Random destruction of wings

[QuesoExplosivo]

I was flying along, when suddenly:

Javascript is disabled. View full album

my wing exploded. It said it crashed into the VAB, which I was a few kilometers away from at the time. Infuriatingly, I was playing on hard mode. This is an 0.25 unmodded install on Mac.

I was able to force-quit KSP, and go back to an earlier autosave, so it wasn't as bad as it could have been.

Edited October 14, 2014 by QuesoExplosivo

[RexKramer]

This is a known issue. It is difficult to reproduce this reliably, seems to happen randomly.

It certainly provides some excitement when a wing suddenly blows up!

While your screenshot indicates the wing thought it collided with the VAB, other buildings have been indicated as well, including the runway.

[Mitchz95]

I got this yesterday in 0.24. My spaceplane was in orbit, and the mission log says they crashed into the Launchpad.

[General Rarity]

I find that it tends to happen during physics warp.

Ex. Shuttle Columbus ran out of fuel, and had to use RCS to de-orbit. I physics warped and Columbus decided that he must crash into the VAB somehow... lost half of my life-support and a whole wing.

[luizopiloto]

this happens since 0.22... and they didn't fixed this @#$% bug Yet... and it's even worse now... everytime you deorbit a craft it have a 50% chance of begin destroyed by the VAB or the Launchpad... from space or close to runway... Some ppl reported this things happens when orbting the Mün... ¬¬"

[Torquemadus]

The triangular wing parts sometimes fall off my planes.

[Commander Jebidiah]

this is aknown issue but you were lucky enough to survive it some people have hit the VAB while in orbit

[steve_v]

Sooo, I see this is a "known issue" and it looks like it's still around. My spaceplane just "collided" with the launchpad on landing... while ~1/2 way down the *runway*

Does anyone know what's causing this?

[Claw]

No, unfortunately we don't know what's causing it. I have suspicions but this bug has only happened to me once (and I space-plane fairly often).

The problem isn't proving it exists (there's no shortage of pics of explosions or broken craft), but I can't seem to get people to explain their habits so that I can reliably reproduce this bug. I even tried creating a whole thread dedicated to this topic where people could post their mission profiles. It had zero responses.

If you guys really want this bug to get looked at harder, then the bug reports for this one must be extremely detailed. "I built a plane, flew it to space, came home and BOOM." doesn't help on this one. I already know that flying a space plane to space and back is the gist of the problem.

The kind of reports that are needed are "I built a plane, pictured here [insert pic], which has these kinds of parts. I flew a usual space plane profile into LKO, then transferred to the Mun at the first available transfer opportunity. I established orbit, did some science and released a small probe, then flew home. This happened 2km from the runway. No save/reverts/quickloads, and I never left the mission."

Incidentally, that is the type of mission profile I suspect has most of the necessary elements in it. So if that sounds like a familiar profile, please start filling in the gaps. This bug is annoying, and surprises people out of nowhere. But until it can be reliably reproduced, it's going to get left in a "yep, it's a known thing" status.

Cheers,

~Claw

[Farex]

Greetings!

I also experienced this. Please check this, could be the reason for the random wing-crashes:

When your Wings crash, please go to your Spacecenter and check your launchpad. There should be a little bit of damage to it, like a destroyed tank besides the pad or a destroyed flagpole. The launchpad itself is intact, but parts of it's model are damaged /destroyed.

Every time I have a damaged launchpad, my wings keep on getting crashing when my SSTOs come back to the KSC for a landing. After repairing the damage to the launchpad, this bug never occurred again until my launchpad was damaged again.

Needs confirmation.

[Renegrade]

The problem isn't proving it exists (there's no shortage of pics of explosions or broken craft), but I can't seem to get people to explain their habits so that I can reliably reproduce this bug. I even tried creating a whole thread dedicated to this topic where people could post their mission profiles. It had zero responses.

I sincerely doubt it's a reproducible thing. I've had that happen all of once or twice in literally thousands of hours, and the times it did happen weren't any different from times when it didn't happen.

(My programmer's gut feeling tells me it has to do with the re-origin-ifying system (I think Squad calls it 'Krakensbane'?) and well, whatever places the buildings on Kerbin's surface. It wouldn't surprise me to find out that the craft and the building set weren't glitchily ending up near 0,0,0 at extremely random times, at least in terms of collision data.)

[Padishar]

I sincerely doubt it's a reproducible thing. I've had that happen all of once or twice in literally thousands of hours, and the times it did happen weren't any different from times when it didn't happen.

(My programmer's gut feeling tells me it has to do with the re-origin-ifying system (I think Squad calls it 'Krakensbane'?) and well, whatever places the buildings on Kerbin's surface. It wouldn't surprise me to find out that the craft and the building set weren't glitchily ending up near 0,0,0 at extremely random times, at least in terms of collision data.)

This is a possible cause that has been proposed before but krakensbane only operates when the vessel is above 750m/s which is unlikely when on final approach to the runway...

[vexx32]

The last time I saw a semi-reproducible case, it required a vessel that could go to solar orbit, return to Kerbin, and then land on the runway (or crash into it inexplicably at 20,000m, anyway). It's a very odd one, but making a reproducible case is very much possible. It's just not at all easy.

Trouble is, without a solidly reproducible case it's much like a wild goose chase.

[Starwhip]

Happened to me the other day in my K-Prize challenge.

Flight profile:

Runway to LKO

LKO to Minmus Intercept

Minmus Orbit -- Release probe

Land on Minmus -- Release beacon

Take off

Transfer to Kerbin

Overshoot runway

Turn around on an unbalanced COM-COL

Smack into Launchpad about 10 meters above Runway on landing approach.

Javascript is disabled. View full album

Here's the mission screenshot portfolio.

[KerbMav]

A "flight recorder" option for KSP might help. A mod or built-in protocol function listing everything that happens, loading of new polygons included.

Has anyone been able to repeat the incident? Meaning having a quicksave shorts before it happened, quickloading and have it happen again (and again)?

[Claw]

Happened to me the other day in my K-Prize challenge.

This... This kind of report is perfect. The details in here are what is needed to build a sort of knowledge base. This is a fairly typical profile that I see for this event. And it's nearly always the wings, or the occasional solar panel. Your flight log actually shows more wing parts destroyed than I typically see (probably because that info gets left out).

One question I always have, was there any quicksave/quickloads during this mission?

I have suspicions about what might be causing this, but I need to keep seeing more info like this. A flight logger is a good idea, but the scope of data to collect needs to be trimmed down.

Cheers,

-Claw

Oh yes, and what tier is your runway at?

[Starwhip]

Yes. There were quicksaves. In fact, once I reloaded to land it legitimately (Without the bug, K-Prize doesn't allow parts to be removed). The bug did not repeat.

And it's a sandbox mode game. Destructible facilities.

[Claw]

I guess I mean to ask, did you ever quickload before the wing crash event? If so, do you remember where you were when you did the last quickload?

[Starwhip]

No. I flew the whole mission without a quickload. Had the ship not reentered unbalanced, I would not have had a quicksave.

In retrospect, that was probably a bad idea.

The only quicksaves I made were when I realized the CoM and CoL were out of alignment (The ship entered the atmosphere like an overexcited top) and when close to landing.

[Claw]

No. I flew the whole mission without a quickload.

Yeah, I ask because this is also seems to be a relatively common piece to this puzzle. You can see my thoughts on the subject two months ago here, but I received no response from the forumites. On top of that, this bug is so rare I have been focusing elsewhere.

Again, thanks for the info.

Cheers,

~Claw

[Starwhip]

Hmm. Entering other SOI's seem to trigger the bug?

Interesting. I could see if I could get some reproducible results.

What I think may be happening with the quickload scenario, and the reason that I did not experience the bug when I reloaded, was that I reloaded within Kerbin's SOI. Within the atmosphere, even.

I shall run tests.

[Renegrade]

This is a possible cause that has been proposed before but krakensbane only operates when the vessel is above 750m/s which is unlikely when on final approach to the runway...

Yes, but that's assuming it's working properly, and assuming it's not engaging or partially engaging at random. Plus I'm sure there's more to krakensbane and the coordinate system than Squad has explained. We have here a collision detection error of some kind, and barring weirdness like dangling pointers or such (not supposed to be possible in these buzzword-laden, performance-killing 90s languages, right?), that kinda puts suspicion on anything that has anything to do with coordinates.

Really though, I think Squad probably needs to do a full-press manual code and design audit, challenging every assumption that every line of code makes.

(By the way, I've actually implemented something not unlike Krakensbane in my own historical and more modern designs, and there's ALWAYS some crazy little gotcha waiting when you start nesting coordinates. I've been fighting with these damned floating point numbers since the 80s...)

Anyhow, my own experiences with the attack of the VAB were both simple, light planes that were basically flying around KSC exclusively at short range, but I've seen this happen in many different places, from LKO, all the way to halfway to Jool on ShimmyTheJJ's stream (his entire large transfer craft impacted with the VAB)

[Daveroski]

I suspect that the culprit can be narrowed down still further.

It seems, at least to me, to be a wing fault.

Maybe the questions we might ask, instead of "Why is it doing such and such?" could be, "Why isn't it ever starting with the root part?"

Why not the Command pod? Why not the air intake in front of the command pod? (in the case of an in-line command pod).

It doesn't seem to happen with an under carriage unless it is wing mounted and part of the cascade or a communications device which as the undercarriage is usually radially attached to the central stack.

I started quick-saving before a re-entry because it happened to me. On a return from Eve orbit. Before that I was willing to accept disaster through pilot error and deliberately did not save the game so that the mission would feel more dangerous and exciting. Having the wing break off from hitting the VAB while still on approach destroyed my mission and upset me enough that I now that that precaution. Now if I make a mess of a landing, the temptation will be to quick-load. I think I will be strong enough to resist it but if the mission is involved and complex, I don't know. Perhaps I'll falter.

It seems to be consistently happening with a radially attached wing part and the children of that part.

There seems to be a verifiable constancy. There must be a reason why it is these parts and children of these parts and not others.

Maybe we should ask, "Has anyone had the same problem with a command pod or a ventrally attached communications device?" If Not.. Why not? In the code, what affects them differently or not at all?

[basic.syntax]

I suspect that the culprit can be narrowed down still further.

It seems, at least to me, to be a wing fault.

Maybe the questions we might ask, instead of "Why is it doing such and such?" could be, "Why isn't it ever starting with the root part?"

Great point! I've noticed same: always starts with a wing. And since an aero overhaul is incoming, it will be interesting to watch if this problem mysteriously disappears.

[vexx32]

I don't think it's an aero issue, to be honest, but I suppose it is a possibility.