles again.  As Fred was seeing, these events did not come
in.
>
> However, then I stopped the reporting task, removed the filter, and then restarted the
task.  Then I generated numerous provenance events (RETRIEVE, DOWNLOAD, etc).  But no events
came in.
>
> So maybe this has something to do with starting/stopping the reporting task in addition
to adding the filter.
>
> Fred, can you file a Jira?  I can add my comments to it.
>
> Thanks,
>
> -Drew
>
>
>> On Oct 5, 2017, at 9:51 AM, Mark Payne <markap14@hotmail.com<mailto:markap14@hotmail.com>>
wrote:
>>
>> Hi Fred,
>>
>> I have used the reporting task with specific event types listed, but I've not run
into issues personally.
>>
>> I am curious if you have tried looking for more common event types, such as RECEIVE,
CONTENT_MODIFIED, or ATTRIBUTES_MODIFIED?
>> EXPIRE events only fire when FlowFiles get aged off from a Connection due to not
being processed within the
>> Connection's "FlowFile Expiration" threshold, and DOWNLOAD events are only triggered
when a user downloads
>> the contents of the FlowFile to look at it manually. So both of these are fairly
rare events.
>> RECEIVE, ATTRIBUTES_MODIFIED, and CONTENT_MODIFIED, on the other hand, will generally
happen constantly.
>>
>> Thanks
>> -Mark
>>
>>
>>
>>> On Oct 4, 2017, at 4:50 PM, Fred S <fs9387287@gmail.com<mailto:fs9387287@gmail.com>>
wrote:
>>>
>>> Dear community.
>>>
>>> I'm currently trying to set up a SiteToSiteProvenanceReportingTask to keep track
of DOWNLOAD provenance events.
>>>
>>> If I set it up without any filters at all everything gets sent to my dedicated
NiFi instance for handling these messages. When I set a filter in the "Event Type" property
(I've tried "DOWNLOAD" and "DOWNLOAD, EXPIRE") - nothing.
>>>
>>> I've been looking around for anyone experiencing a similar issue, without any
luck. Have anyone else set this up successfully without having to log *all* provenance events?
Due to the number of flowfiles I have flying by each day it would not be feasible to go with
that approach.
>>>
>>> I see this issue on both NiFi 1.3.0 and 1.4.0. For what it's worth I've also
tried to accomplish this both with and without security/ssl.
>>>
>>> If anyone else needs any additional information about my setup, please let me
know.
>>>
>>> All the best,
>>> Fred
>>
>



Mime
View raw message