Stefan Lambrev wrote:
I'm having very strange problem.
I have near 200 sockets reported by netstat -An, which are NOT reported
by sockstat and fstat.
All of them look like (output from netstat -An) :
ffffff0169282000 tcp4 0 0 192.168.13.12.4965
192.168.13.3.8080 FIN_WAIT_2
I'm trying to figure out what keeps them active (tcpdump shows traffic
from 192.168.13.12 to 192.168.13.3)
The application that create them is stopped before more then a week but
request are still "flying" around.
FIN_WAIT_2 means that one side has closed the TCP connection and is waiting
for the other side to finish sending any remaining data and acknowledge the
close of the connection by sending a FIN. The TCP stack can remain in this
state until the other side acknowledges or resets the connection, even though
the local process is gone. (In fact, it's not unusual for a process to close
and then quit.)
The TCP stack is supposed to wait 2*MSL (ie, 120 seconds), and then move to
TIME_WAIT and then CLOSED, IIRC, if it doesn't hear anything back.
I have access to both IPs - 192.168.13.12 is http balancer and
192.168.13.3 is apache server (both were restarted)
Sounds like a glich in the TCP state engine of the load-balancer, probably.
This sort of thing happens when using LBs somewhat frequently....
--
-Chuck
_______________________________________________
[email protected] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-net
To unsubscribe, send any mail to "[EMAIL PROTECTED]"
