Hello Joe,

I am gonna explain you the whole picture of what I got and what I would like to 
have. Right now we I receive a file (In fact I receive several files) in a 
shared net unit called “Z" then manually I move the file to a local folder 
called “data” and finally I execute a batch file, this batch file consume 
(read) the file and after read it move it to another folder called “imported”. 
That’s why I just try to invoke a batch program once the file has been dropped 
on the “data” folder.

I have changed Max attribute length and set it to 256 and also leaved the 
Output destination attribute empty. Even so again I haven’t been able to 
execute the batch file (there is no new file called foo.txt on my desktop).

Have you any idea what I am doing wrong? I am pretty sure should be an easy 
fix. Please fell free to make any comment or suggestion regarding to my case.

Thanks in advance.

> El 2 mar 2016, a las 17:36, Joe Percivall <[email protected]> escribió:
> 
> Hello,
> 
> Welcome to NiFi!
> 
> I just tried running an ExecuteStreamCommand processor with the properties 
> you have (I created a script and modified the paths to point to a folder that 
> exists) and two things jump out. One, the Max attribute length must take an 
> integer. If you set it to be a path the processor will be invalid and you'll 
> see a yellow warning icon in the top left of the processor. This means the 
> processor will not run and you'll see the flowfiles queue up in the 
> relationship preceding it.
> 
> Second, the Output Destination Attribute is only for when you want to output 
> the results of the command to an attribute instead of the content of a new 
> flowfile (useful for running a command to find the character encoding of the 
> contents). Using an integer for the max attribute length I am able to 
> correctly run the script.
> 
> As a helpful hint, you can see the description of a property by hovering over 
> the light blue "?" icon in the configure processor tab. Also you can see the 
> documentation for the processor by right clicking on it and selecting "usage" 
> from the list.
> 
> Also what will you eventually be doing with your script? The way the 
> ExecuteStreamCommand is designed to work is by taking in a FlowFile and then 
> running an external command on it. So you may make your flow more efficient 
> and user friendly by putting the ExecuteStreamCommand between the Get and Put.
> 
> Hope that helps,
> Joe
> - - - - - - 
> Joseph Percivall
> linkedin.com/in/Percivall
> e: [email protected]
> 
> 
> 
> 
> On Sunday, February 28, 2016 4:53 PM, jose antonio rodriguez diaz 
> <[email protected]> wrote:
> Hello All,
> 
> I am just getting started with apache Nifi doing a kind of PoC (Proof of 
> Concept) My DataFlow is compose as follow
> GetFile->PutFile->ExecuteStreamCommand
> 
> The idea is move a file from a folder to another one and then execute an 
> script. The first step (move the file from one side to the other) works 
> perfectly but I haven´t been able to execute the script. The script is very 
> simple I just want to create a file on my desktop.
> 
> the script called script.sh is located on my Desktop ($HOME/Desktop/script.sh)
> 
> #!/bin/bash
> 
> echo "This is a test" >> /Users/joseantoniorodriguez/Desktop/foo.txt
> 
> 
> 
> 
> Also the ExecuteStreamCommand is configured as follow
> 
> Command Path: /Users/joseantoniorodriguez/Desktop/script.sh
> Ignore STDIN: true
> Working directory: /Users/joseantoniorodriguez/Desktop
> Argument delimiter: ;
> Output destination attribute: /Users/joseantoniorodriguez/Desktop —> ¿Is this 
> necessary?
> Max attribute length: /Users/joseantoniorodriguez/Desktop
> 
> 
> The file I’m using to test are both csv about one of 324KB and the other 22MB.
> 
> After execute I could see the file has been moved from one folder to the 
> other but I did´t see any foo.txt file on my desktop, also I did´t see any 
> error on the Flow.
> 
> Could anybody give me a hand with this I am pretty sure this should be a 
> ridiculous error or misconfiguration. BY the way the OS is Mac OS X.
> 
> Thanks in advance.

Reply via email to