Ignore the issue. I set the concurrency to 2 last time. When I reset it to 1, 
the result is 1, 2, 3, 1, 2, 3. 


Happy New Year!


Kui


------------------ Original ------------------
From:  "oppo";<[email protected]>;
Date:  Mon, Jan 1, 2018 01:51 PM
To:  "oppo"<[email protected]>;

Subject:  Re: Data Distribution in RPG of NiFi Cluster





I just search and find NiFi 1.2 already add batch size in RPG. Finally I find 
it in "Remote Ports" of context menu in RPG. 
I set the batch count with 1 which I assume it will dispatch event one by one. 
But when I list queue, I found the left files were all dispatched to the last 
node "slave3". See the following snapshot: 1,2, 3, 3, 3, 3. 
It is not what I want. I hope it is 1, 2, 3, 1, 2, 3.
Any idea?





------------------ Original ------------------
From:  "oppo";<[email protected]>;
Date:  Mon, Jan 1, 2018 11:12 AM
To:  "users"<[email protected]>;

Subject:  Data Distribution in RPG of NiFi Cluster




Hi, 
 
I have three nodes in NiFi cluster. Slave3 is the primary, slave2 is the 
coordinator.
 
I visit web UI of slave2 node, and create a process group, add listSFTP and RPG 
to do the data distribution. The input port of top diagram will forward to the 
group and link to FetchSFTP. But I found all 6 files from ListSFTP were just 
forwarded to one node slave1. See attachment.
 
I expect 6 files should be forwarded to slave1 and slave2 (or including slave3) 
in round-robin way. 
 
Does anyone know how is going on here?
 
Thanks and Regards.

Kui

Reply via email to