Michael Wells
Assigned
Reported
Comments
Markku, if you can get me the .jar with that I'd be happy to help test. I have a server right now we're mercifully needing a fix for.
I would also suggest to make it so that the mob would path away randomly before attempting to find another route if the player hasn't moved. That way, if it is a route error, the new location should solve it.
Also, does this apply to villagers too? I've noticed that villagers seem to love steps.. They will literally just walk up and down steps all day and night if they are indoors (like in a chapel or library) . I would suggest the same thing there in that routing as well... If they can't find an endpoint for what they are wanting to go (null route) their endpoint should change or they should path elsewhere and reattempt the route (say max 3 attempts before picking another block to path to).
I see, this is a bukkit issue. I just cranked an old server back up and didn't realize bukkit was now using its own minecraft API. When I disable bukkit, it connects fine.
My server is too having the same issue:
Quad Xeon 4.2Ghz quad core system with 18GB ram.
Top indicates Java isn't using that much more CPU than normal and we have MySQL and Minecraft installed on SSD and minecraft runs from a 10GB ramdisk. This kind of lag shouldn't happen, this server borders on supercomputing.
I can also confirm that it is a null pathing issue. When a zombie agros a player or villager and cannot find a route, whether it be one zombie or 100, the server lags.
I'm going to assume and go out on a limb, not being a java expert but a PHP, C++ expert that the issue lies in too much CPU time being taken to find a new route.
A simple fix for this would be for a zombie to try to find a route once or twice, if that fails, it either attempts to find another target, despawns, or idles as an ambient monster.
Please fix this asap, as many server owners like myself are very much considering rolling back to 1.5.
As a map maker, PLEASE fix this soon!