in team-work culture profile image for Brendan Caffrey by Brendan Caffrey
3 things I learned the day I dropped the accounts table

3 things I learned the day I dropped the accounts table

I really identify with the mantra "move fast and break things". Everything in moderation of course - like don't move so fast that you break all the things. But as an engineer, I've had my greatest success by bringning things to the business/management and effectively saying: hey, I built all this cool stuff - do you think you can sell it?

Early in my career I was learning the ropes as a DBA at a financial company. I was working on a new data load that would create new Accounts in our database. To simplify things and protect the innocent this is how it went down: I was loading the data into an Accounts_Temp table and working in a staging connection string. I was writing some raw MS SQL to Truncate Accounts_Temp and do a BULK INSERT. Of course if you happened to forget that _Temp part and inadvertenty ran that against the wrong connection just truncated the primary accounts table of a major financial company.

So yeah - a lot of things went wrong. I shouldn't have had truncate access to prod. I should have realized I was on prod. Lots of things should have been different. This isn't about all the things that went wrong. This is about all the things that went right that day.

It takes a few seconds. Then the realization creeps in. "That wasn't production...was it?" Followed quickly by, "I can fix this. I can fix this. I can fix this." You most certainly cannot fix this. At least not on your own.

In that moment you have choices. I paused, pannicked, and ran right into my boss' office. "I just dropped the accounts table, what should we do?!?". In the following 2 hours I learned more about being a DBA than in the previous 6 months. We did a partial, live restore of a single table with minimal data loss. It. Was. Awesome.

The things I learned that day:

  1. Make tragic mistakes hard to do. Don't let people make drastic mistakes easily. Things like truncating data or deploying to master should have checkpoints along the way.

  2. When you make a big mistake tell someone right away! Imagine if I hadn't. The team might have spent hours - or longer - trying to figure out what happened. Is there a software bug? Were we hacked?

  3. Let people make mistakes. If I was afraid of getting fired or berated for my mistake - I might have acted differently. Because I knew my boss was supportive of me, I knew I could come to him. I could tell him what I did wrong. I didn't fear being mocked or belittled or fired due to this one mistake.

As I've progressed in my career I've found the first one is pretty easy. It's a computer problem - we have gotten pretty good at solving those. The tricky part is the balance. How can you protect your team from mistakes but still give them freedom to move quickly?

The second two are people problems. Those are more complicated.

Make sure your team knows they won't be dragged across the coals for their mistakes. When something does go wrong - focus on the problem - what happened, how can we solve it. Don't focus on who did it. Today it might be Developer A but tomorrow it could easily be Developer B. What are the processes and procedures that allowed the mistake to happen.

Any conversations about negative performance should be in private. Never call people out in front of their peers.

I try to emphasize to my team - I'm looking at the long view, not the short view. You're not getting fired for having a bad day. We all have bad days. What's your month look like, your quarter - I bet you've contributed significantly.

As a mistake maker - have some confidence along with your humility. Yes I made this mistake. But I had the right intentions. I was trying to do the right thing. I want to fix this. I want to get better. How can we all learn from this? Write up a retrospective. Reccommend policy changes to prevent this from happening again.

We've all made mistakes. Let's create an environment where the really bad mistakes are hard to do and mistakes are both forgivable and recoverable.

About the author:

Hello! I'm Brendan Caffrey. A software engineer and leader currently working as Engineering Director @ Litmus where I lead teams making awesome tools for email professionals.