Bratty Redhead

the sarcasm is free!

Reason #eleventyseven to Automate

About 4 years ago, I was working for a remote client, building a large, fragile Yantra/WebSphere environment.  The client had chosen to use the WebSphere embedded messaging engine (instead of MQ) for systems critical order processing.  The project was big, I was busy, traveling to the client site every week and I had a constant backlog of work.  It seemed like there was never enough time or people to get all the work done.

One day the dev team came to me with their list of 15 queues that needed configuring in dev.  I was already overrun, had no time and typically working 12 hours or more on the days I was onsite.

If you’ve ever worked with WebSphere, you know that you pretty much get two choices for configs: the GUI console or scripts.  The scripting language is based on either jython or tcl but the WebSphere specific DSL is a bitch to deal with if you don’t know the syntax really well.

I really wanted to write some WebSphere scripts to install and configure the queues, but I told myself I just didn’t have time, what with all of the probable debugging and figuring out how to deal with the syntactical challenges.  I figured it would take me less time and I’d just kick out the queue configs super quick via the GUI.  The GUI is tedious but not hard and I didn’t have time to play with scripting, I thought.

Except.  All the naming was wrong apparently.  All the queues had to be deleted a few days later and re-created.  I told myself the second time that it wouldn’t happen again and I’d just be wasting valuable time, writing scripts that wouldn’t be used again.

Except.  They had to be recreated a few times in the Test environment, for several reasons.  And then eventually they were revamped again to match up better with something else.  Not to mention we had to do the work on the queues in prod.  I hear that, after the project went live, they reconfigured the queues a few more times after that.

This situation is one of the more glaring moments where I can look back and say, man, you should have taken the couple or 3 hours to write some automation scripts for a seemingly trivial task.  This would have made sense for several reasons:

  • 3 hours of concentrated (and let’s face it fun) time to get some scripts written, instead of numerous occasions being interrupted to do trivial admin work by hand
  •  If I had written a script, anyone on the team or even someone from the noc could have done the work instead
  • I might have left something after the project that was useful in the long run as a legacy and as an example for further scripting for the admins who stayed behind

I was just telling this story last week and thought that it could really stand repeating for posterity.  Taking the extra time to write a script that automates a lame manual process is NEVER time wasted.  Not ever.  And if anyone tries to tell you differently, you’re probably too smart to be working with them.