I have a vanilla server, recently (excitedly) upgraded to 1.7.2 (did not run any snapshots since 1.6.4). Since then I have had problems connecting to my own server. So background and extra info:
I recently moved into a crappy area of Colorado (bandwidth speaking) and went from a decent 16Mbps to about a sad 1Mbps down and about 3Mbps to 1.5Mbps up (don't ask how that works) sort of situation.
This happened around Sept. 22nd, so from then on, no problems till the 1.7.2 update on Oct. 25th. On the 26th I MCEdited the world to delete chunks where no one had built. Uploaded the world back to the server same day as updated to 1.7.2, and was able to connect no problem. World seemed fine, everyone said the update went well their things were still there and was excited for the update.
I played a bit but nothing crazy and things seemed to be all right. Then I started getting disconnected. But not only would I get disconnected, the server would crash completely. Nothing in the error log, and I finally decided to run outside a screen and the only thing in the console was 'killed'
I'm on a digitalocean server with 1GB ram, 30GB SSD, and Ubuntu 12.04.
Since this has started happening, none of the other players have had an issue. So it seems to point to something about my network. I reduced my start script to the bare minimum:
screen -S minecraft java -Xmx610M -Xms512M -jar minecraft_server.jar
And still have the issue. I have used three computers on the same network all with the same problem. I recently took one machine and put it on a better network and had no problems.
At this point, most the time time I join for a few seconds and then disconnect and kill the server, albeit sometimes I can join and as long as I don't leave, I can stay connected for a couple days.
I switched back to 1.6.4 and don't seem to have any issues at all.
I tried upgrading my server to a 2GB ram and 64GB SSD (same start script) and still had the same problem when using 1.7.2.
When I'm on windows (7) the client disconnects with error message:
Internal Exception: java.io.IOException: An existing connection was forcibly closed by the remote host
On Mac OS 10.8.4 my error message is:
Internal Exception: java.io.IOException: Connection reset by peer
At this point I can honestly say I have tried absolutely everything I can think of, it's not a big deal for me as I will only have this problem for a couple months till I move to NY but I fear others (probably with slow internet/ping) may not be able to join in any multiplayer games. And it's scary that someone like me could attempt to join any server and completely crash it. Let me know if you want me to try, because so far I haven't, GGG and all.
Comments 5
I spent the last few days, completely reformating my server and I loaded on Ubuntu 13.10, also did a fresh install of the latest java on my machine, and I seem to be fine now. Since the upgrade it also resolved the info message I was getting on starting the server:
io.netty.util.internal.PlatformDependent <clinit>
INFO: Your platform does not provide complete low-level API for accessing direct buffers reliably. Unless explicitly requested, heap buffer will always be preferred to avoid potential system unstability.
This is reported in: https://mojang.atlassian.net/browse/MC-36810
However I can't confirm it that was from the ubuntu or the java update. I tried upgrading my java before the server reformat and I couldn't get it to work properly.
@jon, every time I had the problem I was no where even to close to 50% ram usage. I was normally in the 20-30% range, so not sure if we were dealing with the same issue.
As far as my issue is concerned I would say we could consider it resolved, even though not knowing the why always sucks.
More details. Just checked syslog, and found the following relating to excessive RAM usage for the two crashes today:
Nov 9 17:38:54 xev2 kernel: [992050.234014] Out of memory: Kill process 5646 (java) score 849 or sacrifice child
Nov 9 17:38:54 xev2 kernel: [992050.234063] Killed process 5646 (java) total-vm:4674276kB, anon-rss:1798104kB, file-rss:0kB
Nov 9 23:12:00 xev2 kernel: [1011997.750309] Out of memory: Kill process 4277 (java) score 823 or sacrifice child
Nov 9 23:12:00 xev2 kernel: [1011997.750350] Killed process 4277 (java) total-vm:4941608kB, anon-rss:1697984kB, file-rss:0kB
Minecraft server is the only java process running on the machine.
Before the crash occurs, there's no warning signs RAM is running tight, suggesting that when it goes wrong it happens quickly.
I've not yet found a way to reproduce it reliably. However, the two crashes today were after my brother logged in, found it was night, and immediately slept in a bed.
Also, $ java -version
java version "1.6.0_27"
OpenJDK Runtime Environment (IcedTea6 1.12.6) (6b27-1.12.6-1ubuntu0.12.04.2)
OpenJDK 64-Bit Server VM (build 20.0-b12, mixed mode)
I hadn't noticed earlier, but I also had that 'netty' warning message.
Also, since yet another crash, I decided to update Java - from the OpenJDK 1.6 to Oracle's 1.7
$ java -version
java version "1.7.0_45"
Java(TM) SE Runtime Environment (build 1.7.0_45-b18)
Java HotSpot(TM) 64-Bit Server VM (build 24.45-b08, mixed mode)
I'll keep a close eye on things and see if the issue still occurs.
Not quite sure this is a duplicate of MC-38565, but since the Java upgrade (details above) it has been stable.
Also, the java upgrade fixed a separate issue I occasionally saw - 100% CPU usage on the server, when idle and no users connected.
I've seen something similar on my server here.
I have a private server, used by myself and my brother, running on Ubuntu 12.04 LTS. The server machine is a VM with 2GB RAM allocated to it. A few other things are running too (e.g. webserver), but nothing high load.
The minecraft server is started with the command line "java -Xincgc -Xmx1024M -jar minecraft_server.jar nogui"
The first problems I noticed were severe issues on the server - so bad I couldn't log in. It appeared as if something was swapping like mad, causing the problem. The swap size was vastly reduced (from 4GB to 512MB) to limit swapping issues.
Now (and twice so far today), the minecraft process has crashed with the 'Killed' message. There are signs that RAM usage was high just before the crash, with some swapping - but I haven't yet managed to catch it in the act.