Read time out

[cuddylier]

Why do I get the error "Read time out" when using dynmap and my server freezes and needs restarted? It is definately dynmap.

[Carve]

I second this issue, it seems to be an issue with rendering threads going crazy mike, I'll look and see if I can spot anything out of place in the last few rev's. If I find anything I'll submit a pull request and let you know in IRC.

[mikeprimm]

I need some details on actually producing this - clearly, its not being seen many places, as I'm not hearing much beyond this (and, God knows, people complain quickly on this sort of thing). Also, on render threads 'going crazy' - not clear what you mean, nor how this would drive anything with the foreground thread behavior timing out.

[Carve]

mike,

I've ran into read timed out issues in the past week and I've also done some research with the NoLaggExamination plugin and it's showing Dynmap (on a regular 500 tick examination) is taking upwards of 200ms, compared to 20-50ms on any other demanding plugin, the only other plugin that could possibly be causing this is Towny, but even it was only utilizing a 75-100ms time range. I'll see if I can provide you more details when you get off work / get on IRC just message me.

thanks, Carve

[cuddylier]

I do in deed have towny, after I removed dynmap my server hasn't crashed once for 3 days whereas with dynmap it was crashing every 5 hours. There was absolutely no errors so no help there. I have 13GB RAM assigned to the server and the server was using only about 30% CPU with dynmap installed with 30-40 players online so hardware isn't the issue.

[mikeprimm]

@Carve: 200ms total in 500 ticks, or 200ms at a time? 200ms out of 500 ticks is pretty tiny, given that what we need to do on the main thread is almost entirely requesting chunk loads. If 200ms at a time, what granularity is it showing (is it processing a specific event, or a single scheduler/call sync call)?

@cuddylier: "crash" is a desperately not useful statement - I really need to hear more details on what did happen, and with what version of us and of CB. Understand that this code is running on literally tens of thousands of servers, and the forum isn't overloaded with complaints (its actually quite quiet with regard to server crashes or hangs), so the details are everything on this. If the server hung, and you can get to the server console, check if you can do a 'Ctrl-backslash' and get a thread dump, so we can see where the server was processing when it hung.

[cuddylier]

There is absolutely no errors, my server just suddenly says "Read timeout" in the console and the server is frozen and accessible.

[flyingflare]

This sounds similar to what I have just begun to experience. When I view the map online from Chrome now, If I do some moving around or zooming, suddenly my computer hangs for at least 30 sec and I get kicked from the server, but it doesn't crash. Here is the story in steps:

1. I set a border of 5000 to my world and pre-generated about 50% of it, than ran a full render from dynmap. Previously, I had not pre-generated the world and a small server of about 10 people on quite a powerful rig had no problems rendering and viewing HiRes online.
2. After pre-generating the world and fully rendering it, now, if I view the map online and scroll around or zoom different levels, my computer will quickly hang for more than 30sec, become totally unresponsive, and I will get kicked from the server with a read-time out error on the client. There is no error on the console. After the computer becomes responsive again, i can continue looking at the map for a short time again, but then it will stop the computer again, same situation, even if I'm not logged into the game.

I am running craftbukkit 2149 and dynmap 36.3

[cuddylier]

My server experiences the hanging and the read timeout, maybe what your trying to get at. My client has no problems, just connection lost when the server hangs.

[ghost]

I had a similar problem, after disabling the dynmap plugin it didnt occur anymore, Here are the errors that I got, then the server just froze, no one could get online anymore and the server needed a restart:

2012-05-02 12:54:46 [SEVERE] java.lang.NullPointerException 2012-05-02 12:54:46 [SEVERE] at net.minecraft.server.NetServerHandler.sendPacket(NetServerHandler.java:723) 2012-05-02 12:54:46 [SEVERE] at net.minecraft.server.EntityTrackerEntry.updatePlayer(EntityTrackerEntry.java:235​) 2012-05-02 12:54:46 [SEVERE] at net.minecraft.server.EntityTracker.updatePlayers(EntityTracker.java:146) 2012-05-02 12:54:46 [SEVERE] at net.minecraft.server.MinecraftServer.w(MinecraftServer.java:556) 2012-05-02 12:54:46 [SEVERE] at net.minecraft.server.MinecraftServer.run(MinecraftServer.java:449) 2012-05-02 12:54:46 [SEVERE] at net.minecraft.server.ThreadServerApplication.run(SourceFile:492) 2012-05-02 12:54:46 [SEVERE] Unexpected exception java.lang.NullPointerException at net.minecraft.server.NetServerHandler.sendPacket(NetServerHandler.java:723) at net.minecraft.server.EntityTrackerEntry.updatePlayer(EntityTrackerEntry.java:235​) at net.minecraft.server.EntityTracker.updatePlayers(EntityTracker.java:146) at net.minecraft.server.MinecraftServer.w(MinecraftServer.java:556) at net.minecraft.server.MinecraftServer.run(MinecraftServer.java:449) at net.minecraft.server.ThreadServerApplication.run(SourceFile:492)

[mikeprimm]

@fplattel - this is a Bukkit bug, recently fixed (see https://github.com/Bukkit/CraftBukkit/commit/60497e0a2a837b9caa9733ee87a2d34c69f95c51) BUKKIT-1579 is the ticket

[sirrus86]

@flyingflare - I'm having the exact same issue, only I'm viewing a dynmap of someone else's server. Zooming or clicking anywhere in the map maybe two or three times causes dynmap to freeze and the server connection to timeout for 30sec to several minutes. What's strange is I'm the only person I know of on my server with this issue.

My computer has high specs, I have a stable internet connection, and the server is located near me, yet we have a player in the Netherlands with a sub-par computer and spotty internet who's never run into this issue. What's the deal?

[mikeprimm]

Closing this out - any of you guys still having issues with 0.38 release code, open separate defects: the descriptions here are more or less incoherent, and don't suggest anything common.