CraftBukkit 1.2.5-R1.0 is now available!

504 Internal Errors mean the server took to long to respond.
It could be to being overloaded but it works for me, the servers aren't down as they wouldn't just do that with out notification. Most of the time(I said most) its due to the client not reviving a valid/no response while the server has sent the data back.If you still get this error and you just want to run your server, run it in offline mode with an authentication plugin like xauth.

so now the problem is that i couldnt connect to my 1.2.4 server with a 1.2.5 client.(all original/clean version)
no i am not a noob i've been running servers from 1.1-R1 for 20+ppl. so dont question me about my knowledge of this. my 1.2.3 server is functioning as we speak.
all i wanna know is if the quote above is absolutly right and i need to check again and try using another computer, or if is proved wrong.


==========and==========
why is my old post about this problem deleted, with no messages/warnings? a too dumb question?
if you're gonna delete this one please simply tell me that this post is stupid and i will not ask anymore.

I ran my 1.2.4 server for a full week (upgraded to 1.2.5 this morning) and all my client (20 people) were on 1.2.5 without having a single issue, while I can't say that the protocol did not change (I'm not that hardcore yet) I can confirmed that it worked for us.

thanks a lot. it turned out that i used a wrong minecraft.jar(modded version which i forgot to put a tag on it).lol it's stupid.
and to moderaters please take the trouble to tell people why posts are deleted, which will be nicer.

Hmm...
I'd expect them to tell you something about it, but just saying that won't fix your problem go talk to a moderator.

sorry about my not being aware of the rules, and i will use help section next time if needed.
but i see many help requests on page 2 and 3 why delete mine only...?

You're probably talking about my posts. I was informing everyone of bugs I was noticing in 1.2.5 to help the developers be aware of the problems. The fact that people tried to give me advice doesn't mean I was asking for help.

buckit is cool

99 comments, but complaint ain't one.

Good job Craftbukkit team

Dev team - Can you update the status of where you are as it relates to the additional CPU load incurred with the release of 1.2.5-R1.0 which I reported on April 4th (post #16) which was extensively expanded on by Palisade. I am running 1.2.4-R1.0 until more information is released.

Is it possible to comment on the below listed action items?

• Does the the team acknowledge the existence of the additional CPU load.
• Is the additional CPU load the new acceptable norm.
• If not the intended norm, is there a timeline for resolve.

Thanks

I'll try to help answer some of the questions:

1. EvilSeph stated they're looking into it, so they acknowledge it.
2. It has existed in past versions, but maybe it hasn't been as noticeable until now. They've been very focused on bug fixes and features that I think maybe they didn't notice the CPU usage creeping up. I don't think they want it to be the norm, which is why they're investigating it.
3. As a developer, this question annoys me a lot. When you ask for a timeline, you're assuming that software development is as simple as doing your laundry. It is a lot more like trying to track down a serial killer. Go and ask your local homicide department the exact date they're going to catch the killer.

Unfortunately, I got sidetracked with my plugin and server maintenance that I didn't get a chance to jump on IRC and help investigate myself. It is worth noting that it only seems to occur on a subset of CPUs, so the bukkit team might not even be seeing the problem on their side. I never found out if that's the case or not, they were interested in me helping analyze it I think because I'm at least seeing the problem. Others here have reported not seeing the CPU consumption problem at all.

Hajimoto described that for him it sits above 60% most of the time and it's dragging his server to its knees, so he's sticking with 1.2.4 for now. Though, personally I'd like to see a graph of bukkit 1.2.5 cpu usage on his system over a long period of time. That might not actually be his average usage.

For me, I'm seeing a different picture of cpu usage. You can flip back to page 2 to see a graph I provided of cpu usage of bukkit 1.2.5 vs bukkit 1.2.4 and fresh vs complex world loaded.

For me, 1.2.5 performs poorly compared to 1.2.4 in the first few minutes and then over time performs slightly better than 1.2.4. Comparing both of them to the official minecraft server is where the problem is brought into sharp relief, though. The official is averaging 12% early on and maintains this throughout, regardless if the world is fresh or complex. I've also tried bukkit without plugins, and there is a very small improvement in cpu usage, but it never reaches the official server's 12%.

That's the best description of the problem from my side. I can't speak for Hajimoto or anyone else, until we see their own statistics.

I never paid attention to the cpu usage on past builds. Also I started playing minecraft/bukkit (i skipped playing the official minecraft altogether and didn't bother with single player) last month when bukkit 1.1 R6 was out so that was my first exposure to the game. I never tracked the resource consumption at the time, though I wish now I had.

I do know that my copy of the 1.2.4 had high cpu, I haven't tested all builds of it. It might have been introduced before 1.2.4 or during I don't know. The 1.2.4 build I used was the most recent dev build. I do think it's a bit much to ask me to go through and test every single build myself. I suppose I could go grab a really ancient version of bukkit and run it and see what happens. Like.... *pulls one out of the hat* 1.0.1-R1.

Unfortunately, my friends are playing on my server all the time so finding a good moment when I'm not busy and the server is empty to test this in. I want a clean environment with no other processes running, which is hard to pull off. And, I'm loathe to just shut the server off on them to test things like this because then I won't have any users at all. I'm still recovering from the quick update pace that reduced me to 2 - 3 users from 8 - 10 (ya I know, it wasn't very popular to start with, but it's brand new). The graph I made on page 2 was me jumping at the opportunity when I noticed my server was suddenly empty.

I could try it on another computer instead of my main server, but I'm not sure I'll see the same results.

I've just run bukkit 1.0.1-R1 here on my work computer (WinXP, jre7) and after the initial world builds I'm seeing between 1% and 20% CPU usage., with it mostly hanging around 9%. Again, I'm not sure how relevant these numbers are to my server's.

Now I'll grab 1.2.5 and run it here too. I'm seeing 6% to 20% with it mostly sitting at 13%. Note: After awhile it does improve and goes down to 1% to 9% with it mostly sitting around 5%.

This is just me eyeballing bukkit's CPU usage, we should probably run a professional benchmark or something and track it over time.

This box is a Dell Latitude D620, WinXP SP3, Intel Core 2 T7200 @ 2GHz with 2.49GB RAM, it's a laptop.

Ya, I know, my job sucks.

Again, vastly different from my server which is running Ubuntu on a Xeon Quad-core @ 1.66GHz and 8GB ram in a server environment. This server isn't very fast but it's just some old hardware, and it's only running bukkit and mumble. My gaming machine is significantly faster at home but I'm not going to bother running the server on there [core i7].

I have went through all the builds and 1.2.5-R0.1 (Build #02136) is when the massive CPU load started.

@TnT: Since my post which isolated the build in which the additional CPU load was introduced, has there been any headway in identifying or narrowing down the culprit?

Is there a procedure or sequence that I can perform on my server that will produce additional information to assist you in this? If so, please let me know and I will be glad to help.

Looks Solid.... Using 2155 and CPU loads are back to normal, WOO HOO! Great news.

Thanks!

good job !!!

Still getting 46% to 63% here with that build.

It's definitely after 1.0.1. I'll look at it more when I get home.

What??? still the this version... I am still using 1.2.4 because of the overload issue with 1.2.5. When are they goin to fix the problem?

You should try the latest dev version and see if their fix has solved it for you. It has fixed it on Hajimoto's side, but everyone's computer will be different. For example, I've never experienced the 100% CPU usage all the time that he has. For me, after a long time of the server running the CPU usage drops off. Also, for me it happens on 1.2.4 exactly the same, albeit with slightly more (5% - 10% more) usage on 1.2.5.

You might try 1.2.5 and find it isn't causing you any problems. Don't use me as a baseline.

As others have noted, I have CPU overload issues with the 1.2.5 release. As a result, I'm still using 1.2.4 with no problems for players who have updated to 1.2.5.

I have tested the CPU usage on my server, and on a second one for testing purposes at home with no plugins.

I have not tried a dev release as of yet, but will report back with the results as it appears this has helped some others with this issue.

Please bring that to the devs. It's really time to fix it now.

Your comment wasn't very descriptive but I can tell from the video you're upset about the animal collision causing them to choke to death in walls. This is a known problem in Minecraft itself and they're still working on fixing it.

For now you can put fence rows inside the walls to keep them from suffocating.