EntCount - What changed on Dec 08, 2012

Discussion in 'Community Discussion' started by Aikar, Dec 3, 2012.

  1. When I logged out of smp9 last night, both the debug screen's All: and Entcount had been stuck on zero for about fifteen minutes. I went back this morning and entcount's still stuck at 0/250.

    Smp8 Wild is like this too. 0/250, with one or two animals popping up preiodically. Since I'm under hte ocean, this must be squids.

    On smp3 Nether there always seems to be about 30 phantom mobs that are there immediately when I log in and I can only get up to about 60 mobs total.
  2. I found out thatdue to the decreased spawn radius, the "World mob cap" (the one the entlimiter and such was aimed to curb due to spawners) had been hit...

    since the distance is half the vanilla, the mob cap formula was essentially halfed.

    so spawners could hit the world mob cap again...

    this is what everyone was experiencing, and so ive upped the modifier (which is simply used for the world cap...) and holy crap... I'm surprised noone's complaining there is too many mobs now.

    this makes perfect sense for the less than normal mob counts now.

    problem is fixed...
    Pab10S likes this.
  3. I'm kinda confused and Im too lazy to read through the comments to find the answer :p
    So if I want a huge sheep farm in the wild then how far away does it have to be from my mob grinder so both could have the 250 mob cap? (Is it 4 chunks away as in 64 blocks away?)
  4. What kind of grinder? Is it a natural one? If it is a natural dark room grinder, you can never have 250, it can only reach 90. If it is a spawner grinder then you need to keep the sheep out of a 5 block radius of the grinder. (s-spawner, #-sheep, - Chunks)

    S - - - - - #

    If you are going in a straight line its easy. 16 Blocks per chunk, 5 chunks, 80 total blocks.
    Pab10S likes this.
  5. Ohhh ok thanks
  6. I will be making some tweaks to thin back out the density... I'll try to keep the spawn radius at 4 but i likely will test it with 5 and see how things are.

    I consider 5 chunks a 'solvable problem' with slabs/torches... so I'm not going to base that setting around farms.

    Right now mobs are really way too thick, so i'll tweak it.

    But now I at least understand why densities were so much lower in prod than my testing -- I never had spawners full to 250/250 cap in other parts of the world affecting mob cap...

    With the modifier raised thatll def not be a problem and gets me better testing in my test server (which I realize I was hitting mob cap there too w/o the spawners! Explains confusion I had)
    ZBSDKryten likes this.
  7. 1 Question: Will it destroy the economy any more?
    PandasEatRamen likes this.
  8. I was playing yesterday when there were about 35-40 players logged in on smp6 and I think I've experienced the lag you're talking about - for the first time. That kind of lag makes the game hardly playable, so it is clear that this is kind of top priority.
    (BTW I did not cause the lag, I was collecting sand and none of my grinders were in use :))

    Instead of giving Aikar a break, :)
    let's try to help.

    For now, I won't use grinders if there are more than 20 people on the server.
    For Iron farms, it is possible to spread out the villagers a bit - would that help? Or put them into 1x1 holes?
    For gold farms, none of the designs I know should cause lag. (?)
    For grinders with spawners, I guess that having all the mobs in 1x1 hole is better than collecting them in a larger area - is that so?

    Aikar, I guess you've already gone trough diagnosis of the problems several times - please don't mind few questions and hints ...

    Could we somehow - just to test - signal the server where the mob collection point is, perhaps with a sign?
    Could you optimize - again, just to test - the mob movement then?
    Could you build some statistics about mobs per area density and correlate it with the lag?
    Could you optimize mob movement for areas with high density? (If it turns out that this is really the problem.)
    Could you limit the time needed to compute mob movement, so its CPU usage is limited, i.e. if there is not enough time (or CPU power), mobs would not move that much. Perhaps just to test?

    Thank you for you work! Grazie mille!

    For me, I don't really wont to turn MC playing into work, but I'd invest few hours to help testing.

    It is easy to guess that most of the community does not have an insight into inner life of the MC server and CB, and also no practice in profiling and debugging code, but still, it is worth describing the problems to the community. By describing a problem you're (re)ordering associations in your mind. I expect that there are people who understand something about software development among us :), and often, a layperson will give you a valuable hint.

    - can you / did you profile the relevant code / functions?
    - is the server multi-threaded in the mob movement part?
    - how many threads are there normally? How many CPU cores in use, out of how many? When lag is there?
    - can you influence mob movement code? Profile it? Check (again) if / where it is creating the bottleneck?
    - what JRE are you using on the server? Did you tune the JRE? Garbage collector?
    - I assume there is plenty of RAM and no swapping at all. (Tune kernel swappiness etc?)
    - Can you copy a server to a dedicated physical machine to test?
    - Simulate the load by identifying problematic areas and arranging that those chunks are loaded?

    Last but not least I want to encourage you to invest most of the time into diagnosing and the least part of time into tweaking :)
    Good luck!
  9. Can someone give me that in short form?
  10. PS: I really enjoyed more mobs out there.
    That was fun! :)
  11. "Did you turn it off and back on again?" "How about turning it this way?" "Did you stand on your head while doing it?"

    I just can't believe he's actually posted and not used his favorite E word.
    jkjkjk182 and mba2012 like this.
  12. haha sounds like two people trying to flip beakers. one outside flipping them and other one in house yelling at him/her lol
    PandasEatRamen likes this.
  13. Hmm, it seems since the update I am also experiencing game breaking lag, (I'm talking 1-2 fps)
    For the reference of Aikar and Smooch, I'm getting it at the iron farm in pazzo (which now means I am stuck there)
    Perhaps it is something that can be checked out to see if it really is the cause of my problem?
  14. I took a look. There was a jam of golems that could have been causing the lag. I've unjammed them which might help. I never had any lag in the area except when I first tp'd in and found the chunks a little slow loading (the reason I saw the jams).
    RainbowChin likes this.
  15. Yeah, that seems to be an issue we have at Pazzo. Had to let iSMOOCH know once and he found 313 stuck in there and I've had to unclog it two or three times since then because too many were jammed. It's like 3 try to go through at once and get stuck and back them all up.
  16. Seems like that would be a bonus for the person doing the unclogging. At 3-5 ingots per golem, that's pretty good pay.
    Dwight5273 likes this.
  17. You bet :D
  18. I think you guys are experiencing the effects of faster spawning...

    The system use to spawn once every 5 ticks, now its once every 2 ticks, so overall spawning is quicker.

    I bet the golems are spawning too fast now. maybe reduce the villagers?