Anyone at Ticketmaster, what is your core tech stack and what failed to scale or limit requests? Also what are you doing differently today? Asking out of pure curiosity from an Infra architect. TC 400 goog
I bet it’s an on-prem machine using an obsolete version of ms sql server
The speed at which the tickets sold out; I don’t think humans were making decisions slowly
I was getting 503s so yeah the services were down. It was actually the graphql API that was throwing 503s
The real question here is What incentive do they have to make the system work better, when they monopolize the industry and will make money on sellout shows either way?
The issues were far from just a slow moving and paused queue. The website itself was not even responding for most to get that far.
It depends on where they failed. If dns failed then they got nothing unless it was cached in their OS. If the server failed then they might get an error from the website. Ditto for the db failing to retrieve the event or some other query. After the event ticket’s row is locked in the db they’re supposed to be in the queue so something failed before this point. The queue would still need room for them in memory or something.
Actually I’m wondering too because at least on the west coast I was checking the code it was paused for an hour then it cleared 25k people in 40 minutes but some people still got 500 errors trying to access. Also, why in the world would you hand out over 25k codes with a max of 6 tix per purchaser? That’s more than the stadium can hold.
I mean they quite literally had the code in the page resources and you only had to look up users in line to find the wait time and the number I kept refreshing to see if it was moving and it didn’t until around 4:10. I did find it odd it was easy to find though but I guess it’s not really a security risk.
Not to mention they probably only sold < half of all tickets in these pre sales. There was limited inventory, with the rest sold for the general sale on Friday — which I suspect will be a shit show as well.
There really isn’t a core stack. Multiple ticketing systems from web to box office are on different stacks and have to synchronize. But the core original technology was built in VAX assembly for the original ticketing machines. It’s just been modernized since. Even emulating in some teams (Not sure if they still are) in Rust containers.
Since they have to sync is that why they didn’t know they were out of tickets until after?
Hi, quick question here cuz there’s barely any threads on ticket master around. Does ticket master/ live nation provide work cell phones to any specific level and above? 🙏🏻
Pretty sure it’s just a Windows 95 machine with dial up