TCP connection from Server is getting reset intermittently

[Amit K]

We are trying to upload a file from one solaris box to HP UX 11.11 server using TCP communication.
if we try to upload the file 5 times. connection is getting timed out 3 times.
As per application team; they suspect that the socket connection is getting terminated.
--------------------------------------
"when it receives data on the same socket, it sends a TCP RST with cause "new data received when detached". The issue here is that the remote side (Portal) does not send a FIN segment to gracefully terminate the connection"
this is an intermittent behavior.

Can anybody help me on this?

I had also attached trace logs. Please check the same.

[Michael Steele_2]

HI

I don't think you need to get down into the packet commands. When you do then you're questioning the O/S itself and you have to assume the O/S works.

a) traceroute
b) verify same speed and duplex of all hops
c) swap out the cable between the server and the switch with a known good cable
d) use a different port on the switch

Also, use netstat and check for collisions and bad packets. Lets see how many you're collecting.

[Matti_Kurkela]

Duplicate thread.

Based on the trace, the problem looks to me like a HTTP keepalive connection is timing out just as advertised by the server, and the client-side is not dealing with it correctly.

The client tries to send further POST requests using an old connection that is already half-closed. At that point, the server cannot answer any more, so its only choice is to force-close the connection with a RST.

More detailed analysis at:
http://forums.itrc.hp.com/service/forums/questionanswer.do?threadId=1373920


MK

[Michael Steele_2]

keepalive is to the default router and may cause a reboot of the box if not patched properly. You would be getting time out alarm or a server not responding to ping alarms, for that is what a keepalive is, a ping to the default router.

Question: Are you using load directors to this box?

I've seen a bad load director / balancer cause the 3 servers attached to it act like this, but the keepalive was 30 seconds and getting delayed travelling through the load director