Author Topic: Wine issues with posting to master server and some other stuff  (Read 1644 times)

I'm hosting a blockland server on ubuntu server 12.04 LTS.

It has worked fine until recently when it has experienced higher traffic. After a bit, some bits of networking appear to stop working or something, and whenever the server tries to post, I get an error similar to this:

Quote
fixme:winsock:WSACancelAsyncRequest (0xf5a9),stub

You can connect to the server directly with IP and see chat messages, but you can never get past the "Connection Accepted" message. The players inside the server can still play perfectly fine without any obvious issues.

I'm sure I'm not the only one running blockland servers on Linux, so is there anyone with an idea of what's going on here?

noedit: running as root doesn't make a difference

I ended up just switching to Windows Server.



Strange. I've had 32 players using that and had no issues.

Okay, there's still something goofy going on.
Even after switching to windows server, there are issues with authentication and posting to the master server.

It will fail and retry, but only sometimes actually work. It's strange, because all the servers appear to be working fine everywhere else. I don't see any obvious cause because it seems to randomly decide when to work.

noedit: I get a line in the console saying "Connect error: 10055" whenever it fails.

noedit: I get a line in the console saying "Connect error: 10055" whenever it fails.
Interesting.
Does it give 'WSAENOBUFS (10055)' as well?
Because that's apparently the only error code that corresponds to that. It's apparently a 'no buffer space available' notification.
In a nutshell, it's the following:
Quote
Error 10055 means that Windows has run out of TCP/IP socket buffers because too many connections are open at once.

Post the result of cat /proc/sys/fs/file-max

Post the result of cat /proc/sys/fs/file-max
The server is running on Windows Server 2003 x64 now.

Interesting.
Does it give 'WSAENOBUFS (10055)' as well?
Because that's apparently the only error code that corresponds to that. It's apparently a 'no buffer space available' notification.
In a nutshell, it's the following:
Strange.

noedit:
There were apparently three connections on SSH that weren't properly closed, and I've now turned that off entirely since I don't need it anymore.

Other than that, there is a python script listening locally, but afaik the only thing left is the Blockland server. Strange. I guess I'll just have to see if the problem is solved.

noedit again sorry :(

The only issue at this point appears to be with connecting to the authentication servers. Even if I catch it during an incident, the server can successfully post to the master server, but the authentication connection fails every time.

And just now, it has actually failed all five times on authentication again. Weird.

Connect to auth.blockland.us and give results.

Connect to auth.blockland.us and give results.
With a TCPObject or just on a webpage? Regardless, it's not having issues at the moment. Next time I catch it happening, I'll get a netstat report or something. Right now everything is working fine. The issue seems to go away for a bit when I restart. I think the issue may begin once there are more players on the server.