Hi - 

I recently converted to Jenkins. 

First, let me say how easy the interface was to set up and work with - 
great job to the developers for their work on it! 

I've had no issues setting up the environment, moving some cron jobs to 
Jenkins, and getting them to run on the dates/times I expect them to, with 
the dependencies intact, producing artifacts. 

The obstacle I'm facing now is: digesting and processing data produced by 
my builds. 

So, say I have a report (basically a perl/python program that analyzes some 
research data) ABC. I schedule ABC using Jenkins, and Jenkins does a great 
job of informing me when ABC 
1) runs successfully for a given date YYYY-MM-DD or 
2) runs successfully for a given date YYYY-MM-DD but produces some standard 
error or 
3) completely bombs 

The scenario I need to tackle is when the build completes 
successfully/cleanly, but, within the output of the report (artifacts) ABC 
produces, there are 'exception' conditions that require investigation. 

I have many jobs where the output reports/artifacts are text files, each 
file with important info that requires processing and flagging if certain 
conditions are met. 

For example, let's say ABC produces the following results about users and 
their bandwidth usage on a given day: 
User,MaxAllowed,Used
Anne,100,70 
Joe,100,150

So Joe was allotted 100mb of data on a given day, but he used 150mb. I'd 
like some post-build job to analyze the artifacts and flag Joe as having 
breached his limit on the given day. 

What I'm really asking is: what's the best way to plug in a post-build 
process (hopefully one that already exists), that allows for digesting and 
processing of Jenkins build results, whether the artifacts be text files or 
xml files? Eventually, I'd like to categorize Joe's scenario of having 
exceeded his bandwidth limit as an 'exception,' and populate a webpage that 
shows a list of jobs, their Jenkins run status (success or failure), and 
whether the jobs produced 'exceptions' - i.e. # of instances of users 
exceeding their bandwidth limit, per the example in report ABC. 

Any advice would be greatly appreciated. 

I guess I should add that mine is entirely a unix shop, with the following 
languages used for report generation / development: 
- c++
- perl 
- python

Reply via email to