Final Fantasy XIV hits a major instanced roadblock in Stormblood early access

Everyone knows there are going to be issues with an expansion on launch, and Final Fantasy XIV: Stormblood is no exception. Players have hit up against a pretty nasty one, however, even beyond the usual matters of queues and server loads; something seems to be causing issues with instanced content, including a very early part of the main scenario that’s causing huge population bottlenecks around one questgiver.

Producer and director Naoki Yoshida has been doing his best to keep players updated, but at this point getting past this early instance is largely a matter of clicking for entry repeatedly and hoping to not get the now-dreaded error message stating that the instanced battle could not be started. It’s not exactly surprising that the game would have some issues with early access, but players are left unable to progress beyond the first two areas of the expansion without some appreciable luck (and the instance servers aren’t at their most stable even for other purposes). The issue is being addressed, so keep your eyes peeled for updates when the logjam is finally broken.

SHARE THIS ARTICLE
Code of Conduct | Edit Your Profile | Commenting FAQ | Badge Reclamation | Badge Key

LEAVE A COMMENT

83 Comments on "Final Fantasy XIV hits a major instanced roadblock in Stormblood early access"

  Subscribe  
newest oldest most liked
Subscribe to:
styopa
Reader
styopa

From the very first time I tried FFXIV, the idea of significant character advancement bottlenecked behind quests tied to doing instances seemed ill-advised.

Reader
Loopy

Is this a first time a lot of you are experiencing launch day of an expansion? I’m not sure why everyone has such short term memory when it comes to expectations. Take a break, wait a couple of days, and things will normalize.

Justin Olivetti
Staff
Justin Olivetti

Maybe it’s because Morrowind and Legion raised expectations regarding smooth expansion rollouts. Then again, it’s a good sign of a game’s popularity that it has these issues.

Reader
Loopy

I don’t know.. for every 1 “smooth” launch there are 5 congested, laggy, buggy ones – and i’m ok with that and have grown to expect it. But, like you said, it shows that the game is popular and the numbers are showing so that’s always a good thing.

Skoryy
Reader
Skoryy

Popular game has servers jammed at expansion launch. More at 11.

Reader
Pentalyn Priestess

Worst launch for an MMO in a long time…what utter crap. My opinion of Squeenix just dropped down the tubes. Just another money grubbing corp looking to milk the public out of cash.

Reader
Chuck Finley

None of this sentence made sense at all. You’re just one of those gamers that cry at every little thing when something doesn’t work out 100% all the time.

Reader
Kickstarter Donor
Patreon Donor
Loyal Patron
BalsBigBrother

It’s happened before and it will happen again, yes it sucks, Yes Square Enix should know better and I am sure Yoshida is doing what he can but sometimes even with the best will in the world things go wrong.

It is easy to look and say well they should do this or they should do that but you are not in the middle do not know what resources they have access to or what limitations they have on what is possible for them.

All I can say is I hope that the issues are resolved as quickly as possible so you all can get back to enjoying a game you enjoy enough that you went in the the early access of the new expansion. Be calm and good luck to you all.

Reader
Giannis Papadopoulos

Btw there is another problem in ffxiv.. for those lucky to get past the msq, do not do the primal! If you do, you will not be able to login with your character any more…

http://forum.square-enix.com/ffxiv/threads/329048-Character-broken-after-first-SB-primal%21-I-need-help-ASAP?p=4208570&viewfull=1#post4208570

Reader
Kawaii Five-O

For anyone just now seeing this, it was fixed a couple of hours ago when they brought the Ruby Seas zone down for maintenance.

Reader
Giannis Papadopoulos

oh thats great! :)

Reader
Jack Kerras

Important note: this appears to be connected to using emotes while transitioning away from the instance server.

Reader
Giannis Papadopoulos

so all these people on the forum post they did an emote while clicking the exit ? They seem too many for such a coincidence. Except if the game does an auto-emote while you complete the dungeon…

Reader
Kawaii Five-O

That includes being in /sit when going into the instance. Not that uncommon.

Reader
Loyal Patron
Patreon Donor
Kickstarter Donor
Little Bugbear

Wow that sucks! Thanks for posting this. Only one person I know has done primal (He got through). So I guess the rest of use will have to the beware primal.

Reader
Giannis Papadopoulos

Yeap and the fix will not come fast.. they need to make a patch for it.. probably a week+ :(

Reader
Giannis Papadopoulos

First time in my mmo life was too lucky! I got past both bottlenecks in my first try!

Reader
Ronivan Fontanez

SE is pretty much doing nothing to solve the issue. Yoshida’s explanation disgusted me, it was like a big fu%$# you players, pay your subscriptions and be quiet. If they really wanted to fix it, they would have by now, they are just sitting still and letting the problem fix itself.

Reader
Jack Kerras

They’re absolutely doing something. The same thing everyone else (save a few biggies) does when the launch day spike hits and everyone is mad:

Ride it the fuck out.

It’s all you can do. It’s that or reinvent your entire server architecture on a game that’s already running in a handful of datacenters all over the world, which will cause its own completely discrete and totally-impossible-to-predict problems, not to mention costing an astronomical amount of money and shredding the tech staff you are demanding the impossible of.

The issue isn’t that they’re not doing anything. The issue is that their servers are not automatically backed by a mountain of extra hardware that sits and gathers cobwebs from T+5days on into forever.

You can’t buy enough servers to absorb a launch spike, with the exception of AGS, Microsoft Game Studios, and Blizzard… and Blizzard is -purely- by the grace of having orders of magnitude more money than anyone else. They don’t run a commercial cloud service they can cannibalize to absorb spikes like this, they just run a TON of immensely popular games, and three tend to empty out a bit when the fourth is getting hit super hard. Oh, also, they have incredible techs who have built a pretty amazing infrastructure that allows their servers to shift around horrifying amounts of load.

It takes great techs and an immense amount of headroom (which no one can feasibly afford save companies who run a commercial cloud offering also) to do things like this. Squenix doesn’t have that. They aren’t ignoring you and me and our woes about FFXIV’s instance servers being totally fucking destroyed, they just can’t fix it without the board of directors firing them all for shoveling cash into a furnace by the truckload.

Reader
Jason

I see this kind of reasoning a lot, and often times it is probably true but it completely precludes the possibility that the company or any of its team has made a mistake. Which is…exactly what’s happened in two of the three fixed issues plaguing Stormblood in this case.

The post-instance disconnect/lockout bug is caused by a feature they added which wasn’t fully tested on uneven surfaces. And we know now that the instance instability and disconnects are due to a “Server program issue.” Two big problems that seem to be related to connectivity but both were programming issues and 100% Square’s fault.

TLDR: There’s a good chance that the last issue is their fault as well since the first two were. And at the very least we know how they designed the new content created the issue. Assuming launch errors are never the company’s fault isn’t good for consumers or the industry. Don’t set the bar low; only lazy companies benefit when you do.

Skoryy
Reader
Skoryy

The post-instance disconnect/lockout bug is caused by a feature they added which wasn’t fully tested on uneven surfaces.

I work in retail tech on systems tested by two seperate QA teams. Don’t get me started on what they’ve both missed once the product hits the field. Heck, we had to roll back production code just this past week.

Never forget Murphy’s Law of Combat: No plan ever survives first contact with the enemy. Or in this case, the customer.

Reader
Jack Kerras

It’s really not a mistake. The mistake would be buying a truckload of servers you then can’t put to use for anything.

I work in hosting and I know what initial usage spikes look like; if you buy for those instead of buying for what your actual usage is going to be, you spend a boatload of money buying nothing.

Also, the name of ‘server program issue’ has no real meaning; this is a piece of non-information you’ve been given and you’re interpreting it the way that you want to. I’m not interpreting it at all; I know what happened, because what happened is an extinction-event meteor made out of humans hit their servers at the same time, which is what always happens during xpac launches in any popular video game.

Further, the phrase ‘fully tested’ is truly ridiculous; nothing is even starting to show its cracks until it’s in the wild, full stop. Zero testing is worse than some testing or extensive testing, but there’s still nothing that will stop Day 1 short of having a shipping container full of powerful servers that you’re willing to throw into another project a week after your product launches.

That’s why Microsoft launches one of their biggest game properties with new Azure backing and has -zero- matchmaking issues despite the previous installation being plagued by same; they bump their Azure headroom upgrade for that quarter a little extra, allow Halo 5 to devour a chunk of that headroom, and then when the population falls down to normal levels, they sell those machines as cloud hosts to their customers. I am dead certain you’ll see this from AWS’ new game studio also, -especially- if they want to enter console spaces; the dreaded-yet-ubiquitous NAT issue is basically solved by using faux-dedicated cloud infra and that is fucking wonderful.

So:

Yes, this is Square’s fault. They don’t have enough servers.

No, they should not solve the problem by throwing enough machines at it to make it work Day 1, because I would like them to continue to have a video game I can enjoy, and making foolish investments while attempting to solve temporary and widely understood problems is not conducive to same.

This is not me setting a low bar.

If you wanna bitch about game companies doing things wrong, there are way better trees to bark up than ‘well they should buy enough servers to handle launch day’.

No, they shouldn’t. Most of them couldn’t even if they were fool enough to try. The ones that can only can because they -also- rent multiple warehouses full of servers to other businesses on a daily basis, and grow their install base so fast that SSD production facilities literally cannot keep up with them.

Reader
Jason

There’s no other way to put this: you don’t know what you’re talking about in this instance. I’m guessing you don’t play XIV and/or haven’t read the notices put out by the team. I’ll happily explain:

The first bug I referred to happened because they implemented a feature that preserves persistent poses started before and after an instance without testing if it could cause a character to become stuck in an instance if it was initiated on uneven terrain. That was an abject failure of testing.

The second “server program issue” most likely refers to an individual Daemon in their web of server processes that started to chug and became the focal point of the slowdown. They fixed it and gave a notice that the problem won’t occur again after.

BTW, I woke up today and of course the third problem is now miraculously fixed too. We’ll have to see if it resurfaces again later under load, but I doubt it. The problem persisted constantly for the last 48 hours regardless of fluctuating load after all.

These issues would’ve occurred whether or not they had more instance servers available. I hope you actually read what I wrote this time and process it instead of going off in another tangent.

Reader
Jack Kerras

Oh hey! In that first instance, I just missed the fuckin’ memo. :) Thanks.

amkosh
Reader
amkosh

Sorry dude, This is utter BS. IF you’re ONLY solution to a what is a scaling problem is to throw hardware at the problem, then you sir, need to resign and find a job where you can’t fuck with technology.

Guess what, scaling *is* hard. But it is also a known problem with many known solutions (besides the idiots guide to throw more hardware at the problem)

There is an economics problem here. Spending money on quality people who can scale is not cheap, and in some ways probably more expensive than just tossing HW at it, which will eventually cause some other single point bottleneck to appear.

Unless people are willing to make it uneconomical by denying them revenue, then it will never change. And that likely will never happen.

Reader
Jack Kerras

There’s a huge difference between ‘make it scale better’ and ‘make it scale SO well that it can handle the launch spike without stumbling’, and those are orders of magnitude in terms of difference.

Further, we’re basically talking about scaling out a legacy system on top of everything else, and unless they want to move to a whole new architecture (which is an entirely different can of worms with a discrete set of issues), they’re going to have a difficult time maintaining their shard format.

Also, yes, it’s difficult to hire the right people for things like this; if you don’t have a crazy specialist at this sort of thing, it’s even difficult to tell who is and isn’t -also- a crazy specialist, and hiring wizards when you aren’t a fucking wizard and don’t know what one looks like is really, really hard. It’s not purely a money thing.

Scaling is definitely hard, but making a system this substantial scale to 5x the amount of people it’s ever going to see again – and gettting a 5x normal concurrency on an xpac launch for an MMO isn’t that unusual on Day 1 – is a completely different can of worms than scaling out your smallish music-production site so that it absorbs the impact of getting hit by a bunch of people on an album launch.

Scaling simple things up to absorb spikes and to not actually fall over when a lot of things hit it is totally normal. Scaling legacy MMO servers up when they have to remain compliant with a console’s needs, when they have no intention of moving over to a megaserver format, when they get hit this much harder and by this much more people than usual… man, that’s not the same thing as ‘a scaling problem’. It’s fundamentally the same problem at its core, but that does not mean that ‘just hire a team more capable of scaling’ is going to solve the launch-day hype of an MMO. Too many more people than usual to handle things without buying more servers, full stop.

You can handle greater-than-average concurrency and server congestion in more ways than just ‘add hardware’, of course you can. There is a big difference between a growing playerbase and consistent, predictable congestion increases over time and the launch of a large, paid content patch. Even Steam – a significant minority of players in FFXIV’s case, since they ran for years without it and Steam xpacs can’t be bought on non-Steam accounts – literally doubled its concurrency overnight. If you’re serving a website, sure, that’s doable. If you’re running an MMO, multiplying your logins is too fucking hot.

There’s a point where merely making better software is not enough, and MMO launches are recognizably one of those points. Launch issues happen not because of laziness or benightedness, but because people get fuckin’ excited and stomp on the server as a gigantic gestalt. This big spike is -never- repeated afterwards short of xpacs, and a server structure that doesn’t scale as readily or as obviously (see: EVE, ESO, Wildstar) will have issues wherein a ton of shards just -empty out- after only a few days. First-day login spikes are always an unprecedented number of people; day 2 or 3, concurrency is much more normal.

Rustyblades
Reader
Rustyblades

I feel like I should be sitting next to a dead dwarf with a bucket on a well in some dungeon – “there is no escape, can’t get out, can’t get out….” since when you are going into an instance, the game keeps trying to put you… into the instance… but if there isn’t room it instantly kicks you out to the start screen… smooth move Square… smooth.

Add that to the queue to actually get into the server… like waiting to get punished.

Reader
Michael Gerringer

I am stuck in the instance login loop now. I will just wait till the rush dies down. i am still the the Heavansward content right now anyways.

Reader
Michael Gerringer

It is pretty frustrating. Same thing happened at launch, Heavansward, and now Stormblood. I wish they would have this type of thing sorted by now but that is ok. I understand that is the way it is with MMOs. It just sucks I have the next couple of days off and by the time things die down I will be back at work.