[1.3b60+] Network/Boot Bug Discussion

[Vry]

- Maptool Build Number 1.3b62
- OS Vista Home Premium 64 Bit SP2
- MTStartup Memory Settings 1024 Meg
- Networking Card Realtek RTL8168C(P)/8111C(P) Family PCI-E Gigabit Ethernet NIC (NDIS 6.0)
- Router Linksys WRT54G
- Video Card NVIDIA Geforce 9800 GT
- Java version 1.6.0_07-b11
- Make/Model of PC HP m9500t
- Internet speeds Download: 6.42 Mb/s
Upload: 0.98 Mb/s
Ping: 77 ms
- If you use other chat apps along with Maptool mIRC
- If you use any VOIP apps (voice) along with Maptool None
- State if you were hosting the session or a client. Host (at first, then both)

Last night I had a session going to test an isometric map. While I was tabbed away in photoshop fixing some problems with the map someone connected but never said anything. He was there on the list for a few hours but I don't think he was actually connected as the network indicator wasn't flashing that it was uploading (it was a 27 meg campaign file so it should have taken a while).

After changing to a new version of the map I tried connecting with a second client. The second client locked up on trying to load the campaign and never showed the connection at all on the host client. Shortly after, the host client locked up as well.

[dorpond]

Vry, let me see if I understand this correctly. You were hosting and then you tried connecting another local client? Was that on the same PC or another in your local LAN? Also, when you say someone connected and you didn't see the upload lights flashing, were you using the repository at the time? The reason I ask is because I recently identified that when using the repository, the link lights don't flash.

Thanks for the update.

[Vry]

The second client that failed to connect was on the same PC as the host.

It was able to connect fine earlier in the day. But after that other person connected (not sure if he clicked the wrong server or was just looking around at open servers) I wasn't able to connect even from the same computer.

No repository. Locally hosted campaign consisting of a 19.5 meg jpg map and a couple dozen .png tokens with a few object .pngs.

[dorpond]

Very interesting.. Thanks for the info up Vry.

One last question if I may - how many other clients were connected at that time?

[Vry]

It was just the host/GM, that other person who came in but likely wasn't really there anymore (don't remember his name, Dark something), and then the second client on the same machine with the GM that tried to connect.

This wasn't an actual game, just a test.

[Slander]

I was going to ask if there was any kind of trace output for Maptool when I happened to check the Maptool folder and found the log file.

These are three different errors that were written to the log on the day we had problems. The errors occurred several times throughout the night, but the only errors [in the log] were of these three types. Unfortunately, at this point, I can't correlate an error in the log with a particular problem, or if these errors are even specifically related to the problem, so I don't know how much this will help.

Code: Select all

2009-12-03 20:38:43,640 [server.MapToolServerConnection:60] ERROR - Handshake failure: java.net.SocketException: Connection reset
java.net.SocketException: Connection reset
	at java.net.SocketInputStream.read(Unknown Source)
	at java.net.SocketInputStream.read(Unknown Source)
	at com.caucho.hessian.io.HessianInput.read(HessianInput.java:1657)
	at com.caucho.hessian.io.HessianInput.readObject(HessianInput.java:1083)
	at net.rptools.maptool.server.Handshake.receiveHandshake(Handshake.java:50)
	at net.rptools.maptool.server.MapToolServerConnection.handleConnectionHandshake(MapToolServerConnection.java:53)
	at net.rptools.clientserver.simple.server.ServerConnection$ListeningThread.run(ServerConnection.java:225)
2009-12-03 20:38:43,687 [server.MapToolServerConnection:60] ERROR - Handshake failure: java.net.SocketException: Connection reset
java.net.SocketException: Connection reset
	at java.net.SocketInputStream.read(Unknown Source)
	at java.net.SocketInputStream.read(Unknown Source)
	at com.caucho.hessian.io.HessianInput.read(HessianInput.java:1657)
	at com.caucho.hessian.io.HessianInput.readObject(HessianInput.java:1083)
	at net.rptools.maptool.server.Handshake.receiveHandshake(Handshake.java:50)
	at net.rptools.maptool.server.MapToolServerConnection.handleConnectionHandshake(MapToolServerConnection.java:53)
	at net.rptools.clientserver.simple.server.ServerConnection$ListeningThread.run(ServerConnection.java:225)
2009-12-03 22:07:21,578 [server.MapToolServerConnection:60] ERROR - Handshake failure: com.caucho.hessian.io.HessianProtocolException: unknown code for readObject at 0x61 (a)
com.caucho.hessian.io.HessianProtocolException: unknown code for readObject at 0x61 (a)
	at com.caucho.hessian.io.HessianInput.error(HessianInput.java:1690)
	at com.caucho.hessian.io.HessianInput.readObject(HessianInput.java:1170)
	at com.caucho.hessian.io.MapDeserializer.readMap(MapDeserializer.java:114)
	at com.caucho.hessian.io.SerializerFactory.readMap(SerializerFactory.java:361)
	at com.caucho.hessian.io.HessianInput.readObject(HessianInput.java:1153)
	at net.rptools.maptool.server.Handshake.receiveHandshake(Handshake.java:50)
	at net.rptools.maptool.server.MapToolServerConnection.handleConnectionHandshake(MapToolServerConnection.java:53)
	at net.rptools.clientserver.simple.server.ServerConnection$ListeningThread.run(ServerConnection.java:225)
2009-12-03 22:07:32,515 [server.MapToolServerConnection:60] ERROR - Handshake failure: com.caucho.hessian.io.HessianProtocolException: unknown code for readObject at 0x61 (a)

Am I correct in thinking that the only exceptions written to the log file are actual exceptions that get thrown to the client UI? If so, are there any application switches I can turn on that might write a more detailed trace file as the app runs? I know for the most part the problems we had didn't match up with any thrown exceptions. It was just working one minute, and then not working the next.

If there is such an option, I'd happily run my games with a detailed log when we start back up to help pin down the problem. Since we play on a LAN, I could get everyone in the group to do the same.

[LaserWolf]

I was using b62 today and, while I didn't have the usual problem, it was similar.

Very early in today's session 3 out of 4 clients were simultaneously reported disconnected and the Skype call also dropped at the exact moment. Interestingly, the only client not reported as dropped is one of the players who I believe has caused the crash before (and whose connection was a little flaky today). Skype recovered immediately and I received a text and was able to resume the call only moments later. Players were unable to reconnect to my MT server, though, and attempting to disconnect the server threw the mysterious empty errors. I had to kill the app and restart.

I'd call this maybe a minor improvement, because at least this time I knew right away that something was wrong. Minor inconvenience that Skype was taken down along with it, though.

I've asked this player before about his setup and didn't get a response. I'll let you know if I find out anything more.

[Mrugnak]

Here's that follow up I promised for the GM computer:

> > - Maptool Build Number:
b61

> > - OS:
WinXP SP3

> > - MT Startup Memory Settings:
Default?

Ed. note: He uses the jnlp launch link from the MT website, apparently with no changes.

> > - Networking Card:
Atheros AR5006X Wireless Network

> > - Router:
D-Link DI624M

> > - Video Card:
Radeon x1600 Mobility

> > - Java version (example: build 1.6.0_17-b04):
Java SE6 U15 6.0.150.3

> > - Make/Model of PC:
Acer Ferrari 5000 w/ AMD Turion

> > - Internet speeds (http://www.speedtest.net/):
15MB/S download; 2 MB/S upload

> > - If you use other chat apps along with Maptool:
OpenRPG 1.7.1.

> > - If you use any VOIP apps (voice) along with Maptool:
None.

I had just created a new map and turned on Fog of War when the disconnect happened.

[Merkuri]

Our group just had a lockup, playing with b62. The GM and one player are on the same network, one player is in Japan, and a third player is out in the US somewhere. The "somewhere else in the US" player lost his internet connection (at which point, as far as we can tell, the game continued on). He regained it and tried to reconnect, and it told him his username was already in use.

At some point after the disconnect, not longer than a few minutes (possibly when the disconnected player tried to reconnect), nobody was able to receive new messages from the server. It took us a while to figure it out. The GM had to restart MapTool. The rest of us were able to simply disconnect and reconnect.

The important part, I think, was that the problem player disconnected several messages before we all received the last message from the server. The three of us who didn't lose internet all agreed on the last message that was sent, but the person who did lose internet didn't see that message, nor the three or four messages before that. I have a feeling we all lost our connection to the server when he tried to reconnect and got the "your name is already in use" message.

Edit: Another piece of information... that same player got disconnected due to internet problems three or four times before this and no one else was affected. He did not get the "your name is already in use" message during these other times. He only got it during the instance where the server locked up.

[Thanlis]

We just had a b62 lockup, as per Merkuri's description (but not the same game). The player who lost connection was in the US. We relaunched the server, and said player is OK, but another player is now experiencing lag.

[Merkuri]

Just happened a second time, same game, same player.

The problem player got this message when he disconnected:

ava.lang.IllegalArgumentException: Cannot set a null TableModel
at javax.swing.JTable.setModel(Unknown Source)
at net.rptools.maptool.client.ui.ConnectToServerDialog$4.done(ConnectToServerDialog.java:192)
at org.jdesktop.swingworker.SwingWorker$5.run(Unknown Source)
at java.awt.event.InvocationEvent.dispatch(Unknown Source)
at java.awt.EventQueue.dispatchEvent(Unknown Source)
at net.rptools.maptool.client.swing.MapToolEventQueue.dispatchEvent(MapToolEventQueue.java:24)
at java.awt.EventDispatchThread.pumpOneEventForFilters(Unknown Source)
at java.awt.EventDispatchThread.pumpEventsForFilter(Unknown Source)
at java.awt.EventDispatchThread.pumpEventsForHierarchy(Unknown Source)
at java.awt.EventDispatchThread.pumpEvents(Unknown Source)
at java.awt.EventDispatchThread.pumpEvents(Unknown Source)
at java.awt.EventDispatchThread.run(Unknown Source)

He said he also got a "name already in use" message again when he tried to reconnect. And once again he disconnected about five messages before the rest of us did.

[dorpond]

Three words: What the frig...


Man, I sure would like to know what is going on but I can't seem to figure out why some groups have all these issues while others don't. The only common ground seems to be a problematic client who drops internet a lot.

So let me see if I understand this: these problems cropped up for all of you at B60? In other words, if you go back to b56 or whatever, the problem goes away completely? Or do we not know that for sure?

[Thanlis]

So let me see if I understand this: these problems cropped up for all of you at B60? In other words, if you go back to b56 or whatever, the problem goes away completely? Or do we not know that for sure?

I do not know that for sure, but I never had the problem before b60 across over ten games hosted.

Detailed report:

- Maptool Build Number: 1.3b62
- OS: Mac OS X 10.6.2
- MTStartup Memory Settings: 1024M max,512M min, 8M stack
- Router: Linksys WRT5110, Firmware Version: 1.0.04
- Network card: Airport (i.e., wireless)
- Video card: ATI Radeon X1900
- Java version: 1.6.0_17
- Make/Model of PC: Mac Pro
- Internet speeds: Download: ~20 Mb/s
- Upload: ~3 Mb/s
- If you use any VOIP apps (voice) along with Maptool: Ventrilo and Skype
- State if you were hosting the session or a client: Host
- Macros: yes, Veggiesama's 4e framework

[Phergus]

I notice that the Hessian library we are using is 1.5 years behind the current library.

IIRC, the last time Trevor updated Hessian lib versions he had to do some rework of MT for it to work.

But for the adventurous there are 3.2.0 and 3.2.1 versions on the hessian.caucho.com website plus 4.0.1 and 4.0.2. Experiment at your own risk.

[LaserWolf]

I hadn't experienced this problem before b60, but I don't know if rolling back would help. Interesting comment from Merkuri about the issue possibly occurring on reconnect. I'll see if that matches my experience.