I'm managing a Gitlab instance for 10k users since 3 years, we had zero unplanned downtime in the last 3 years, I just had to apply the upgrade every 3 months and it's running smoothly ever since
It’s why we run certain services in my department ourselves rather than central IT, because central IT thinks that the best time for planned downtime is 6pm, which is hilarious as for our department 5pm-11pm is the most critical time, and the best time to make changes is between 9 and 12 when the services are least important (not that we have a full outage, but we can have a degraded outage - our internet service for half of our equipment can take upto 20 seconds to fail over, and not every end device is dual connected)