The problem is when they have more services than sysadmins. While the sysadmins are busy upgrading the git server, the logging infrastructure suffers. They pivot to work on that, now the CI/CD server is down/slow/randomly breaking. But the sysadmin that knew the ins and outs of it left last quarter so the new sysadmins don't want to touch it. Oh, and management doesn't prioritize any of this stuff, so actually jumping two versions on the git server is a much bigger, fragile ordeal now than it was a month ago.