Board index FlightGear Development Aircraft

FlightGear 3.6 RC crashes with Boeing 777 and multiplayer

Questions and discussion about creating aircraft. Flight dynamics, 3d models, cockpits, systems, animation, textures.

FlightGear 3.6 RC crashes with Boeing 777 and multiplayer

Postby legoboyvdlp » Mon Oct 12, 2015 2:08 am

Split off from the topic Please test the 3.6 RC.

Well, it is most likely that your computer does not like the 777; you do not have enough VRAM; enough RAM; a good enough GPU or CPU...
KL-666 showed me how to replace the MP 777 with a low-poly AI model... let me see if I can find it.
Last edited by Johan G on Thu Oct 22, 2015 1:11 am, edited 2 times in total.
Reason: Split off from the topic "Please test the 3.6 RC"; Clarifying that it relates to the 3.6 RC
User avatar
legoboyvdlp
 
Posts: 7001
Joined: Sat Jul 26, 2014 1:28 am
Callsign: YV-LEGO
Version: 2018.3.1
OS: Windows 10 HP

Re: Please test the 3.6 RC

Postby someguy » Mon Oct 12, 2015 3:09 am

Meanwhile, see if moving/deleting the 777 from the Aircraft folder helps.
User avatar
someguy
 
Posts: 1653
Joined: Tue Nov 25, 2008 5:54 am
Location: USA
Version: 2016.2.1
OS: Mac OS X 10.11

Re: Please test the 3.6 RC

Postby legoboyvdlp » Mon Oct 12, 2015 12:32 pm

User avatar
legoboyvdlp
 
Posts: 7001
Joined: Sat Jul 26, 2014 1:28 am
Callsign: YV-LEGO
Version: 2018.3.1
OS: Windows 10 HP

Re: Please test the 3.6 RC

Postby clrCoda » Mon Oct 12, 2015 6:20 pm

Thanks legoboyvdlp and KL-666 for the AI 777 file with lower resource load. Can't wait to try it. :)
Ray St. Marie
clrCoda
 
Posts: 1228
Joined: Wed Apr 07, 2010 11:04 am

Re: Please test the 3.6 RC

Postby jsb » Tue Oct 13, 2015 7:07 pm

I believe a Xeon with 3.4GHz and 8GB RAM *should* be sufficient.
And yes, (re)moving the 777 folder kind of 'solves' the problem, except I cannot see it anymore *sigh*
jsb
 
Posts: 219
Joined: Sat Oct 25, 2014 8:17 pm
Location: Hamburg, Germany
Callsign: D-JSB
Version: next
OS: Win7/Linux

Re: Please test the 3.6 RC

Postby legoboyvdlp » Tue Oct 13, 2015 7:20 pm

Jsb, did you see my post above?
Just copy/paste into your FGDATA.
User avatar
legoboyvdlp
 
Posts: 7001
Joined: Sat Jul 26, 2014 1:28 am
Callsign: YV-LEGO
Version: 2018.3.1
OS: Windows 10 HP

Re: Please test the 3.6 RC

Postby jsb » Wed Oct 14, 2015 4:55 pm

Yes, thanks. That works.

Edit: interpolate works, problem was a side effect
jsb
 
Posts: 219
Joined: Sat Oct 25, 2014 8:17 pm
Location: Hamburg, Germany
Callsign: D-JSB
Version: next
OS: Win7/Linux

Re: Please test the 3.6 RC

Postby KL-666 » Sat Oct 17, 2015 2:03 am

After reading this 777 discussion, i tested 3.6 again.

- Model viewing the 777 crashed fg3.6
- Making 777's into AI and viewing did not crash
- Viewing other planes never crashed fg3.6

Ergo, there is something in fg3.6 or the 777 that causes a crash when viewing the 777 from the outside. So everyone involved in either fg3.6 or the 777, think hard about the changes you made that causes fg to crash when viewing 777's over MP.

Oh, and do not start about system resources, i can hold 100 777's in my memory.

Kind regards, Vincent
KL-666
 
Posts: 784
Joined: Sat Jan 19, 2013 1:32 pm

Re: Please test the 3.6 RC

Postby sanhozay » Sat Oct 17, 2015 8:51 am

Crashes are unlikely to be fixed unless someone who has the problem can compile from source with debug and consistently produce a backtrace that identifies the line of code that causes the crash.

What's the recipe for the crash? Is it as simple as loading up a 777 (any variant) and switching to model view? Or does it have to be someone else's 777 viewed over MP in model view?
sanhozay
 
Posts: 1207
Joined: Thu Dec 26, 2013 11:57 am
Location: EGNM
Callsign: G-SHOZ
Version: Git
OS: Ubuntu 16.04

Re: Please test the 3.6 RC

Postby Richard » Sat Oct 17, 2015 9:05 am

I've had a few lockups on MP with 3.6 RC - I thought that the 777 was the cause but I've also had it when no 777's are within 100 miles. I have seen console load errors (missing textures / models) that might be related.

If I get a lockup I will run FG direct from VisualStudio (next/3.7) and at the same location with the same players (e.g. a FGUK flightnight) my built version doesn't lockup. I haven't updated my libraries or done a Cmake rebuild for a while - so it might be a 3rd party dependency.
Richard
 
Posts: 706
Joined: Sun Nov 02, 2014 10:17 pm
Version: Git
OS: Win10

Re: Please test the 3.6 RC

Postby KL-666 » Sat Oct 17, 2015 11:47 am

Hello Sanhozay,

The crashes are strictly over MP. I can be flying a 777 myself and look at myself. But if i Model view another 777 over MP the crash happens immediately.

The sequence of events is that i get to see the other 777 for a minute, while fg is frozen. And then it crashes. Unfortunately there is nothing about it in the log.

LOD settings are:
- Aircraft: 10000 m (the default i believe)
- Interior: 1 m (i do not want to be hindered by interiors)

Kind regards, Vincent
KL-666
 
Posts: 784
Joined: Sat Jan 19, 2013 1:32 pm

Re: Please test the 3.6 RC

Postby Hooray » Sat Oct 17, 2015 1:48 pm

as was said, it would be good to have a reproducible test case, one that preferably does not use many other features/settings. For starters, it would be good to exclude MP factor, e.g. by changing tanker.nas to instantiate a 777 model and see if that reproduces the issue. If it does, other factors should ideally also be excluded, i.e. location, main aircraft etc.
And once you have something that you can post here (as in exact instructions/settings), we can try to come up with a gdb backtrace.

If the tanker.nas approach shouldn't work for some reason, you could hook up two fgfs instances manually and check your working theory by putting a 777 in front of the ufo.

For instructions, see: http://wiki.flightgear.org/index.php?ti ... or_testing
Please don't send support requests by PM, instead post your questions on the forum so that all users can contribute and benefit
Thanks & all the best,
Hooray
Help write next month's newsletter !
pui2canvas | MapStructure | Canvas Development | Programming resources
Hooray
 
Posts: 11329
Joined: Tue Mar 25, 2008 8:40 am

Re: Please test the 3.6 RC

Postby KL-666 » Sat Oct 17, 2015 2:19 pm

Hooray wrote in Sat Oct 17, 2015 1:48 pm:it would be good to have a reproducible test case


You got one, but you do not want to do it. Go fly MP and model view 777's in your surroundings.

Hooray wrote in Sat Oct 17, 2015 1:48 pm:And once you have something that you can post here (as in exact instructions/settings), we can try to come up with a gdb backtrace.


Really? Only then? Well i do not have a problem you need to solve. I work around it by letting 777's show up as AI planes. You (the developers) have a problem when you release this thing. I'm only tipping you where your problem will be. Just test a bit more on MP yourselves.

Kind tegards, Vincent
KL-666
 
Posts: 784
Joined: Sat Jan 19, 2013 1:32 pm

Re: Please test the 3.6 RC

Postby Hooray » Sat Oct 17, 2015 2:26 pm

KL-666 wrote:Well i do not have a problem you need to solve. I work around it by letting 777's show up as AI planes. You (the developers) have a problem when you release this thing. I'm only tipping you where your problem will be. Just test a bit more on MP yourselves.


umm, let me try to phrase this politely: a MP test-case is generally understood to be hardly "reproducible", i.e. if you want to see this fixed, you need to come up with a narrowed-down test case. I told you of several options to check your working theory (if it's MP related, if it's AI related (the MP system is using the AI system internally) and if it's related to the aircraft you mentioned).

In other words, it is your call - but please don't expect anybody to care given your attitude.
If you would care, you would make a proper bug report and attach the requested information.

Personally, I don't give a damn ... and in fact, most "developers" don't even use MP at all.

BTW: and thank you for proving my point (the posting following mine came from you ...):

Subject: Access to MPserver12
Hooray wrote:You only need to take a look at the forum (or the mailing list) and compare the number of active contributors with the number of MP users, and then look at their attitude ("entitlement"). The subset of both groups is extremely tiny, and usually those were FG contributors before they became MP users. The opposite seems pretty rare, while the number of MP users causing tons of work on the forum (think support) is remarkably high.

The whole "gaming" aspect of FG is not working in its favor unfortunately (cost/gain).
That is not to say that anybody is considering to phase out such functionality, it just seems increasingly obvious that MP/VA users cause a ton of work for contributors (excluding obviously those who are contributing in some way)
Please don't send support requests by PM, instead post your questions on the forum so that all users can contribute and benefit
Thanks & all the best,
Hooray
Help write next month's newsletter !
pui2canvas | MapStructure | Canvas Development | Programming resources
Hooray
 
Posts: 11329
Joined: Tue Mar 25, 2008 8:40 am

Re: Please test the 3.6 RC

Postby KL-666 » Sat Oct 17, 2015 2:36 pm

Hooray wrote in Sat Oct 17, 2015 2:26 pm:given your attitude


Likewise to you.

Kind regards, Vincent
KL-666
 
Posts: 784
Joined: Sat Jan 19, 2013 1:32 pm

Next

Return to Aircraft

Who is online

Users browsing this forum: No registered users and 21 guests