Jump to content
  • Sign Up

Reducing Bugs in Future Releases


Recommended Posts

I’ll start off by saying I don’t wish to be negative, but as a player, I have some concerns.

 

This episode of season 4 was delayed, and we had a post about “adjustments to how we (ArenaNet) approach developing each episode that will help us hit our quality standards more consistently”. Unfortunately, with this release the players experienced quite a number of bugs, some of which are still not yet fixed (such as the portal scroll won’t go in the portal tome, and raid wing 3). Personally, the amount of bugs took away from my enjoyment of this episode.

 

My question being, what are you doing to make sure the players don’t experience this amount of errors on a release going forward?

 

(much better title, ty )

Link to comment
Share on other sites

I get the impression that some of the bugs were spontaneous, and other bugs were due to unfinished content. I know it sounds like a conspiracy theory, but I can't trust PR messages anymore for games in general. It just felt strange to hear: 'So, we had this game-breaking bug appear when we moved it from our test to live servers.' This hasn't exactly happened before on this scale. The only thing that comes close is the launch of Path of Fire but for an expansion launch, that was understandable. It feels too: 'Put out the fires, deal with it later' for me. It might seem a bit rude, but that's my perspective of it. I'm glad that most of the bugs have disappeared, but I'm a bit 50/50 on the whole matter.

Link to comment
Share on other sites

I'm actually more concerned about if whatever caused the bugs, might force the devs to restrict themselves in the future.

I mean the only fix we read about in patch notes for the connection issue was reducing the size of the Instance (which, granted was pretty much half the whole map, if not more). So does this mean, in the future devs will have to restrain themselves from creating awesome experiences, like the scouting mission was, for fear of creating a new error, or was that also addressed in other ways not in the patch notes?

I'd love to read an honest recount of what happened, what caused it, and what were the consequences in terms of how the game is developed in the future, if any.

Link to comment
Share on other sites

> @"ReaverKane.7598" said:

> I'm actually more concerned about if whatever caused the bugs, might force the devs to restrict themselves in the future.

> I mean the only fix we read about in patch notes for the connection issue was reducing the size of the Instance (which, granted was pretty much half the whole map, if not more). So does this mean, in the future devs will have to restrain themselves from creating awesome experiences, like the scouting mission was, for fear of creating a new error, or was that also addressed in other ways not in the patch notes?

> I'd love to read an honest recount of what happened, what caused it, and what were the consequences in terms of how the game is developed in the future, if any.

 

Obviously, I'm not a developer, but I get the impression that in the future, they'll split the instances in half and make it a two-part process like in Sandswept Isles. So we'll do one instance, load screen, next instance, rather than it being seamless. Personally, if it makes things easier to run/render, I think two instances is completely fine.

Link to comment
Share on other sites

> @"Twyn.7320" said:

> > @"ReaverKane.7598" said:

> > I'm actually more concerned about if whatever caused the bugs, might force the devs to restrict themselves in the future.

> > I mean the only fix we read about in patch notes for the connection issue was reducing the size of the Instance (which, granted was pretty much half the whole map, if not more). So does this mean, in the future devs will have to restrain themselves from creating awesome experiences, like the scouting mission was, for fear of creating a new error, or was that also addressed in other ways not in the patch notes?

> > I'd love to read an honest recount of what happened, what caused it, and what were the consequences in terms of how the game is developed in the future, if any.

>

> Obviously, I'm not a developer, but I get the impression that in the future, they'll split the instances in half and make it a two-part process like in Sandswept Isles. So we'll do one instance, load screen, next instance, rather than it being seamless. Personally, if it makes things easier to run/render, I think two instances is completely fine.

 

Sure, but for example, Forearmed is Forewarned, couldn't be done that way. Maybe they could remove that area where the portal was, and have a loading screen there, but most of the mission is around the south half, which is still huge.

Link to comment
Share on other sites

> @"ReaverKane.7598" said:

> > @"Twyn.7320" said:

> > > @"ReaverKane.7598" said:

> > > I'm actually more concerned about if whatever caused the bugs, might force the devs to restrict themselves in the future.

> > > I mean the only fix we read about in patch notes for the connection issue was reducing the size of the Instance (which, granted was pretty much half the whole map, if not more). So does this mean, in the future devs will have to restrain themselves from creating awesome experiences, like the scouting mission was, for fear of creating a new error, or was that also addressed in other ways not in the patch notes?

> > > I'd love to read an honest recount of what happened, what caused it, and what were the consequences in terms of how the game is developed in the future, if any.

> >

> > Obviously, I'm not a developer, but I get the impression that in the future, they'll split the instances in half and make it a two-part process like in Sandswept Isles. So we'll do one instance, load screen, next instance, rather than it being seamless. Personally, if it makes things easier to run/render, I think two instances is completely fine.

>

> Sure, but for example, Forearmed is Forewarned, couldn't be done that way. Maybe they could remove that area where the portal was, and have a loading screen there, but most of the mission is around the south half, which is still huge.

 

Well, how it'd work, and it'd be a bit strange at first. We'd do the bit around the Village, and there would be a gate leading to the South Half. At the gate, we'd have another instance start point, and it'd load the next bit as its own instance. Once we finish that bit, we'd go back to that gate, and travel back through to the Village and the Portal to do the third bit. It'd feel a bit strange, but obviously, it's all about loading instances.

Link to comment
Share on other sites

Just to be clear, I'm not talking about ONLY the large map instance bugs. I get that this was something specific, but along with this, we also saw a number of other bugs with the episode. For some examples:

 

* New code altering Raid Wing 3 (another error that went through, but these are adding up. This isn't the first time code has made it to live before intended)

* Yellow border in raids (probably the same code that bugged wing 3)

* New portal scroll, which can't be added to the tome (this really should have been caught)

* Multiple Deepstone fractals in one daily cycle (2/3 recs one day, and a teir + rec another day)

* Deepstone "no falling" achievement (people were getting this with falling, and others not getting it while not falling)

* Event on the new map needed for the story bugged so enimies didn't spawn (you needed to kill 5-10 enimies, but I forget the full details of this event)

* Failed event being needed for a collection (no reason for this not to get caught and corrected, we have been down this road enough times)

* No loot from Meta enimies (we've been here before)

* Beetle collection targets being killed far too fast for people to get cradit, such as the spider and SW beetle (there should be some realization that a large number of players are going to be after these targets on release)

* Login error1083 (maybe a little related to the instances , as it gave the same error)

 

These are just the bugs I personally ran into, so you can check the reddit bug thread for others (I haven't read through it). I do realize some of these may not be "bugs", but rather more general quality.

 

Asking for zero bugs on release is unreasonable, and mistakes do happen. In this episode, those mistakes really added up. I'm not angry about it, I'm not leaving Guild Wars 2 about it, and I'm not blaming QA (or anyone else). What I do want is some restored faith and communication that ArenaNet will try to prevent this from happening again. [There was a statement about quality beforehand](https://en-forum.guildwars2.com/discussion/41775/season-4-episode-3-is-on-the-horizon#latest "https://en-forum.guildwars2.com/discussion/41775/season-4-episode-3-is-on-the-horizon#latest"), so I hope they are willing to answer to that now.

Link to comment
Share on other sites

> @"Ewon.5903" said:

> Just to be clear, I'm not talking about ONLY the large map instance bugs. I get that this was something specific, but along with this, we also saw a number of other bugs with the episode. For some examples:

>

> * New code altering Raid Wing 3 (another error that went through, but these are adding up)

> * Yellow border in raids (probably the same code that bugged wing 3)

> * New portal scroll, which can't be added to the tome (this really should have been caught)

> * Multiple Deepstone fractals in one daily cycle (2/3 recs one day, and a teir + rec another day)

> * Deepstone "no falling" achievement (people were getting this with falling, and others not getting it while not falling)

> * Event on the new map needed for the story bugged so enimies didn't spawn (you needed to kill 5-10 enimies, but I forget the full details of this event)

> * Failed event being needed for a collection (no reason for this not to get caught and corrected, we have been down this road enough times)

> * Login error1083 (maybe a little related to the instances , as it gave the same error)

>

> These are just the bugs I personally ran into, so you can check the reddit bug thread for others (I haven't read through it).

>

> Asking for zero bugs on release is unreasonable, and mistakes do happen. In this episode, those mistakes really added up. I'm not angry about it, I'm not leaving Guild Wars 2 about it, and I'm not blaming QA. What I do want is some restored faith that ArenaNet will try to prevent this from happening again. There was a statement about quality beforehand, so i hope they are willing to answer to that now.

 

Agreed. I also don't think finger pointing is kind of wasteful at this point, we already lost those hours waiting for the fixes, they won't bring them back, or try to compensate for it (but that's a different issue).

 

I do agree that at this point the only thing we can do is ask what is/was done to prevent a situation like this from arising again, knowing full well that accidents, and incidents, do happen, and the only thing you can do is try your best to prevent them and their consequences.

Link to comment
Share on other sites

> @"Twyn.7320" said:

> I get the impression that some of the bugs were spontaneous, and other bugs were due to unfinished content. I know it sounds like a conspiracy theory, but I can't trust PR messages anymore for games in general. It just felt strange to hear: 'So, we had this game-breaking bug appear when we moved it from our test to live servers.' This hasn't exactly happened before on this scale.

 

It may not have happened to ANet before, but I assure you that this is an annoyingly common factor of bugs in production for developers. I have worked on some of the largest systems in the world, where significant efforts to literally never go down are spent, and ... we have been down for hours, even days, at times.

 

A big factor in that is that seemingly unrelated changes can cause failures because you tested on one version of the underlying OS, and prod got updated to a slightly different version that, whoo, bug! This is most likely to happen when you have recently changed servers ... and, hey, look, ANet recently move to AWS from their previous self-hosted stuff...

 

The other big factor is that something which works fine when you test it with a couple dozen people, and works fine when you generate simulated load of thousands of people, turns out to have bugs that take it down when 50,000 people all start hitting it at once. The bug just doesn't manifest until the specific set of things happen at the same time, which you never hit in testing because you are not really all that good to simulating human stuff happening at that scale.

 

Finally, we know that one of the issues was that a system which managed instance size was killing things: basically, something that was there as a watchdog to prevent problems with one instance bringing down the whole system. Unfortunately for them, it considered every instance of the new map to be "broken" and ... well, sometimes the root cause of your outage is a system meant to prevent outages.

 

None of this is unique to ANet, and while they may or may not have hit it before -- and believe me, this was a super-terrible launch -- the reliability issues are entirely believable, and are consistent with real world experience in other large scale systems.

 

> The only thing that comes close is the launch of Path of Fire but for an expansion launch, that was understandable. It feels too: 'Put out the fires, deal with it later' for me. It might seem a bit rude, but that's my perspective of it. I'm glad that most of the bugs have disappeared, but I'm a bit 50/50 on the whole matter.

 

I'm also glad, and I hope they appropriately change things to avoid this ever happening again. That said, no matter how many of the smartest people in the world you have spend literally centuries worth of person-time effort on making sure you never make the same mistake twice ... your system *still* goes down because something else breaks, and billions of people yell at you.

Link to comment
Share on other sites

×
×
  • Create New...