I put a lot of thought in to this. While it MAYBE possible, it'd require some serious time spent from a development perspective. Remember this would be two different servers, meaning two different server instances. There is no shared features so the two servers would need to sync entities, communicate to each other, and likely require extra client side mods. This would be a massive undertaking.
the reason i asked this was because of the recent code found in forge that was causing lag World gen lag fixed in CoFHCore, possibly Forge later? : feedthebeast I think TDT 1.0.5 or 1.0.6 has this fix
*sigh* people attacked my post for saying my mobfarm as a result would only run for 10 minutes a day, this would be because thats how long i'd spend at the mobfarm, and because of NATURAL MOB MECHANICS, Spawn - Minecraft Wiki
Here's my 2 cents: Many of the proposed solutions mentioned here and elsewhere seriously affect game-play. Hansi did a good job at summing up some of the major impacts and potential work-arounds, but ultimately the "solutions" change the game completely. These changes would work, but only because people would become so disenchanted with the server/pack configuration that they would find a more "normal" Minecraft server to play on. To put it another way: The "solutions" are not scalable either. If we were to grow to 100 players on the server, we would encounter the same issues even with three servers handling the loads. For these reasons, I believe the only correct way to handle these issues is to attack the root cause(s). We should perform the testing, identify the largest contributing factors, then implement testing and fixes that do not completely wreck gameplay on the pack. I think that Slind and Willi have been approaching the issue in a manner similar to what I would recommend. They've been gathering performance data and sharing it with each other and the community so that we can solve this issue together. Continuing on in that spirit, I would recommend researching the following: 1. Has Java been considered as a source of some problems? I've heard that Java 8 has addressed some performance issues in 1.7.10 and 1.8. 2. The data presented of the 24-hour mob-spawn test occurred during the largest holiday in the world. What those graphs fail to show is that the number of players on the server was 1/3 to 1/2 of our normal player count. I cannot argue that the mobs appear to be a major source of the CPU lag that we are experiencing, but the data presented is somewhat misleading. Maybe we can conduct another short test during peak usage to see if the results are conclusive? 3. I believe that there may be a mixture if issues happening that are leading to poor performance. While mob spawns were disabled, I experienced some very bad crashes and lag on the server that were mostly isolated to the morning. It did not appear to be a TPS issue. /tps showed consistently high readings. I think that to some degree the crashes and instability present in the modpack are being lumped in with the CPU usage issue and as a result we may make drastic changes on the server only to be faced with the same problems.
Squids and Bats are now disabled. Only those two just peaked 50ms. (remember 50ms = all we have for 20tps not even considering the normal strain from players, worlds..)
Someone mentioned that the lag caused by the mobs are due to their pathfinding code, can't the mob's ability to move be limited or totally disabled all together?
the data has been compared with the player count: While this should be taken into consideration, such a drastically difference is highly unlikely. Furthermore are mods mainly developed and tested with java 6 or 7, while oracle states that all java 6,7 apps run with java 8 it isn't different than with windows xp -> vista, vista -> 7, 7 ->8 => especially not with the experimental way some of the mods are written with. Also take into account that the mean time is about at only about 1/14 without mods. It is true that the server was down for quite some time and later on an exploding me system did put it into its knees. We could run another test in order to get more intel around this. But in my opinion I got all the information I needed, with the comprehension of ms update time. Do consider that normal lag spikes from all sort of sources (e.g. ME autocrafting, world load/unload, autosave, chunkload/unload==player login/logout/teleport/rt) appear more drastically than under heavy load. Taken a moment ago with 35 players online after disabling squids and bats which improved the entity update time by 50ms: Be aware we have an ent count per player of around 44, which is low, very very low. The average of vanilla is at around 80-120.
Im a bit confused. Or im not understanding. When you say 50ms = all we have for 20tps, What exactly does that mean? ms= ping correct?
i thought 50ms was 1 tick its from what i gather from this Can't Keep Up - Improving Minecraft Server Performance Our Can't keep up! message is caused by the dreaded LAG. Basically, minecraft does everything in TICKS. Each tick is intended to last 50MS and must handle everything in game during that 50ms window. During each tick, your minecraft server must make mobs move, handle all player actions that occurred within the last 50ms, notify all players of all changes, and many more things. Adding plugins to the mix means that even more things must be done each tick. When each tick starts to take more than 50ms, you start experienceing Lag. Lag can have wide-reaching consequences. Severe Lag can cause the sun and moon to slow down, and make days and nights take longer. Mobs move and react more slowly. Instead of moving smoothly, you'll notice players jump from place to place. Cooking items in a furnace will take longer. Lag significantly deteriorates your players' gaming experience.
nothing personal, but I feel that I need to clear my head: Willi and me have spent around 30 hours on this issue. We have enough for it. We have tested so much have written custom solutions to limit the mob spawn. Started the modpack/server around 150 times (15-20h) to test out different solutions, try the different versions of forge, cauldron and test mods by them self's. We need a solution now! and I think you all agree. I do fully understand the concerns around splitting it and the gameplay part that will be missing, but yet I haven't heard/found any solution that would come only close to the proposed one. Unless you can come up with something we haven't tried yet and sounds promising I would like to focus on the idea. I'm not interested in spending another 20 hours on this issue and ending up without a solution, especially as we need one NOW.
slind if you disabled squids for now, could you put ink sacs on the market? cuz we need ink sacks for alot of things
Slind's trying to get a solution running and you talk about ink sacks xD[DOUBLEPOST=1419611062,1419610873][/DOUBLEPOST]What java is the server running on? If its java 8, how about trying it on Java 7
pretty sure they are using 7[DOUBLEPOST=1419612738][/DOUBLEPOST]i still see a lot of bats on my mini map. Anyone see squids?
