I’m having an insane time trying to grasp how a company that makes this much money isn’t performing weekly DB replication and continuous testing MONTHS in advance for this launch. How do you not have a dedicated test server?
The fact that Fwoibles said that the cadence was different this time around because the live DB prevents them from testing while Classic Anni is alive, is baffling to me. DAY ONE DB engineering should have taught you to have test databases and you can literally replicate live to a test DB so you mimick real load(s). This is like first grade level for development. How does this go overlooked? I don’t claim to know their architecture fully so if anyone has more info I’m all ears. But this just feels like horrible process engineering and a money hungry c-suite.
5 Likes
You probably hit it on the head. They didn’t give them the proper resources to do this and the devs are trying to finagle it the best they can.
1 Like
You’ll play by tomorrow. It’s time to just grow up and wait and maybe do something else. The amount of complaining is astonishing.
Meaningless response. No one is pissed. I asked a question. You failed to answer it lol. Should look elsewhere to troll.
with all these ICE raids the devs are out protesting
6 Likes
Yea I definitely feel for the devs working on this. I’m more disappointed that management can’t get their process together.
pretty hypocritical calling anyone else a troll
It probably just comes down to money…as in how can they make more of it. In this case they got our drug (WoW) and we’ll wait for it!
1 Like
Lemme sit back and let you explain this phenomenal line of logic.
well lucky for you, your ability to grasp it doesn’t matter, so you don’t need to stress about it!
1 Like
And you’re complaining about the complaining about complaining when will the cycle end!?!?
The TL;DR I got from skimming the Twitter thread is that they are forced to complete a full server infra migration from 0 → 100 in 24 hours due to underlying complexities / inherent incompatibility with 'the new world’ (new server architecture).
From a dev perspective, this is an absolutely insane undertaking and I truly hope the devs are getting some rest and are hydrated over there. This is a herculean effort by industry norms.
1 Like
Another important thing to consider is they likely are in the trenches of 20+ year old server infra (the layer Dante failed to document).
No idea what the server setup is like but I can easily foresee an ungodly amount of bare metal spaghetti code making this three billion times harder than spinning up a docker container and calling it a day.
Thanks for that rundown! Yeap, definitely seems like the devs got screwed by poor management and bad process engineering. Gotta feel for the grunt work the devs have to do here. Been in those trenches before with management breathing down your neck for their bottom line.
The bigger issue I don’t understand is Fwoibles’ comment about not being able to test the actual infra changes since the classic DB was supporting Anniversary. Like why not replicate the way most companies do? I can only assume it’s because this would cost some more money.
1 Like
yea, it’s probably some technical detail e.g. can’t do a ton of R/W without compromising existing server integrity. my assumption is that for some reason they had to wait for a complete shutdown before beginning db operations to safely execute the migration steps. it’s usually some boring technical edge case in my experience.
Good, more power to them.
Yea that’s fair. If they wanted to, they could have a separate servers for testing but I digress.
Difficult to say what’s really up without being on the inside.
Good luck to you Desu !