acts for unsent data > 0

[Mott Given]

How significant are 4 acks for unsent data occurring during 1 day on a V2250 with 12 CPUs? I ask because some sources say this should only be 0. The host is HP-UX v11.00. This occurred after the host was moved to a different Cisco switch.
Below are stats from "netstat -s" for tcp:
2723930 packets sent
2355815 data packets (855358939 bytes)
2131 data packets (780510 bytes) retransmitted
368085 ack-only packets (104872 delayed)
0 URG only packets
5 window probe packets
0 window update packets
25111 control packets
2514388 packets received
1832853 acks (for 855227454 bytes)
13835 duplicate acks
4 acks for unsent data
1817142 packets (361639573 bytes) received in-sequence
0 completely duplicate packets (0 bytes)
63 packets with some dup, data (22484 bytes duped)
672 out of order packets (271806 bytes)
0 packets (0 bytes) of data after window
1 window probe
661554 window update packets
4 packets received after close
0 segments discarded for bad checksum
0 bad TCP segments dropped due to state change
8107 connection requests
4222 connection accepts
12329 connections established (including accepts)
13632 connections closed (including 1535 drops)
1441 embryonic connections dropped
1820897 segments updated rtt (of 1820897 attempts)
1239 retransmit timeouts
40 connections dropped by rexmit timeout
5 persist timeouts
96 keepalive timeouts
96 keepalive probes sent
11 connections dropped by keepalive
0 connect requests dropped due to full queue
1384 connect requests dropped due to no listener

Mott Given

[Ron Kinner · Accepted solution]

I don't think it's anything to worry about. Every time I see a nonzero value I also see the same or larger number in "packets received after close." So I think it's just a case of an application dying and taking its TCP connection with it before it was completely closed.

Ron

[Mott Given]

Thanks, Ron.

Mott