3/ Watch the dashboard live for the first hour.
Spread out background jobs. 250 push notifications at once caused our second spike.
Have a downtime message ready.
Write down what broke while it's fresh.
I launched a new product and ~250 people jumped on at once.
CPU maxed out. Server fell over. ๐ญ
The circled bit is 200% CPU and panic setting in. The gap is the server being switched off and upgraded. ๐
3/ What I took from it:
You won't find your real limits in testing. Real users will show you.
"It crashed because too many people wanted it" is a good launch-day problem.
Being ready to recover beats looking perfect.
I struggle with this daily ๐ฃ
How much juice do you squeeze?
* Being content vs. hungry
* Rest vs. pushing
* Acceptance vs. criticism
* Enjoying what youโve built vs. extracting more of your potential