Anyone else experiencing crashes looking like this?
Crashlog: [url]http://pastebin.com/33mi1tUb[/url]
Willox said it was a model, so I removed a vehicle I added recently (even though it's a very popular workshop addon which hopefully doesn't crash for everybody).
After removal of the vehicle model the crashes decreased. It's not every day anymore, there can be a few days without a crash and then suddenly two-three crashes in one day - often with a short time between (1 hour).
Anyone have a tip how to debug this? Apparantly it's caused by TraceLine use. I logged all of them, but the few crashes I had didn't show any pattern.
[URL="https://github.com/ValveSoftware/source-sdk-2013/blob/master/mp/src/game/shared/animation.cpp#L522"]Can we get this removed?[/URL]
[t]http://i.imgur.com/tLgUBGs.png[/t]
Or atleast print what model is causing problem, printing causes several frame drops and con_filter doesn't work on srcds
[QUOTE=gonzalolog;50955490][URL="https://github.com/ValveSoftware/source-sdk-2013/blob/master/mp/src/game/shared/animation.cpp#L522"]Can we get this removed?[/URL]
[t]http://i.imgur.com/tLgUBGs.png[/t]
Or atleast print what model is causing problem, printing causes several frame drops and con_filter doesn't work on srcds[/QUOTE]
That's important for making SNPCs, but it's annoying that you can't know what model and animation is causing it though. It should print the model name and animation. Also it would be good too if it printed the entity as well
It'd be great if it was actually [URL="https://github.com/Facepunch/garrysmod-requests/issues/691"]exposed to the Lua state[/URL] as well so not only would pose param calculation be a lot simpler, but it could be triggered to see what model/anim was causing the error.
[QUOTE=LegoGuy;50936420]Honestly, I don't see the point of redirect servers. Isn't that just a waste of resources and time?
[editline]23rd August 2016[/editline]
Like I'm serious, someone tell me why they should exist.[/QUOTE]
Seems like Facepunch does not agree either.
[IMG]http://i.imgur.com/xRKBI0h.jpg?1[/IMG]
[QUOTE=Datamats;50967626]Seems like Facepunch does not agree either.
[IMG]http://i.imgur.com/xRKBI0h.jpg?1[/IMG][/QUOTE]
-snip I'm dumb, he's talking about the redirect servers-
[QUOTE=Datamats;50967626]Seems like Facepunch does not agree either.
[IMG] http://i.imgur.com/xRKBI0h.jpg?1[/IMG][/QUOTE]
Robotboy forgot he doesn't care
[QUOTE=Gurrazor;50930399][IMG] https://dl.dropboxusercontent.com/u/10974609/Thriving%20Ventures/Miscellaneous/Permission.jpg[/IMG][/QUOTE]
[QUOTE=Kevlon;50968438]Robotboy forgot he doesn't care[/QUOTE]
Kevlon forgot how to use his eyes since the picture clearly states "I do not mind as long as [b]its an actual srcds instance and doesn't fake any info.[/b]".
[editline]29th August 2016[/editline]
[del]Not only that, they are also collecting ad revenue for every hit on those "redirect" servers.[/del]
[QUOTE=Robotboy655;50968518]Not only that, they are also collecting ad revenue for every hit on those "redirect" servers.[/QUOTE]
I do?
[QUOTE=Robotboy655;50968518]Kevlon forgot how to use his eyes since the picture clearly states "I do not mind as long as [b]its an actual srcds instance and doesn't fake any info.[/b]".
[editline]29th August 2016[/editline]
Not only that, they are also collecting ad revenue for every hit on those "redirect" servers.[/QUOTE]
I agree with the first part, however I'd like to see proof of the latter statement.
[editline]29th August 2016[/editline]
[QUOTE=gonzalolog;50955490][URL="https://github.com/ValveSoftware/source-sdk-2013/blob/master/mp/src/game/shared/animation.cpp#L522"]Can we get this removed?[/URL]
[t]http://i.imgur.com/tLgUBGs.png[/t]
Or atleast print what model is causing problem, printing causes several frame drops and con_filter doesn't work on srcds[/QUOTE]
Could you put it in a GitHub issue so this doesn't get forgotten? This is really important.
I could've sworn you you used ad f . ly for the loading screen, but I guess either I am wrong or you changed it.
[QUOTE=Robotboy655;50968631]I could've sworn you you used ad f . ly for the loading screen, but I guess either I am wrong or you changed it.[/QUOTE]
Doesn't adfly work on a skipping-to-get-the-content basis? Would that even work in a splash-ad context?
[QUOTE=code_gs;50968814]Doesn't adfly work on a skipping-to-get-the-content basis? Would that even work in a splash-ad context?[/QUOTE]
It wouldn't. I've absolutely no idea what Robotboy is talking about. adfly has never worked as a loading screen. It just wouldn't.
edit: God dammit this is still the fucking next update thread you guys. Really?
Sadly that such semi derails happen often when there isn't much going on with updates.
So let's talk about something else:
[QUOTE]Switched ClientLeafSystem from ushort to uint (from upstream)[/QUOTE]
What does that mean? Is it something about viscluster having higher limits or something?
You know how people always say not to use Material every frame? Why not make Material cache the return value internally?
It's already being detoured (clientside only, but that's where it matters anyway) - [url]https://github.com/garrynewman/garrysmod/blob/master/garrysmod/lua/includes/extensions/client/render.lua#L3-L18[/url]
and here's how it's done [url]https://www.lua.org/pil/17.1.html[/url]
Before anyone says anything, I know that Material has a very small/negligible performance hit, but I can't think of a reason not to do this assuming it doesn't break anything. Changing a key value in a material will affect all instances of that material anyway from my tests, so might as well make them all share the same instance. And when it's not used it gets garbage collected.
[QUOTE=NeatNit;50974526]You know how people always say not to use Material every frame? Why not make Material cache the return value internally?
It's already being detoured (clientside only, but that's where it matters anyway) - [url]https://github.com/garrynewman/garrysmod/blob/master/garrysmod/lua/includes/extensions/client/render.lua#L3-L18[/url]
and here's how it's done [url]https://www.lua.org/pil/17.1.html[/url]
Before anyone says anything, I know that Material has a very small/negligible performance hit, but I can't think of a reason not to do this assuming it doesn't break anything. Changing a key value in a material will affect all instances of that material anyway from my tests, so might as well make them all share the same instance. And when it's not used it gets garbage collected.[/QUOTE]
Correct me if I'm wrong, but that would probably break things that modify the material.
I.E. anything that uses blur, or maybe you have two materials that use the same path, but you've changed a key value on one of them — would it not conflict then?
[QUOTE=NeatNit;50974526]You know how people always say not to use Material every frame? Why not make Material cache the return value internally?
It's already being detoured (clientside only, but that's where it matters anyway) - [url]https://github.com/garrynewman/garrysmod/blob/master/garrysmod/lua/includes/extensions/client/render.lua#L3-L18[/url]
and here's how it's done [url]https://www.lua.org/pil/17.1.html[/url]
Before anyone says anything, I know that Material has a very small/negligible performance hit, but I can't think of a reason not to do this assuming it doesn't break anything. Changing a key value in a material will affect all instances of that material anyway from my tests, so might as well make them all share the same instance. And when it's not used it gets garbage collected.[/QUOTE]
Theres absolutely no reason to create a material every frame.
[QUOTE=Zeh Matt;50976201]Theres absolutely no reason to create a material every frame.[/QUOTE]
Yeah, that's why [QUOTE=NeatNit;50974526]Why not make Material cache the return value internally?
It's already being detoured (clientside only, but that's where it matters anyway) - [url]https://github.com/garrynewman/garrysmod/blob/master/garrysmod/lua/includes/extensions/client/render.lua#L3-L18[/url]
and here's how it's done [url]https://www.lua.org/pil/17.1.html[/url]
Before anyone says anything, I know that Material has a very small/negligible performance hit, but I can't think of a reason not to do this assuming it doesn't break anything. Changing a key value in a material will affect all instances of that material anyway from my tests, so might as well make them all share the same instance. And when it's not used it gets garbage collected.[/QUOTE]
[QUOTE=Z0mb1n3;50974733]Correct me if I'm wrong, but that would probably break things that modify the material.
I.E. anything that uses blur, or maybe you have two materials that use the same path, but you've changed a key value on one of them — would it not conflict then?[/QUOTE]
In my tests, when you modify a material it affects all instances of that material.
I see why it would be useful but I feel like it would encourage bad practices, just like people continue to do GetConVarString( "convar" ) ; they would keep doing render.SetMaterial( Material( "mymat" ) ).
And besides, calling Material again reloads the material from disk if there were any changes, if you were to cache that you'd have to call mat:Recompute() or something; and I'm not sure if that even reloads it from disk.
Is there anything that can be done about the [I]massive[/I] client freeze that happens when props that have been materialized with a material that isn't on your client are spawned?
[QUOTE=unrezt;50978163]Is there anything that can be done about the [I]massive[/I] client freeze that happens when props that have been materialized with a material that isn't on your client are spawned?[/QUOTE]
My. God. That is an issue I hope gets fixed quickly, because, one of my friends made a firetower (you can find it in Post Your WIP) and every single time he spawns it, more often than not, it'll make my game hang for roughly 10 seconds. Extremely annoying, especially when coming out of world portals or cordon, where the materials are re-rendered.
(And he says that all of it is hl2 materials :what:)
Any reason why windows defender would want to analyze \modules\halo.lua - This is a first for me, lol.
[t]http://i.imgur.com/YrQQmnb.png[/t]
[QUOTE=Invule;50991924]Any reason why windows defender would want to analyze \modules\halo.lua - This is a first for me, lol.
[t]http://i.imgur.com/YrQQmnb.png[/t][/QUOTE]
Probably random selection.
[QUOTE=Invule;50991924]Any reason why windows defender would want to analyze \modules\halo.lua - This is a first for me, lol.
[t]http://i.imgur.com/YrQQmnb.png[/t][/QUOTE]
Oh no, it has uncovered out secret! In reality we are Illuminati and that files contains some Illuminati code that allows us to turn on your microphone and camera at will.
[editline]3rd September 2016[/editline]
Fun Fact: Your Microphone is probably turned on right now!
[QUOTE=code_gs;50992268]Probably random selection.[/QUOTE]
Combined with lots of people having this file on their system perhaps.
They're looking for Halo leaks
Is there any news on this bug report?
[url]https://github.com/Facepunch/garrysmod-requests/issues/324[/url]
NetworkVarNotify not being called on client after the server sets it?
Or rather, should we just use the SetNWVarProxy and use the NWVars2 system?
[url=http://pastebin.com/wwdDEQUF]any idea why this server in particular is crashing like crazy?[/url] running debian, same code as on the other two servers and it started just a few days ago
and apparently because it's running on linux, there aren't any mdmp files created
[QUOTE=Giraffen93;51000412][url=http://pastebin.com/wwdDEQUF]any idea why this server in particular is crashing like crazy?[/url] running debian, same code as on the other two servers and it started just a few days ago
and apparently because it's running on linux, there aren't any mdmp files created[/QUOTE]
[code]
#0 0x00000050 in ?? ()
[/code]
:v:
this has to be some sort of very bad bug that's either creating a thread at the wrong addresses or a buffer under/overflow or just some general bad stack disalignment somewhere
Sorry, you need to Log In to post a reply to this thread.
