Sometimes - I should say often, this occur 2 times out of 4 - when client using dist live webroot sample hard disconnect (i.e. closing browser window) I start to see many entries per second in edge logs :
2009-08-2320:58:1179286(w)2631008Asynchronous I/O operation failed (close: shutdown failed on conn 0x28c6a510 socket 59 with 104 (Connection reset by peer)).-
2009-08-2320:58:1179286(w)2631008Asynchronous I/O operation failed (close: shutdown failed on conn 0x28c3eca8 socket 58 with 104 (Connection reset by peer)).-
2009-08-2320:58:1179286(w)2631008Asynchronous I/O operation failed (close: shutdown failed on conn 0x28c34ae0 socket 42 with 104 (Connection reset by peer)).-
2009-08-2320:58:1179286(w)2631008Asynchronous I/O operation failed (close: shutdown failed on conn 0x28c3e268 socket 60 with 104 (Connection reset by peer)).-
2009-08-2320:58:1179286(w)2631008Asynchronous I/O operation failed (close: shutdown failed on conn 0x28c38498 socket 41 with 104 (Connection reset by peer)).-
2009-08-2320:58:1179286(w)2631008Asynchronous I/O operation failed (close: shutdown failed on conn 0x28c6ae60 socket 61 with 104 (Connection reset by peer)).-
2009-08-2320:58:1179286(w)2631008Asynchronous I/O operation failed (close: shutdown failed on conn 0x28c6a510 socket 59 with 104 (Connection reset by peer)).-
2009-08-2320:58:1179286(w)2631008Asynchronous I/O operation failed (close: shutdown failed on conn 0x28c3eca8 socket 58 with 104 (Connection reset by peer)).-
In the same time I see a fmscore process jumping up to 80% CPU. Syslog also start to work hard, even with local0.* directed to /dev/null
I let run the process several hours ad decided to kill it. -TERM didn't work -KILL do the job. It is then not possible to start a new live session, even from another IP/computer, and I see the following in core log, tens by second also :
2009-08-2320:59:1579071(e)2581394Failed to wait for process condition: errno(22).-
2009-08-2320:59:1579071(e)2581394Failed to wait for process condition: errno(22).-
2009-08-2320:59:1579071(e)2581394Failed to wait for process condition: errno(22).-
2009-08-2320:59:1579071(e)2581394Failed to wait for process condition: errno(22).-
2009-08-2320:59:1579071(e)2581394Failed to wait for process condition: errno(22).-
2009-08-2320:59:1579071(e)2581394Failed to wait for process condition: errno(22).-
2009-08-2320:59:1579071(e)2581394Failed to wait for process condition: errno(22).-
2009-08-2320:59:1579071(e)2581394Failed to wait for process condition: errno(22).-
2009-08-2320:59:1579071(e)2581394Failed to wait for process condition: errno(22).-
2009-08-2320:59:1579071(e)2581394Failed to wait for process condition: errno(22).-
2009-08-2320:59:1579071(e)2581394Failed to wait for process condition: errno(22).-
Then it is a fmsedge process that start to eat all CPU and even when killed the only way to go is to restart FMS.
This is on a QuadCore 4GB RAM host, and I'm the only connected user, so no ressources issues afaik, http proxy is sent to another server that still answer during the high load (GigaEthernet between the 2 hosts), but requesting FMS host with http doesn't answer (browser time out). The apache server receive no request from FMS proxy btw.
Any tips ?