A small success"It's amazing that you can deploy on a Friday at 5pm, then take off."
Last Friday, around 3pm, I discovered that the code I was working on for a client had the potential to allow SQL injection attacks. The fix wasn’t difficult, and after about half an hour I had a PR ready to be merged.
The question then became: Do we merge and deploy now, right before end-of-day on a Friday? Or wait until Monday at the next planned release?
We discussed for a while, and the details don’t really matter.
What does matter is that we decided to go ahead with the deployment on Friday. But to wait until after 5pm, so that the 2-minutes of downtime that come with each release, wouldn’t happen during business hours.
So the team lead deployed the software, then said:
Deployment is complete. I gotta run. I hope nothing breaks. See you next week.
But the clincher is what happened next. The CTO said:
It’s pretty amazing that you can deploy on a Friday at 5pm, then take off immediately. That wasn’t even thinkable a month ago.
A month before, deployments were happing at 3am, and involved a bunch of post-hoc manual testing.
Can you deploy a security patch at 5pm on a Friday, then take the weekend off?
If not, what’s standing in your way? What steps can you take to improve the situation?
What did we change?
What changed in the last month to give us the confidence to deploy at 5pm on a Friday?
Your low standards don't represent an engineering failure on my part
What is it that makes you uncomfortable deploying at 5pm on a Friday? Let's talk about that.