Arriving At Destination ... On Cloud


A Sense Of Achievement

Arriving At Destination - Chequered Flag

Earlier this month I described the intention to implement “page like” functionality for DilemmaLearning.com and gave a progress update the following week. The former sparkle in my eye has now been achieved (yay!) so I want to share some of the highs and lows and what I learned from doing this.

I’d decided to use a MySQL database (hosted by AWS RDS) to store the page like data and that this would be read and updated directly from the client-side browser using JavaScript to make API calls to back-end AWS Lambda functions (NodeJS fronted by AWS API Gateway). This was all new stuff to me and an irresistible opportunity to delve into the technologies and learn by doing.

Forgive all the specific technologies named. This article is about what I learned from the experience and really not at all about the actual technologies. They are named purely for context.

Timeless Skills

Debugging is a timeless skill and well worth any investment you make in learning how to do it well.

There were several situations where a lack of “old-fashioned policing” (code reviews, asserts, error handling code and debug info logging) led to far too much wasted time and “wild goose chases”, including blaming AWS for bugs in their Lambda proxy integration code. (Sorry AWS, but even the code I cut and paste is almost always perfect like the rest of my code. 😊)

For example, following the best practice of putting configuration into environment variables caused me to make a spelling mistake in the name of the environment variable used to hold the RDS database user password. This meant the Lambda function would never be able to connect successfully to the database to do anything. It took me ages to find that simple typo bug. Doh!

However, my code wasn’t even getting that far due to a network problem where Lambda couldn’t talk to my RDS instance. When I reviewed my network setup and permissions policies (copied of course) they were totally wrong (pointing at logs in a region my code wasn’t running in and granting access to a Dynamo database that I wasn’t using).

I would have found the bugs much sooner if I’d known which logs to look in on AWS and the monitoring facilities available but this was all new territory and I hadn’t taken any time to find out or ask anyone about what was available or where to look.

In my opinion, the AWS console encourages newbies like me into bad habits (like hack and slash Lambda coding and manual configuration practices). I think I found that out the hard way when I tried to fix a CORS problem and “accidentally” deleted an API call from my API Gateway shortly after go live. Ouch!

Early Learning

As “early learners” we often find ourselves cutting and pasting example code or following instructions verbatim from others. This is all part of the learning by doing journey when unaccompanied.

It won’t surprise you to know that not everything we stumble across on the internet is best of breed (or even working) code and it’s even more frustrating when the instructions you follow turn out to be out of date or wrong. (I now regularly ignore cloud articles more than a year old.)

Gung Ho With Mis-Guided Trust

Warning Sign

We are more prone than ever to trust and use the first thing we find when we encounter a significant lack of understanding in what we are doing. Knowing we do that should really trigger some kind of warning or fear about what we are about to do!

I’m just experimenting with technology building a blog site – pretty harmless – but someone else might be using the same approach to re-work the code for an aircraft or a nuclear reactor etc. Whoa!

This is the Dunning-Kruger Effect at work (thanks again for sharing that with me @andyburgin).

Not Always Expert Help

I admit I did my share of cut and paste and following tutorials blindly as part of this work – some of it from highly esteemed websites known for helping people get answers to their problems or for helping people get started with web technologies.

Some of the sample code I copied worked the first time but subsequent executions failed repeatably because the code was poorly written and didn’t manage connection closure properly.

Others Sharing The Same Pain Does Not Help

Along the way I noticed lots of people encountering the same problems as me who were presumably making the same mistakes as me and possibly even using the same code snippets, tutorials and articles for guidance after typing the same search phrases into their search engine as me.

I rarely go beyond the first page of search results before trying a different search and like most people I pride myself in always knowing the best keywords to include in my searches – yeah right!

Blind Alleys And Rabbit Holes

Problem situations are notorious for taking us down blind alleys (some people call them “rabbit holes” these days) where we waste a lot of time trying to solve one problem after another because we should never have gone there in the first place.

That’s often what happens when you cut and paste “amateur code” when trying to short-cut a learning curve. It happened to me with some NodeJS asynchronous JavaScript. Luckily (and thankfully) I found a better library of sample code.

Don’t Follow A Lost Sheep

I stumbled across an article where someone described in great detail how they and their team spent two weeks to not get something working and eventually blamed it on a bug in someone else’s library. Wow! I’d rather find a solution myself and wouldn’t want to wait anywhere near two weeks to find it. Two days would even be too long. Two hours would kind of be okay. 😊

They say “a problem shared is a problem halved”. That’s often true but it’s always true if you go to an expert!

The Map Is Not The Terrain

Component testing boosts your confidence but it’s not a replacement for a full end-to-end test. I encountered situations where my Lambda functions worked perfectly in the AWS console hitting the “Test” button for the few test cases I defined but then my full end-to-end tests from the browser would fail miserably.

Like

Tim Simpson
30th August, 2019
#LifeAtCapgemini

« Previous Blog Post Blog History Next Blog Post »