MonsterMash v1.5e

[}TCP{Wolf]

Well the new version is installed, mod page updated as is FTP.

Since it requires(!!!!) UTeamFix now, the days of spawnkilling monsters are over (if run in native mode)...

I think you will love it... it makes perfect reuse of UTeamFix, looks better and should run more stable.

The server crash / freeze bug could NOT be hammered out but I gave it less reasons to occur

[}TCP{Wolf]

I am pasting this for Zombie, it is a mail I received from him:



Hi Wolf,

This is the same guy asking about you ServerDog mod months back. Just
wanting to contact you about an issue you mentioned about the MonsterMash mod
and an old issue you've mentioned. I would have posted after the monstermash
announcement but TCP forum blocks average posters in that section from
replying.

Quote "The server crash / freeze bug could NOT be hammered out but I gave it
less reasons to occur."

I have reasonable suspicion that the Server crash/hang problem is a very
old original Unreal bug a nothing with the mod actually creating it. The
issue also mostly always occurs on Coop servers since there are other spawned
monster AI pawns controlled by the server in the map. Since the monstermash
spawns AI controlled pawns like a coop server that's why I think they are
tied.
I'm no programmer but I like to do bug testing and when I hit problems
with any software I test how it was caused. When running my own (Zombie's
DeadZone) Coop server I discovered how summoning monsters and then having
them attack you before disconnecting helps create the server freeze. For
instance, summon about 5 to 10 skaarjwarrior in a room and have them all
attack you while you have a lot of health (like 5000 so ya can't die). When
you get about 3 or more monsters physically attacking you, disconnect from
the server at that moment and I guarantee you will see a server hang. The
monsters that were attacking cause some sort of server infinite loop when the
pawn they were attacking no longer exists. However!, this server hang problem
never happens in a Listen Server, for whatever reason I don't know. Also, the
problem doesn't occur if a player suicides or dies before disconnecting after
a monster attack on a dedicated server as well.
I've talked to the mod creator of JCoop for Coop servers about this as
well but Joss hasn't really attempted to fix it after explaining it. Anyway,
some of the possible ways to fix it we discussed were putting some code in
for the Logout() function for when a player disconnects. Perhaps have the
playerpawn automatically die before the client disconnects? or even supply
some code which could possibly address how monster AI attacks in the script
for when the pawn suddenly no longer can be accessed.

If you have any reason to believe the issue in your mod is just like the Coop
problem you might wanna consider releasing a separate mod package for a
broader audience if you can correct the issue.


-ZD

[}TCP{Wolf]

Okay number one, I changed the forum permissions and I hope regular users can now REPLY into news threads (but not post new ones).


I actually had the idea that it might not be the mod, but I wasn't sure WHAT exactly caused the problem.
I did notice that when bots leave or player disconnect, the freeze bug would occur more likely, but what I checked was logout/destroy functions if there was anything, and yet did not come to the idea it might have something to do with what the monsters are doing at that moment related to the player/bot leaving.

I am grateful for this new inspiration, it will give me new opportunities to check for causes. However, I remember the freeze bug occuring also without any bot or player leaving, faaaar less often though, leaving "players" were usually the freeze moment in most cases.


Now the thing what I did was, that when a player/bot logs out all monsters which are attacking this player get their enemy reference cleared and are given the order to start roaming again.
I do not know if this helps anything, but that was a "shot in the dark in good hope"....

I will however start investigating this a bit more whenever I have time for it... I am still involved in programming lots of other things but this piece of information might be what we need to fix this matter... it is a MAYBE though as a first inspection of the attacking code did not reveal anything obvious. Also I cannot find any obvious reason why it would not happen on listening servers, but only on dedicated... well time will tell (I hope...)

Anyway thanks Zombie