Thank you both Andrew and Pierre. I think I’ll try both approaches on our
integration system to see how each behaves in our typical conditions for
throughput. Cheers, Jim
On Fri, Nov 13, 2020 at 2:50 PM Pierre Villard
wrote:
> Hi,
>
> You can configure your relationship before the ExecuteScript
Hi,
You can configure your relationship before the ExecuteScript processor to
use the load balance strategy to "single node" so that all the data gets to
one node (not necessarily the primary node so you'd still keep the
processor configured to be executed on all nodes).
Pierre
Le ven. 13 nov. 2
I think a better design is for every node to write to this network share
with some form of a partition (node) id in the filename. In a 5 node
cluster you will have 5 parts.
Next, either query over this directory directly with your engine of choice.
Reducing all traffic to 1 node will be a bottlen
My flow files are distributed across my cluster nodes. I am using an
ExecuteScript processor running a python script to write custom log
messages to a log file I maintain. It appears that i am losing records when
all the nodes are attempting to write to the one log file that exists on a
common netw
