Exploding and bending ships while time warping.

[Fengist]

I read this:

Ships with the claw, especially when clawed to something, sometimes experience spontaneous explosions or bizarre bending when using time warp. The devs are aware and are working on it, but there is currently no short term solution to this.

This is happening to all of my ships. I've never used the claw. What I've noticed is that it happens after I've launched and reached LKO and am time warping to apogee. Vessel length makes no difference. I've had it happen with purely stock parts as well as mod parts. Longer length vessels tend to bend before exploding. I have noticed that the Kerbin gravity has some influence on this. If the ship is horizontal to Kerbin, it will bend with the aft end bending directly toward Kerbin while the bow remains horizontal. If you place the vessel vertical facing away from Kerbin, it will stretch it like an accordion but not bend and break. In all cases, the ship becomes useless afterward.

It happned again just a few moments ago and this time, the Map showed the vessel accelerating but the vessel view showed it's speed as unchanging.

The workaround I've found for now, quicksave before time warping. If the ship bends and/or explodes, exit KSP and restart. Reload the quicksave and it's usually fixed.

[MisterFister]

F9 rescue sounds like a good workaround, but I've read elsewhere on the forums that this too can sometimes fail to address the issue. Perhaps a .craft and a screenie? (I've yet to encounter this problem without the klaw, I think.)

I wonder, are you in atmosphere before engaging time warp to apo? I doubt you would be, but if so, you're not in true time-x, you're in phys-x, which carries a whole host of other issues.

[Fengist]

I'll try to grab a screen shot next time but the craft file won't matter. Once it starts doing it, it'll do it with every craft. I've even made a stock craft with 1 engine, 10 fuel tanks and a pod and sent it toward orbit just to test. The only way to get it to stop is exit out of KSC and reload.

Nope, I'm making certain I'm above the 70km mark This is time warp, not physics.

It bumps up to X5, bends the craft then drops back down to x1.

[Fengist]

Ask and ye shall receive.

This was a very large craft I've been working on so it happened in slow-motion.

You'll notice it's as I described. The 'control pod' remains horizontal while the rest of the ship bends toward Kerbin.

Before warp...

After...

[Fengist]

Even if you stop the warp it keeps bending it. If you try to warp the parts that are left, it says it's under acceleration.

Oh, and no claw.

Once one ship bends like this, any ship I send up and try to time warp will bend. I know, I spent over 5 hours last night thinking I'd somehow buried a part that was being clipped and causing this. After I restarted KSP, everything was fine again for a while.

[Fengist]

Just downloaded .90, loaded my quicksave of this ship and it did it again. Tonite when it does it, I'll send up a pure stock ship.

[Claw]

This has been one of those really hard to trace bugs. There doesn't seem to be some sort of obvious trigger. It isn't just time warp by itself, but seems like it's time warp in conjunction with something else...So if you are running into this without the claw, then maybe it has to do with some combination of activities before hand.

[MisterFister]

For further experimentation, I wonder how the root part assignment influences this bug's behavior. Is it the root part that "remains" and the rest of the craft is what "bends"? Try redefining the root part (possible before with a mod, now a stock behavior) and see if the origin of the bending behavior moves to follow the root part.

- - - Updated - - -

Also, perhaps a log output, even though it might not crash or generate an "error" dump.

[MisterFister]

Other thoughts: does camera orientation, camera mode (free / chase / etc.) or camera mod (enhanced chase cam) influence this behavior at all? Craft orientation? Staging distribution / behavior? Part module classes (especially if root part reassignment is a factor)? Time warp activity while on the Launchpad? You already suggest that the problem tends to go away for a while following KSP restarts. I wonder if there's some kind of silent error being thrown, or nulref, that "accumulates" and clogs a buffer of some kind. What about aerodynamic model (stock / FAR / NEAR -- yes I know the problem is after you leave atmo, but the presence of a physics-changing mod like FAR / NEAR might generate a bug like this, idk.) What about Kessler properties, do you have a crowded LKO or otherwise highly populated persistence file? What about RemoteTech -- I know RT2 had some issues for Bob Fitch in his Project Odyssey series on YouTube that related to root parts affecting whether engines fired, and Scot Manley early in Interstellar Quest had spontaneous craft explosions in a version of RT2. What about launch ascent profiles? Do polar orbit launches produce the same issue (physics modeling of the craft with respect to the rotation modeling of the planet surface)? Do you run EVE, BA, or ATM? What about Modular Fuels / Modular Tanks, or any mod that relies on the RSS .dll files (which are surprisingly legion)?

You suggested sending up a test launch of a stock-only vehicle. What about a stock-only launch? Maybe it's an environmental bug moreso than one introduced (or triggered) by any given part or part-mod?

[Claw]

This happens in stock too, although it nealy always involves the claw.

I haven't been able to verify, but my suspicions so far is that this has to do with the kracken's bane fix (or something similar). The craft acts as if the root par is experiencing zero local velocity while the rest of the ship is under its normal local velocities. That would explain the slow bending and resulting ecplosion.

But that is a difficult hypothesis to test because the bug has been so difficult to replicate.

Cheers,

-Claw

[MisterFister]

This happens in stock too, although it nealy always involves the claw.

I didn't mean to imply that this bug wouldn't occur in stock. Simply, this OP seems to be fortunate (?... odd word, I know) to have a somewhat reliable ability to replicate this, since I think he took screenies of the problem AFTER I asked about them. I say, gather as much hard data and full log traces from him while we can to contribute to this effort.

[Crimson Sunrise]

Hm...Looks like a variation of the Cthulu...

[Fengist]

Well, I dunno about fortunate and I dunno about being able to repeat it on purpose but it does happen fairly often and I'll do what I can to help fix this.

You had lots of questions so let's begin.

Right now, that's the only ship I have in orbit. When you asked for screenshots I got 'lucky' that it happened. I loaded up my save game, went to the VAB, launched a random ship, hit F9 and it went to the ship you see in the images, which is the only ship I have anywhere at the moment. I turned on the RCS to orient it horizontally to Kerbin and hit the warp button. You see the results.

Immediately after that, I downloaded .90, repeated this process and it did the same. Here's a log of that:

KSP Log

It's rather large so you may want to look for "[LOG 11:35:00.547] [FLIGHT GLOBALS]: Switching To Vessel Jool 1 ---------------------- " which is when I loaded the quicksave.

Current mods are:

Active Texture

B9 and it's associated bloat

Core Anvil 172

Firespitter

KAS

Kethane

LLL

Infernal Robotics

An OLD copy of RBI Tracks

About the only thing I can say I consistently have done in all of the launches where this has happened is to use an Anvil as the launch vehicle. I doubt this has any relevance though as the last time, no anvil parts were used.

Actually, that's wrong, there was an Anvil upper stage in those screen shots. Dunno.

Not using any physics mods atm.

Once I get it to do it again, I'll send up a pure stock and test a polar launch. Thus far, I've been launching, rotating at around 7-10km toward 90 deg and accelerating to 1,000m/s. I then start switching between the ship and map to keep my angle of ascent the way I want it and wait for apogee to reach where I intend. At first, I was blindsided by this as it would do it while warping in the map and I'd switch back to just the command pod. After that, I started quicksaving before warping.

I typically only change camera orientation when I'm docking so it stays default. I don't warp on the pad.

And I just repeated that same process of loading the quicksave... everything warped just fine.

I'll do some more testing and try to get more logs of this happening.

Edited December 17, 2014 by Fengist

[x5060]

Yeah, I had this issue in .25 as well. I would say its either tied to SAS and time warping or the direction whatever the control part is moving at the time of warping. I cant really launch anything that leaves kerbin orbit as anything more complex than that seems to always be affected by this bug.

[Fengist]

Yeah, I had this issue in .25 as well. I would say its either tied to SAS and time warping or the direction whatever the control part is moving at the time of warping. I cant really launch anything that leaves kerbin orbit as anything more complex than that seems to always be affected by this bug.

I thot so too, but you'll notice in the screen shots, SAS is off, RCS is on.

[Fengist]

And here it's doing it to a pure stock vehicle.

Warping

Testing a polar launch now.

[Fengist]

More results:

At 375km polar orbit

45 degrees N, warped. Bent but didn't break. But after warping it thinks the ship is under acceleration.

Over the pole, Bent but didn't break.

Another note is that on this launch I used about 33% of my fuel so it's mass was considerably less as well.

Both of these were at much higher altitudes than before (I got distracted and overshot 70km)

Testing another equatorial launch at much higher altitudes to see if the Kerbin SOI is having an influence.

Edited December 17, 2014 by Fengist

[x5060]

I thot so too, but you'll notice in the screen shots, SAS is off, RCS is on.

Well SAS has to be tied to directional updates, for some reason those directional updates are no longer made for the root control piece. So I guess it isnt really SAS per se but something affecting the routine that updates the rotation of the ship.

Also, all of my ships are stock, I don't really use any mods or add-ons.

[Fengist]

More testing. (I'm quicksaving before each of these tests and f9 after each)

750km equatorial: Bent, didn't break... full of fuel. Still says ship under acceleration after warping.

1,000km equatorial: Bent, didn't break. Bending less than at 750.

1,500km equatorial: Did not bend, did not break, but now the ship is accelerating (or at least the AP is increasing) even though I didn't engage the engines!?!? Oh and this is really strange. Subsequent warping... the ship remains roughly where it was while everything else warps.

2,000km equatorial: Did not bend, did not break. Everything except the ship warped and according to my screen, it's warping at 50x yet it's altitude barely changed. Ship is still usable after shutting down warp.

2,500km equatorial: Something rather unexpected:

And here's the 'under acceleration effect.'

At 1,000km I warped. The two 'booster's I had strapped to the sides of it were basically following beside the ship as I didn't power the ship up, other than rcs.

Before warping the AP was around 2.600km. About 15 seconds after warping you can see the apogee for the boosters is unchanged but now the ship, even though I engaged nothing, has accelerated and has an AP of 3,000km.

After the 2,500 detonation I went back to the space center, exited to the load game menu, loaded my saved game, went to the VAB, launched the autosave ship, hit F9 to get back to where this ship was at 2,500km. I warped, it exploded.

I then completely exited the game, reloaded ksp, loaded my save game, went to the VAB, launched the autosave, hit F9 to reload this ship at 2,500km, hit warp and... it warped exactly as it should.

As I said, I spent over 5 hours the other day thinking I'd somehow screwed up my craft by part clipping. I discovered, it wasn't my ship. And no, I still can't reproduce it at will but when it starts doing it, the only thing that stops it is completely restarting KSP. Luckily for me, reloading quicksaves works after a restart.

Edited December 17, 2014 by Fengist

[Fengist]

One other thing, my log is now expanding exponentially as I'm doing this testing with the following. It's now at 42 megs!

[EXC 20:08:57.366] MissingFieldException: Field '.Part.uid' not found.

InternalProp.OnUpdate ()

InternalModel.OnUpdate ()

Part.InternalOnUpdate ()

Part.Update ()

Edit --- The log reached over 54 megs, mostly repeating the lines above every few miliseconds in the space of the 30 minutes I was flying that ship ---

Let me know if you need any more information or tests and when it does it again, I'll try to give examples.

Edited December 17, 2014 by Fengist

[Crimson Sunrise]

Did this happen without any mods on? Cause I'm pretty sure the code of 0.90 is different than 0.25. Even if KSP didn't told you which mods aren't compatible, I wouldn't trust that 100%.

[Fengist]

Did this happen without any mods on? Cause I'm pretty sure the code of 0.90 is different than 0.25. Even if KSP didn't told you which mods aren't compatible, I wouldn't trust that 100%.

There were no mods on that particular ship whatsoever. Everything was completely stock down to the RCS ports. I do have mods installed yes.

I'm hesitant to uninstall them just yet as if it is one particular mod (and yea, I'm getting several that .90 says are incompatible) then it would be real nice to know which one. Unfortunately, since I can't repeat this at will yet I couldn't be certain of the results unless it started doing it again.

One idea I have (because .90 says that Kethane is incompatible with Unity and I have Kethane on an orbiting ship.) I want to get it to do it again, save, delete the ship with Kethane and see if it still does it.

Edited December 17, 2014 by Fengist

[Fengist]

Well SAS has to be tied to directional updates, for some reason those directional updates are no longer made for the root control piece. So I guess it isnt really SAS per se but something affecting the routine that updates the rotation of the ship.

Also, all of my ships are stock, I don't really use any mods or add-ons.

All I can say is that I did turn SAS on a few times during the testing and tried it with SAS on and off. I didn't see any difference. Apparently this is having something to do with the SOI's gravity well and/or the distance from the SOI. The lower the altitude, the more pronounced the effect (with the exception of the catastrophic detonation at 2,500km). One thing I'm going to try to do is get this test ship into orbit around the Mun and leave it (I'll have to do this while warp is working). Once it starts doing it again, I want to know if this is just something that happens in Kerbin's SOI (which I doubt) or any SOI.

[Claw]

I didn't mean to imply that this bug wouldn't occur in stock. Simply, this OP seems to be fortunate (?... odd word, I know) to have a somewhat reliable ability to replicate this, since I think he took screenies of the problem AFTER I asked about them. I say, gather as much hard data and full log traces from him while we can to contribute to this effort.

Yeah, sorry. I didn't mean to sound dismissive. Just trying to make sure this doesn't get written off as an add-on problem.

All I can say is that I did turn SAS on a few times during the testing and tried it with SAS on and off.

Of the testing I've done, I nearly certain it isn't clipping related. It doesn't act like other clipping bugs. I'm pretty sure it's not SAS pulling the ship apart either, based on how the ship looks when it's coming undone. Your pictures are exactly the same as the bug on stock installs.

One thing that would also help out when you're posting pictures is to specify which is the root part, and which is the second part that you placed. I'm also thinking there might be some correlation between where the ship comes apart, and where the first two parts are. Seems fairly common that it'll bend at the joint off the second part. And often when it comes apart, the second part gets ripped out completely.

Edited December 17, 2014 by Claw

[Fengist]

For further experimentation, I wonder how the root part assignment influences this bug's behavior. Is it the root part that "remains" and the rest of the craft is what "bends"? Try redefining the root part (possible before with a mod, now a stock behavior) and see if the origin of the bending behavior moves to follow the root part.

- - - Updated - - -

Also, perhaps a log output, even though it might not crash or generate an "error" dump.

I just realized what you're suggesting. Since that first ship I posted has multiple places where it can be controlled, I'll test that theory next time it starts doing it. BUT, I think I've already experimented with that. I do recall with the ship in the first screenshots, I started noticing it that it would bend in different places depending on what part I had where. At times, it would bend in the middle, other times, I'd be left with the root part and a couple of pieces behind it. But the root part always stayed in place while the rest bent.