Server not responding randomly - urgent help needed

[Kicksome]

I'm not sure to go with this, but it's driving me crazy.

The SFS just randomly stops responding to all requests. I have a port monitor setup but it still responds when this happens so I have no way of knowing about it until I wake up to 20 messages from upset people.

I have two different zones and both of them just stop working. These zones use totally different code. I added the second zone for some unrelated things and both just stop working. I have the wrapper set to 1 gig of memory. And it seems to make no difference.

I've tried this on two different physical servers. I can understand why one zone would lock up - due to code. But two totally different zones locking up at the same time? SFS keeps responding to requests - it just does nothing with them.

I just looked at the log closely and tracked it down to this: Once this happens - everything on that SFS instance goes to hell. Any ideas???

INFO | jvm 1 | 2007/12/02 03:17:24 | Exception in thread "EventWriter-2" java.lang.Error: java.net.SocketException: Socket operation on nonsocket: getsockname
INFO | jvm 1 | 2007/12/02 03:17:24 | at sun.nio.ch.Net.localAddress(Unknown Source)
INFO | jvm 1 | 2007/12/02 03:17:24 | at sun.nio.ch.SocketChannelImpl.localAddress(Unknown Source)
INFO | jvm 1 | 2007/12/02 03:17:24 | at sun.nio.ch.SocketChannelImpl.toString(Unknown Source)
INFO | jvm 1 | 2007/12/02 03:17:24 | at java.lang.String.valueOf(Unknown Source)
INFO | jvm 1 | 2007/12/02 03:17:24 | at java.lang.StringBuilder.append(Unknown Source)
INFO | jvm 1 | 2007/12/02 03:17:24 | at it.gotoandplay.smartfoxserver.EventWriter.processEvent(EventWriter.java:320)
INFO | jvm 1 | 2007/12/02 03:17:24 | at it.gotoandplay.smartfoxserver.EventWriter.run(EventWriter.java:72)
INFO | jvm 1 | 2007/12/02 03:17:24 | at java.lang.Thread.run(Unknown Source)
INFO | jvm 1 | 2007/12/02 03:17:24 | Caused by: java.net.SocketException: Socket operation on nonsocket: getsockname
INFO | jvm 1 | 2007/12/02 03:17:24 | at sun.nio.ch.Net.localInetAddress(Native Method)
INFO | jvm 1 | 2007/12/02 03:17:24 | ... 8 more

[Kicksome]

I've looked at the other logs - and the it has always stopped responding to new requests after this error:
INFO | jvm 1 | 2007/12/01 10:45:51 | Exception in thread "EventWriter-2" java.lang.Error: java.net.SocketException: Socket operation on nonsocket: getsockname
INFO | jvm 1 | 2007/12/01 10:45:51 | at sun.nio.ch.Net.localAddress(Unknown Source)
INFO | jvm 1 | 2007/12/01 10:45:51 | at sun.nio.ch.SocketChannelImpl.localAddress(Unknown Source)
INFO | jvm 1 | 2007/12/01 10:45:51 | at sun.nio.ch.SocketChannelImpl.toString(Unknown Source)
INFO | jvm 1 | 2007/12/01 10:45:51 | at java.lang.String.valueOf(Unknown Source)
INFO | jvm 1 | 2007/12/01 10:45:51 | at java.lang.StringBuilder.append(Unknown Source)
INFO | jvm 1 | 2007/12/01 10:45:51 | at it.gotoandplay.smartfoxserver.EventWriter.processEvent(EventWriter.java:320)
INFO | jvm 1 | 2007/12/01 10:45:51 | at it.gotoandplay.smartfoxserver.EventWriter.run(EventWriter.java:72)
INFO | jvm 1 | 2007/12/01 10:45:51 | at java.lang.Thread.run(Unknown Source)
INFO | jvm 1 | 2007/12/01 10:45:51 | Caused by: java.net.SocketException: Socket operation on nonsocket: getsockname
INFO | jvm 1 | 2007/12/01 10:45:51 | at sun.nio.ch.Net.localInetAddress(Native Method)
INFO | jvm 1 | 2007/12/01 10:45:51 | ... 8 more



although it keeps processing old requests but once that users disconnects they can't connect to the server again - no matter what zone.

Is there a way to monitor this so at least I know when it happens? At least then I can reset it. Right now it's a massive problem.

[Lapo]

Hi,
according to Java documentation:
"Socket operation on non-socket.
An operation was attempted on something that is not a socket. Either the socket handle parameter did not reference a valid socket, or for select, a member of an fd_set was not valid.

Can you give us more details about the OS and SFS version that you are using?
Are there other errors in the logs?

[Kicksome]

I'm using Windows 2003 Server with SP2 with 4 gigs memory and 2 dual core amds. I've switched servers about 3 different times and I've always got this error (Windows 2003 servers - some intel chips). I just assumed it was my code that was locking up until I just recently realized that all other zones were locking up too - but the SFS was still up - just not doing anything. I wish it would at least crash so it would restart - or stop responding to requests so port monitoring would work and I would know it was down.

I'm using Sun jre1.6.0_03 - although I've gone through other version of Sun JVM's.

There are no more errors in the log - just this exact error:
INFO | jvm 1 | 2007/12/02 03:17:24 | Exception in thread "EventWriter-2" java.lang.Error: java.net.SocketException: Socket operation on nonsocket: getsockname
INFO | jvm 1 | 2007/12/02 03:17:24 | at sun.nio.ch.Net.localAddress(Unknown Source)
INFO | jvm 1 | 2007/12/02 03:17:24 | at sun.nio.ch.SocketChannelImpl.localAddress(Unknown Source)
INFO | jvm 1 | 2007/12/02 03:17:24 | at sun.nio.ch.SocketChannelImpl.toString(Unknown Source)
INFO | jvm 1 | 2007/12/02 03:17:24 | at java.lang.String.valueOf(Unknown Source)
INFO | jvm 1 | 2007/12/02 03:17:24 | at java.lang.StringBuilder.append(Unknown Source)
INFO | jvm 1 | 2007/12/02 03:17:24 | at it.gotoandplay.smartfoxserver.EventWriter.processEvent(EventWriter.java:320)
INFO | jvm 1 | 2007/12/02 03:17:24 | at it.gotoandplay.smartfoxserver.EventWriter.run(EventWriter.java:72)
INFO | jvm 1 | 2007/12/02 03:17:24 | at java.lang.Thread.run(Unknown Source)
INFO | jvm 1 | 2007/12/02 03:17:24 | Caused by: java.net.SocketException: Socket operation on nonsocket: getsockname
INFO | jvm 1 | 2007/12/02 03:17:24 | at sun.nio.ch.Net.localInetAddress(Native Method)
INFO | jvm 1 | 2007/12/02 03:17:24 | ... 8 more


Smartfox version is 1.5.9 although this has happened with earlier versions.

[Kicksome]

I *might* be wrong about all the zones not responding to requests - it might just lock the one zone, at least in this one case I found. I noticed in this instance when it happened the other zone was still working okay.

INFO | jvm 1 | 2007/11/26 21:28:18 | Exception in thread "EventWriter-2" java.lang.Error: java.net.SocketException: Socket operation on nonsocket: getsockname
INFO | jvm 1 | 2007/11/26 21:28:18 | at sun.nio.ch.Net.localAddress(Unknown Source)
INFO | jvm 1 | 2007/11/26 21:28:18 | at sun.nio.ch.SocketChannelImpl.localAddress(Unknown Source)
INFO | jvm 1 | 2007/11/26 21:28:18 | at sun.nio.ch.SocketChannelImpl.toString(Unknown Source)
INFO | jvm 1 | 2007/11/26 21:28:18 | at java.lang.String.valueOf(Unknown Source)
INFO | jvm 1 | 2007/11/26 21:28:18 | at java.lang.StringBuilder.append(Unknown Source)
INFO | jvm 1 | 2007/11/26 21:28:18 | at it.gotoandplay.smartfoxserver.EventWriter.processEvent(EventWriter.java:320)
INFO | jvm 1 | 2007/11/26 21:28:18 | at it.gotoandplay.smartfoxserver.EventWriter.run(EventWriter.java:72)
INFO | jvm 1 | 2007/11/26 21:28:18 | at java.lang.Thread.run(Unknown Source)
INFO | jvm 1 | 2007/11/26 21:28:18 | Caused by: java.net.SocketException: Socket operation on nonsocket: getsockname
INFO | jvm 1 | 2007/11/26 21:28:18 | at sun.nio.ch.Net.localInetAddress(Native Method)
INFO | jvm 1 | 2007/11/26 21:28:18 | ... 8 more