"Black Hole" Server/map bug. What is it, what causes it?

[R1CH]

Q2 only informs the client about an entity if it has one of 4 things: a model, a sound, an effect or an event. What's happening here is someone is disconnecting with a railgun in hand (railgun sets the player to have a constant hum sound). The mod isn't properly removing the sound once the player disconnects, so Q2 still sends the entity to the client, complete with the last used bounding box which is why it feels like a invisible wall. I believe this is a bug in the 3.20 Game DLL.

[yahoo]

I still believe in my ghost theory..

[peewee_RotA]

I still believe in my ghost theory..

Don't forget about Occum's Razor


If you have equal evidence to support two opposing theories, then the simplest answer must be HAX.

[[BTF] Reflex]

Q2 still sends the entity to the client, complete with the last used bounding box which is why it feels like a invisible wall.

Thanks for the input R1CH
But,  why can't you walk full into it and stop like you could do with a player?




Peewee,  I think you meant Occum's R4zor

[peewee_RotA]

Peewee,  I think you meant Occum's R4zor

Good catch. I forgot how he customized that name. 

[Acer]

There exists a cvar called "zombietime" and it sets the amount of time in minutes (2 by default) before a frozen character is removed from the map.
Normally when people are frozen it`s because they lost their connection or it is timing out. But sometimes people just smash their F10 key while in the middle of movement and action and it looks as if the server only removes the visible models but not the other effects.  Could "zombietime" have anything to do with it and would it make a difference if it was set to something like 0.5 instead of 2 minutes?

[Punk_FAS]

Oh, btw, I've run into this problem a couple times since I made this post.  Once it was right next to the yellow armor box in the mega room on DM1.  Apparently this doesn't show up in a recorded demo.  In the 2 demos I made, I pass over and back through the laggy areas without any appearance of obstruction in the demo.

[R1CH]

There exists a cvar called "zombietime" and it sets the amount of time in minutes (2 by default) before a frozen character is removed from the map.
Normally when people are frozen it`s because they lost their connection or it is timing out. But sometimes people just smash their F10 key while in the middle of movement and action and it looks as if the server only removes the visible models but not the other effects.  Could "zombietime" have anything to do with it and would it make a difference if it was set to something like 0.5 instead of 2 minutes?

The cvar you are thinking of is "timeout", in seconds. zombietime is related to the netcode and has nothing to do with frozen players.

[Acer]

In that case, setting "cl_timeout" to 30 seconds instead of 120 is going to make those lagspots disappear sooner I guess...

But what do they mean then by "frozen character" if it`s related to the netcode?

zombietime
Default: 2
Description: The amount of time in minutes before a frozen character is removed from the map.
http://www.cs.rochester.edu/research/quagents/console_commands.htm

[yahoo]

"zombietime"

Bwahahahaha. Welcome to the realm of The Twilight Zone.

[R1CH]

In that case, setting "cl_timeout" to 30 seconds instead of 120 is going to make those lagspots disappear sooner I guess...

But what do they mean then by "frozen character" if it`s related to the netcode?

zombietime
Default: 2
Description: The amount of time in minutes before a frozen character is removed from the map.
http://www.cs.rochester.edu/research/quagents/console_commands.htm

Not all information on the Internet is correct. Also cl_timeout just sets how long before you disconnect from a unresponsive server, not how long the server waits. That cvar is "timeout".

[Acer]

I believe this is a bug in the 3.20 Game DLL.

Also cl_timeout just sets how long before you disconnect from a unresponsive server, not how long the server waits. That cvar is "timeout".

OK got it. By the way "timeout" is also in that command list and the description is correct, it was just my understanding that was incorrect.