[MC-63802] Server hangs after /stop when players are still online

When I do /stop, my server console will not close. It saves and all that, then I have to cmd-opt-esc (open force quit menu) and force quit the console. When I restart it, everything is saved, but the only problem is that force-quitting just does not feel right. I'll edit this if it comes out not saving. I am running a mac OSX Mountain Loin. Thanks.

P.s. It closes normally if there is no one on the server.

CONSOLE OUTPUT:
[11:22:39 INFO]: Stopping server
[11:22:39 INFO]: Saving players
[11:22:39 INFO]: Saving worlds
[11:22:39 INFO]: Saving chunks for level 'New Creative Server Photos'/Overworld
[11:22:40 INFO]: Saving chunks for level 'New Creative Server Photos'/Nether
[11:22:40 INFO]: Saving chunks for level 'New Creative Server Photos'/The End
[11:22:40 INFO]: Stopping server
at the 'Stopping server' part it does nothing else and won't close unless I force quit the console.

Linked issues

duplicates 1

MC-72806 When using /stop server executes the shutdown sequence twice (and hangs if players are online) Resolved

Comments 27

kumasasa 2014-07-26T07:04:28Z

Cannot confirm, getting that in the console:

D:\MineCraft\server-test>java.exe -Xmx2G -Xms2G -jar minecraft_server.14w30c.jar --nogui
[09:01:52] [Server thread/INFO]: Starting minecraft server version 14w30c
[09:01:52] [Server thread/INFO]: Loading properties
[09:01:52] [Server thread/INFO]: Default game type: SURVIVAL
[09:01:52] [Server thread/INFO]: Generating keypair
[09:01:52] [Server thread/INFO]: Starting Minecraft server on *:25566
[09:01:53] [Server thread/INFO]: Preparing level "Molokai"
[09:01:53] [Server thread/INFO]: Preparing start region for level 0
[09:01:54] [Server thread/INFO]: Preparing spawn area: 2%
[09:01:55] [Server thread/INFO]: Preparing spawn area: 27%
[09:01:56] [Server thread/INFO]: Done (3,319s)! For help, type "help" or "?"
stop
[09:02:07] [Server thread/INFO]: Stopping the server
[09:02:07] [Server thread/INFO]: Stopping server
[09:02:07] [Server thread/INFO]: Saving players
[09:02:07] [Server thread/INFO]: Saving worlds
[09:02:07] [Server thread/INFO]: Saving chunks for level 'Molokai'/Overworld
[09:02:07] [Server thread/INFO]: Saving chunks for level 'Molokai'/Nether
[09:02:07] [Server thread/INFO]: Saving chunks for level 'Molokai'/The End
[09:02:07] [Server Shutdown Thread/INFO]: Stopping server

D:\MineCraft\server-test>

Please attach your console output after issuing "stop"

Chris Hazuki 2014-07-26T20:14:41Z

I can confirm this is an issue. The console output is the same except it doesn't exit.
SIGTERM is ignored so I have to SIGKILL.

Tested on:
CentOS 6.5

openjdk version "1.8.0-internal"
OpenJDK Runtime Environment (build 1.8.0-internal-0)
OpenJDK 64-Bit Server VM (build 25.0-b57-internal, mixed mode)

Chris Hazuki 2014-07-27T17:33:54Z

I just did a few tests and it only happens if people are still connected when you do /stop. Entering the command into an empty server (i.e. 0 players online) will successfully exit.

Also confirmed on:
Kubuntu 14.04

java version "1.8.0_11"
Java(TM) SE Runtime Environment (build 1.8.0_11-b12)
Java HotSpot(TM) 64-Bit Server VM (build 25.11-b03, mixed mode)

Ultimate Omicron 2014-08-08T15:09:54Z

Also happens for me on latest (32b, 32c crashes everyone for me) when there are online players.

This makes zero-maintenance batch-restarters useless, because when I remotely /stop I'm still online and it locks up, requiring me to get to the physical machine.

If command blocks could /stop, this would have a workaround.

A usable workaround, though, would be to have a sniffer program look up a magic word on server log files and kill java when it finds that.

Ultimate Omicron 2014-08-25T19:52:55Z

Was this fixed? I am currently using a sniffer to manually kill the server due to being unable to stop from inside.

17 more comments

kumasasa 2014-08-25T20:50:09Z

Not fixed, see MC-69402

Ultimate Omicron 2014-08-28T00:15:08Z

Hmm, If this is not fixed for release I might have to rely on a separate tool to shutdown the server from inside. I have been using a Game Maker executable to kill java when it finds a passphrase on the server logs to accomplish this. Might write a dedicated tool on a more appropriate platform.

If command blocks could run all commands this would not be a problem too.

angie nantz 2014-08-30T16:48:33Z

this is a problem for us as well, running 1.8 pre-release 3, the shutdown will hang if players are still connected at 'saving players....'... please fix, pretty please 🙂

Ultimate Omicron 2014-08-30T23:25:03Z

Yeah, this is a major problem for remotely managed or zero-config servers. Specially since sometimes the server will overload and then the target machine is saturated, preventing remote connections to stop the server.

Random Ness 2014-08-31T01:20:36Z

Yeah, adding the console output after doing /stop now. Its still going in the pre-releases!

Ultimate Omicron 2014-09-02T04:51:37Z

Well this is bad. Bumping for visibility.

I will now try to find a sequence of ingame actions that crashes the server, and try to automate that.

Hugger 2014-09-03T22:31:39Z

I'm experiencing this issue on my server, or what seems to be a similar issue.

My server has scheduled restarts and as it goes into the restart it 'hangs'.

23:00:08 CONSOLE: [INFO] Saving worlds
23:00:08 CONSOLE: [INFO] Saving chunks for level 'Huggercraft'/Overworld
23:00:08 CONSOLE: [INFO] Saving chunks for level 'Huggercraft'/Nether
23:00:08 CONSOLE: [INFO] Saving chunks for level 'Huggercraft'/The End
23:00:10 CONSOLE: Shutdown Thread/INFO]: Stopping server
23:00:10 CONSOLE: Shutdown Thread/INFO]: Saving players

At this point it's shut-down and has not initiated the restart sequence, I have to manually log in and restart myself.

angie nantz 2014-09-04T01:26:26Z

i had to change my scheduled restart to 'when no one is on' until this is resolved and sneak in a few manual reboots when i can.

rodney cheney 2014-09-06T09:08:04Z

im still getting this on 1.8 release just hangs there scheduled restarts are usless atm

rodney cheney 2014-09-07T15:40:24Z

For anyone using linux and bash scripts to restart servers
this command below will help stop the server so it can run freely again after stop
feel free to remove ; screen -wipe if you dont use screens

ps -ef | grep java | grep -v grep | awk '{print $2}' | xargs kill -9 ; screen -wipe

at least then you can have servers restart and not worry about having to be online to do it your self, cause if you use rcon like i do the next java instace will still be running rcon, which means it may be more then just the fact it dosent stop it could be hanging due to something else

Adrian G 2014-09-07T23:42:05Z

I can confirm this issue too, for me it actually caused all my player files to not save, so for some reason all player files where gone. It wasn't a huge deal since i have frequent backups, but seriously, this isn't a snapshot, this is a final release... this needs to be fixed ASAP.

rodney cheney 2014-09-08T06:05:03Z

here is some log i use a 64 gig dedicated server i get lag alot but i benn getting alot of errors latly here is a error that i noticed during one of the stops

[02:00:31] [Server thread/INFO]: Stopping the server
[02:00:32] [Server thread/INFO]: Stopping server
[02:00:32] [Server thread/INFO]: Saving players
[02:00:32] [Server thread/INFO]: Saving worlds
[02:00:32] [Server thread/INFO]: Saving chunks for level 'world'/Overworld
[02:00:33] [Server thread/INFO]: Saving chunks for level 'world'/Nether
[02:00:33] [Server thread/INFO]: Saving chunks for level 'world'/The End
Sep 08, 2014 2:00:34 AM io.netty.util.concurrent.DefaultPromise notifyListener0
WARNING: An exception was thrown by rk.operationComplete()
io.netty.util.concurrent.BlockingOperationException: DefaultChannelPromise@4bf99325(uncancellable)
at io.netty.util.concurrent.DefaultPromise.checkDeadLock(DefaultPromise.java:397)
at io.netty.channel.DefaultChannelPromise.checkDeadLock(DefaultChannelPromise.java:157)
at io.netty.util.concurrent.DefaultPromise.awaitUninterruptibly(DefaultPromise.java:290)
at io.netty.channel.DefaultChannelPromise.awaitUninterruptibly(DefaultChannelPromise.java:135)
at io.netty.channel.DefaultChannelPromise.awaitUninterruptibly(DefaultChannelPromise.java:28)
at gr.a(SourceFile:199)
at rk.operationComplete(SourceFile:125)
at io.netty.util.concurrent.DefaultPromise.notifyListener0(DefaultPromise.java:682)
at io.netty.util.concurrent.DefaultPromise.notifyListeners0(DefaultPromise.java:607)
at io.netty.util.concurrent.DefaultPromise.notifyListeners(DefaultPromise.java:565)
at io.netty.util.concurrent.DefaultPromise.trySuccess(DefaultPromise.java:413)
at io.netty.channel.DefaultChannelPromise.trySuccess(DefaultChannelPromise.java:82)
at io.netty.channel.ChannelOutboundBuffer.safeSuccess(ChannelOutboundBuffer.java:608)
at io.netty.channel.ChannelOutboundBuffer.remove(ChannelOutboundBuffer.java:344)
at io.netty.channel.socket.nio.NioSocketChannel.doWrite(NioSocketChannel.java:283)
at io.netty.channel.AbstractChannel$AbstractUnsafe.flush0(AbstractChannel.java:694)
at io.netty.channel.nio.AbstractNioChannel$AbstractNioUnsafe.forceFlush(AbstractNioChannel.java:321)
at io.netty.channel.nio.NioEventLoop.processSelectedKey(NioEventLoop.java:515)
at io.netty.channel.nio.NioEventLoop.processSelectedKeysOptimized(NioEventLoop.java:464)
at io.netty.channel.nio.NioEventLoop.processSelectedKeys(NioEventLoop.java:378)
at io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:350)
at io.netty.util.concurrent.SingleThreadEventExecutor$2.run(SingleThreadEventExecutor.java:116)
at java.lang.Thread.run(Thread.java:744)

still a big issue really hope someone fixes this

lapppy 2014-09-08T19:18:28Z

Based on the comment above, it could be related to MC-37586 which is when the server thinks a player hasn't left when they actually have, causing ghost players.

Chris 2014-09-08T19:54:11Z

@Jono it's not related to that bug. If I start a server and join it, then do nothing else except try and stop the server, this bug is duplicated. And like Adrian above said, it has deleted all the player and stat files at one point when the console was forcefully quit which is a major problem, esp. for people not smart enough to make backups.

rodney cheney 2014-09-12T16:22:15Z

chris is correct there two differant bugs and we should not have to make doggy scripts to kill processes to get servers to full stop

lapppy 2014-09-12T21:40:04Z

Fair enough, but let's not forget the possibility that they might be related. Code does some weird things sometimes.

Warren Liddell 2014-10-15T22:44:00Z

Bug is still present 🙂

Dlawso the Really Lucky Rabbit 2014-10-28T20:01:11Z

Is this fixed in 1.8.1-pre3?