Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations SkipVought on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

Critical problem with tcpip (FIN_WAIT_2)

Status
Not open for further replies.

17

IS-IT--Management
Jul 6, 2001
3
0
0
UY
Hi everyone,
we're in trouble with some sockets that never close.

We need to close this connections (shown with the following command: netstat | grep -i node_commander | grep -i FIN_WAIT_2):

tcp4 0 0 f1n3.34148 node_commander.6666 FIN_WAIT_2
......

We don't know hot to associate a PID number to this connections, maybe someone have the answer.

Anoter option is using the ndd command, but it is not provided with AIX. With this command you can find all your connections which are in FIN_WAIT_2 and then kill them. Anyone knows a similar command for aix?

Tks in advance,
norbes.

 
Looks like you are having a problem with application not closing out the socket.
I feel you should try to solve the problem instead of avoiding.

If you want to findout the PID then you can use a freeware lsof which can be downloaded from net. IBM does not support it. But using lsof you can find out the PID.

For permanently solving the probelm check your "thewall" setting and the Network adapter speed (which should be set properly i.e if the switch is 100 Full_duplex then you should set your network adapter to 100 Full_duplex.) There can be many other things. Just this thing I can think from top of my head

 
Thanks anshudad for your fast answer. We have properly seted up our network device. I don't know what you exactly mean with "thewall". We'll look for this later. But now, let me clarify you the following:
The -U option of lsof show us all the sockets, but we can't figure out which of those lines are whar we are looking for. Because none of them says "FIN_WAIT_2".

We solve some performance problem in our app. and that speed up the amount of concurrent connections with the node-commander so this server isn't able to close properly the conections we open. Before this increases in performance, this problem doesn't exist.

I hope you can help us a little more, and we are really glad to receive your help.

Let us know if you need some other details... regards in advance, Norbes.
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top