Patch Install Breaks TCP/IP Connectivity

Question

Today (8/29/02) I installed the latest patch cluster for Solaris 2.8 (Sparc). Ever since I installed it, I have been unable to telnet or establish a socket connection from my Solaris system (port 10014) to a particular TCP/IP device. This is not a computer but a satellite reciever developed to allow a TCP/IP socket connections that receive data from the satellite feed.
I have a second Solaris 2.8 (Sparc) box which I have not applied the latest pactch cluster to yet, and it can connect fine.
When I telnet to the device, Telnet gets hung at the "Trying..." prompt. I checked netstat -a and the state of the connection is SYN_SENT.
I used "snoop" to monitor the connection and the patched Solaris box is sending connection requests but receiving no responses from the device. The unpatched Solaris box is getting a response from the device. Both boxes appear to be sending the same information to the device.
The both Solaris boxes can ping the device.
The use to work on the pached Solaris box, before I applied the patches.
Both Solaris systems and the device are on the same network. No routers or external firewalls separate them (in fact they are connected to the same switch).
1. Does anybody have any idea why this is happening? Is there a configuration that I am missing? (I search through the new files on the patched Solaris box looking for one).
2. Does anybody have a way to uninstall the entire patche set? There does not appear to be an "uninstall_cluster" script in the zip file.

Answer 1

I think I have found what appears to be the problem (although I am not sure how to resolve it).
I compiled and installed tcpdump on both Solaris boxes. I then wrote a little perl script which does the following:
#!/usr/bin/perl
use Socket;
socket( SH, PF_INET, SOCK_STREAM, getprotobyname( 'tcp')) || die $!;
my $dest = sockaddr_in(10014, inet_aton( "147.81.85.155"));
connect( SH, $dest) || die $!;
send( SH, "\r\n", 0);
while(1)
recv( SH, $buffer, 128, 0 );
print "$buffer";
(simply connects to socket and writes out bytes read)
I then ran "tcpdump -vvv -l host 147.81.85.155" and got the following output:
11:50:30.496927 nattyboh.32942 > hcsn-monitor6.wsicorp.com.10014: S [bad tcp cksum 2175!] 2943092708:2943092708(0) win 24820 <nop,nop,sackOK,mss 460> (DF) (ttl 64, id 41884, len 48)
11:50:33.860142 nattyboh.32942 > hcsn-monitor6.wsicorp.com.10014: S [bad tcp cksum 2175!] 2943092708:2943092708(0) win 24820 <nop,nop,sackOK,mss 460> (DF) (ttl 64, id 41885, len 48)
11:50:40.610137 nattyboh.32942 > hcsn-monitor6.wsicorp.com.10014: S [bad tcp cksum 2175!] 2943092708:2943092708(0) win 24820 <nop,nop,sackOK,mss 460> (DF) (ttl 64, id 41886, len 48)
11:50:54.110248 arp who-has hcsn-monitor6.wsicorp.com (Broadcast) tell nattyboh
11:50:54.110661 arp reply hcsn-monitor6.wsicorp.com is-at 0:a0:e1:98:19:e3
11:50:54.110691 nattyboh.32942 > hcsn-monitor6.wsicorp.com.10014: S [bad tcp cksum 2175!] 2943092708:2943092708(0) win 24820 <nop,nop,sackOK,mss 460> (DF) (ttl 64, id 54229, len 48)
11:51:21.110163 nattyboh.32942 > hcsn-monitor6.wsicorp.com.10014: S [bad tcp cksum 2175!] 2943092708:2943092708(0) win 24820 <nop,nop,sackOK,mss 460> (DF) (ttl 64, id 54230, len 48)
Note the "bad tcp cksum" message. Also that the device is responding to an arp request from the Solaris box. When I ran the test script and tcpdump on the unpached solaris system, the tcpdump looks like the following:
11:51:57.435625 arp who-has hcsn-monitor6.wsicorp.com (Broadcast) tell tsingtao
11:51:57.437165 arp reply hcsn-monitor6.wsicorp.com is-at 0:a0:e1:98:19:e3
11:51:57.437205 tsingtao.51997 > hcsn-monitor6.wsicorp.com.10014: S [tcp sum ok] 3544242549:3544242549(0) win 24820 <nop,nop,sackOK,mss 1460> (DF) (ttl 64, id 10901, len 48)
11:51:57.464533 hcsn-monitor6.wsicorp.com.10014 > tsingtao.51997: S [tcp sum ok] 772727198:772727198(0) ack 3544242550 win 4096 <mss 1460> (DF) (ttl 64, id 10904, len 44)
11:51:57.464577 tsingtao.51997 > hcsn-monitor6.wsicorp.com.10014: . [tcp sum ok] 1:1(0) ack 1 win 24820 (DF) (ttl 64, id 10902, len 40)
11:51:57.464926 tsingtao.51997 > hcsn-monitor6.wsicorp.com.10014: P [tcp sum ok] 1:3(2) ack 1 win 24820 (DF) (ttl 64, id 10903, len 42)
11:51:57.466583 hcsn-monitor6.wsicorp.com.10014 > tsingtao.51997: . [tcp sum ok] 1:1(0) ack 3 win 4096 (DF) (ttl 64, id 10906, len 40)
Which is what I would expect.
Any suggestions on how to resolve this problem?