CraftBukkit 1.2.5-R1.0 is now available!

Discussion in 'Bukkit News' started by EvilSeph, Apr 4, 2012.

Thread Status:
Not open for further replies.
  1. Offline

    EvilSeph

    A new CraftBukkit Recommended Build (1.2.5-R1.0) that provides Minecraft 1.2.5 compatibility is now available.

    Will plugins break with this build?
    Unless the plugin developers of your plugins have been keeping up with Bukkit development, plugins will most likely be broken due to the recent optimisation and cleaning of our code. However, if the plugins you use have already been updated to work with at least 1.1-R7 or at least 1.2.3-R0.1, then you should be fine.

    For more detailed information on what is contained in this update, please see the changelog here.

    Download CraftBukkit 1.2.5-R1.0 here
     
  2. Offline

    tremor

    A power cycle. You might need to do this per world if you have multiple worlds. What's probably happening is there may be some world generating changes that aren't being handled properly. This is just a wild guess as of right now, but this solution has worked for my server in the past in regards to chunk error like issues after a version update.
     
  3. Offline

    palisade

    I've moved this to this page in the hopes evilseph gets to see it.... I went ahead and performed Shift-H in top to see the pid of the highest java thread. And, then performed a kill -3 on it to dump the thread information. Then converted the PID to hex and searched the thread dump for the culprit. This might help in analyzing what might be causing the problem.

    "Server thread" prio=10 tid=0x6e255400 nid=0x4fe2 runnable [0x6e1fe000]
    java.lang.Thread.State: RUNNABLE
    at java.lang.Object.getClass(Native Method)
    at org.bukkit.Location.equals(Location.java:420)
    at net.minecraft.server.EntityMinecart.F_(EntityMinecart.java:603)
    at net.minecraft.server.World.entityJoinedWorld(World.java:1264)
    at net.minecraft.server.WorldServer.entityJoinedWorld(WorldServer.java:106)
    at net.minecraft.server.World.playerJoinedWorld(World.java:1246)
    at net.minecraft.server.World.tickEntities(World.java:1153)
    at net.minecraft.server.MinecraftServer.w(MinecraftServer.java:542)
    at net.minecraft.server.MinecraftServer.run(MinecraftServer.java:449)
    at net.minecraft.server.ThreadServerApplication.run(SourceFile:492)

    Empty server, no joins.
    Hope this helps.


    Ya, I re-ran it with the pre-release 1.2.5 and it didn't fix anything. Worth noting that the pre-release jar runs 12%. To test your hypothesis, I decided to completely delete the worlds (I have them backed up) and all my plugins as well, and regenerate them from scratch with bukkit 1.2.5. I'm getting 28% to 45% right now. I'll let it run for awhile and see if that % drops off.

    NOTE: Just in the few minutes I've been running it I've seen it hit a low of 18%, and sitting around 21% most of the time. Shutting it down and re-running it has it sitting between 17% and 28%.

    As a second experiment, I deleted the worlds again and ran with minecraft_server 1.2.5 prerelease to regenerate the world instead. Initially it sits at 20% - 25% cpu, and then quickly drops to 16% - 23%. After a 30 seconds it hit 12% to 15%.

    Third experiment, I ran bukkit 1.2.5 again using the world generate by prerelease 1.2.5. It notified me that it needed to migrate the world, and started doing so. After it said it was done doing this, the cpu sat around 37% to 46% initially. After about 30 seconds it dropped to 23% to 29%. I'll let it run a little longer and see what happens.

    Three minutes later bukkit is hovering at 15% to 20%.

    At six minutes in it's holding between 16% and 20%.

    Worth noting, I went back and ran my 1.2.4 server and it runs 29% to 50%. However, after 10 minutes it has jumped up to 48% to 55%. So, at least for me, it's not a problem between 1.2.4 and 1.2.5 bukkit. But, really a difference between minecraft_server prerelease 1.2.5 and bukkit 1.2.5, with bukkit while using my original worlds sitting at a high % and the prerelease from Mojang sitting at 12% with those same worlds.

    Without plugins bukkit 1.2.5 runs at 15% to 29% with my original worlds, vs prerelease at 12%.

    Nice little graph of what's going on:

    [​IMG]

    Original on the chart means with my original worlds that players have built up statues, homes, castles, etc on. Fresh indicates I deleted the worlds and regenerated them. The mc 1.2.5's overlap, so the cyan line vanishes into the dark magenta one.
     
    Don Redhorse and Miamore like this.
  4. Offline

    thadeadmonkey

  5. Offline

    Bibber

    One question: When will the mob wall glitch be fixed? This occured first in MC Beta 1.8 or so (don't remember correctly) and it is still not fixed. It's really annoying to hear bone breaking sounds all the time when some sheeps are in my cage glitching on the walls or when I'm digging through the underground (especially slimes). Also mobs seem to be able to escape from an area surrounded by fences. I guess it's the same bug as they sometimes glitch and suddenly stand on top of the fence from where they can escape. Please fix this annoying bug.​
     
  6. Offline

    Amaranth

    It may not be 100% fixed with fences and iron bars and such but mobs should no longer glitch in to or on top of regular block walls.
     
  7. Offline

    robxu9

    Wrong Jarfile. You should be using CraftBukkit :|
     
  8. Offline

    Hazzabazza33

    Great, thanks for the quick update. :)
     
  9. Offline

    Juze

    That was quick, thank you!
     
  10. Offline

    dEEPZoNE

  11. Offline

    Hoang

    It was out so fast because they were working on it dev. and beta build throughout the week due to Mojang released their snapshot early for dev. like them.
    Again, thank you very much!
     
  12. Offline

    goldhaxx

    I feel so derpy right now.
    [pig]
     
  13. Offline

    gawelium

    Hi !! Great news !!

    But just one thing. I remember in lastest version (in minecraft 1.7), admin can see in the console when a player join to a server what mods this player have in his minecraft. P.e : gawelium joined with xray_mod
    Why this thing has been remove ? and maybe you can re-add it no ?!!

    Thank's for the answer.
     
  14. Offline

    Hajimoto

    EvilSeph - Any ideas or comment on the additional CPU load in 1.2.5 as opposed to 1.2.4? This doubling of the CPU as crippled my server and forced me to continue with 1.2.4 until a resolve is noted.

    In previous posts I have listed that I have disabled all plugins during this evaluation and also ran the server using Mojang's Minecraft_server.jar, to confirm that it as not a configuration issue.

    Please advise
     
  15. Offline

    feildmaster

    They have too much travel room. If they leave the vicinity of the village more can spawn.
     
  16. Offline

    Phoenix_Skull

    Hey um..... I created my new server, i did everything right, I have had previous experience with creating one (1.8.1 - 1.0) and I recently found out about 1.2.5 update for bukkit. I downloaded, installed it right, and got my server running. Then after about 30mins of building, a sudden glitch appears and i cannot place or remove blocks around where i have just built. If I move 40 blocks away, i can build but, then another 30mins gone past i cant build there and no one else can. I have no plugins, no nothing, all the documents. Please Help!!
     
  17. Offline

    Bibber

    It hasn't been fixed. Mobs still glitch into and on top of regular blocks in 1.2.5. It's really annoying that no dev cares about this bug for months. Sad.
     
  18. Offline

    palisade

    At least the 1.2.5 client is going to be able to connect to a 1.2.4 server this time, so that's a relief.
     
  19. Offline

    Amaranth

    The fix in place is pretty much the best we can do without a protocol (and thus client and server) change.
     
  20. Offline

    Bibber

    Hopefully there will be a real fix with the next protocol update... but I don't really believe because the problem is present for months and nothing changed until now. I can just hope.
     
  21. Offline

    Denni300

    Awesome, but can i use plugins for 1.2.4 for this RB ?
     
  22. Offline

    tremor

    Many are still compatible but I have heard about a few breaking changes that may affect some plugins... Multiverse is potentially one of those plugins that may be affected. Can't be sure I'm sticking with 1.2.4 until I get more information. Since the 1.2.5 clients can still connect to my server, there is no need to rush it.
     
  23. Offline

    EvilSeph

    palisade not really sure what the issue is, but we are investigating possible causes. Please try this build and let me know if there is any change in CPU usage: http://dl.bukkit.org/downloads/craftbukkit/view/01009_1.2.5-R0.1/

    Please be certain to not use any plugins what-so-ever and try and keep the testing environment the same throughout every test we attempt. It would be better if you can get on IRC and we could investigate this in real time. At the same time, how big is the world you are testing on? If it isn't too big, could you upload it for us for testing?
     
  24. Offline

    TnT

    MultiVerse works for me.

    If you are at all curious if a plugin works between updates - test it. You'll find out right away.
     
  25. Offline

    travja

    Users keep getting kicked with a .overflow error... What is causing this?
     
  26. Offline

    Grrrr159

    yes! thanks for updating :D
     
  27. Offline

    grid21

    I have been using Craftbukkit 1.2.5 on my server and it is really stable! I haven't had any problems with it and and all my old plugins from 1.2.3 and 1.2 are working perfectly! I can't say enough how greatful I am to the team at bukkit.org! You guys are ROCK AWESOME!! Keep up the amazing coding you guys do, and I'll keep coming back again and again!
     
  28. Offline

    palisade

    Seems like the problem is still there with the dev build you provided. Here are some readings over the course of a minute:

    32%, 43%, 58%, 60%, 52%, 56%, 45%, 42%, 31%, 39%, 37%, 53%, 75%, 39%, 65%, 38%, 26%, 57%, 27%, 25%, 26%, 38%, 49%, 41%, 49%, 30%, 24%, 32%, 39%, 28%, 35%, 54%, 53%, 35%

    By the way, I think the problem has probably existed in bukkit for some time, we're just now paying attention to it. And, for at least one other user, he noticed a signifcant difference between 1.2.4 and 1.2.5 bukkit. For me, there was only a slight difference between the two. The real problem is there is a significant difference between bukkit and the regular minecraft jar in terms of cpu usage, with or without plugins. Even 1.2.4 had this problem. It'd be interesting to go back and check the bukkit versions historically and see when it might have picked up the high cpu usage. Maybe it's always been there. If not, then a diff between a couple of releases might show the culprit.

    At work at the moment, so I'd have to do the real-time IRC session later on after work when I have some free time.

    The world isn't too large, I made a point of using BorderGuard plugin for performance reasons to keep users from making it too large. I limit the world to a 4000 block radius, which as worlds go is rather small.

    After several minutes, cpu is 39% to 61% btw. I do have plugins loaded in this test. Unfortunately, a few players just joined as I was testing this, so I don't want to shut the server down on them and try without plugins yet. And, running a second server on a different port would be rough on the resources.

    For size reference:

    palisade@recoil:~/mc/world$ du -hs
    372M .
    palisade@recoil:~/mc/world$ cd ..
    palisade@recoil:~/mc$ cd world_nether
    palisade@recoil:~/mc/world_nether$ du -hs
    28M .
    palisade@recoil:~/mc/world_nether$ cd ..
    palisade@recoil:~/mc$ cd world_the_end
    palisade@recoil:~/mc/world_the_end$ du -hs
    12M .
    palisade@recoil:~/mc/world_the_end$ cd ..

    Note: In my earlier post I made a point of starting with a fresh world, no plugins, and bukkit still used more cpu than the regular mc jar. However, it did use less cpu than with my existing world. It's the yellow line on that chart in my post above.

    Here's my world:
    Download.
    md5sum: 19be20d27dddca6d53a54b29786bc44e worlds.tgz

    Okay, they logged off again.... I took out the plugins, here's the results:

    19%, 41%, 27%, 19%, 41%, 70%, 37%, 34%, 24%, 18%, 19%, 24%, 21%, 19%, 48%, 48%, 32%, 27%, 22%, 19%, 18%, 19%, 29%, 25%, 46%, 42%, 37%, 24%, 17%, 18%, 18%, 24%, 23%, 19%, 44%, 40%, 29%

    The min, max and avg here are lower than with plugins, but overall the cpu percentages are still significantly higher than the avg 12% from the mc regular jar.

    You know what would be a cool command line option or server.properties flag to add to future bukkits? Some sort of benchmarking mode so we can tell it to track memory / cpu / players / plugins / etc. while the server is running and then dump statistics into a special log file at any time. And, maybe dump it via a command like /dumpstats [filename] or something. That'd be really handy. I suppose it'd be similar to the LagMeter plugin. But, it could even tell you what each plugin is consuming. That'd also help in finding which plugins are the heavy ones on memory and cpu, and could be a useful toolkit for plugin developers to optimize their plugins with.

    I guess it could be a plugin, but then it wouldn't be able to check statistics without plugins loaded. Maybe for older bukkit versions we could make a plugin that does it, but in future bukkits the support would be built right in. This would also be handy in the mainline minecraft_server as well. Though, I suppose there are probably external tools we could try using as well made for profiling. I know java has some, but its hard reading their statistics.

    It's good we're looking into this, because people tend to use their servers for other things like databases, web servers, email, counterstrike, mumble servers, etc. And, sometimes they're running the server on the same machine they're playing from, so then the server could be stealing cpu time that could be used by the client.

    After a day of running, here are the stats:

    [NO USERS] 13%, 15%, 42%, [USER JOINED] 25%, 23%, 45%, 23%, 29%, 49%, 20%, 27%, 33%, 29%, 27%, 21%, 19%, 25%, 21%, 21%, 43%, 21%, 25%, 23%, 25%, 52%, 27%, 17%, 49%, 19%, 19%, 23%
     
  29. Offline

    superpeanut911

    Great job on the quick update :D !
     
  30. Offline

    ankmagicmn1

    Every time I go to update my server to the new build it like rolls back time on the map. So things that have been built disappear like I have not done anything. Any way to fix this or am I faced with the decision to stick with my version or start a new map?
     
  31. Offline

    4am

    That should not be happening at all, nothing should ever "roll back". Are you using a Ramdisk? How are you stopping your server when you upgrade?
     
Thread Status:
Not open for further replies.

Share This Page