For one reason or another, something monumentally stupid happened to one of my clients' Vultr instances yesterday that took it down for far too long. When we went into the clients' Vultr console to check on its status, the VPS was stopped with a message that it was down due to some emergency thing and technicians were working on it to bring it back up ASAP. Okay, cool, but I saw that at 4pm and it stayed down until almost 10pm when I submitted a support ticket through the client's account asking for an update on the outage.
Right after submitting the report, the VPS hosting their webserver went from "stopped" to "running". It started responding to pings but wasn't serving the website yet, so I tried viewing the console and got in at first but then got a connection error that kicked me out and kept coming back when I tried to reconnect. Attempting to restart the server through the console was met with an error message stating the server timed out so it couldn't be restarted. I was about to try an SSH session when the website loaded in my browser and seemed to be working fine again. I was able to log in, make a full backup, and download it. Shortly afterwards the support ticket was updated with a message noting that the instance should be back up and running now. The entire time, all of my own VPS instances on my account were running just fine.
I'm glad the server came back up, of course, but I can't shake the suspicion that human error was ultimately responsible for the length of this particular server's outage. Though possibly a coincidence, it is pretty fishy that the server came back up immediately after submitting a ticket after being down for so long. It makes me wonder if the actual outage wasn't that long but someone forgot to flip a switch somewhere to start the server back up... And of course, their SLA is vaguely mentioned in the TOS but not actually laid out anywhere I have found so far... except that you waive your right to it if you don't mention it to them within three days.
Throughout the outage, spinning up a backup from their automated server backup system didn't work, despite me doing exactly the same thing earlier that day on a test server. The original VPS was in LA, and the backup instance was in Seattle, but neither of them were loading the website. Also, the LA node was listed as green with no issues here the entire time.
I'm just glad I'm on good enough terms with the client, and the website is small enough at this point, that it wasn't the gargantuan issue it could have been. Still, you better believe I now have the motivation to learn EC2 hosting and revamp my site backup system. Vultr just went from "awesome" to "not for production, or maybe anything" within the space of a few ridiculous hours.