- 论坛徽章:
- 0
|
The first difference I see is that you couldn\'t traceroute to newsmth.org while you could tracert to it. Traditionaly windows tracert uses ICMP and Unix traceroute uses UDP high ports. However, this may have been chaanged. So I need you to do traceroute -v -I -i eri0 -q 1 newsmth.org and then in a separate command, do snoop -v when you are doing just traceroute -v -i eri0 -q 1 newsmth.org. The -I option in the first run forces ICMP protocol. \r\n \r\nAlso, please telnet to port 80 of newsmth.org instead of just telnet and do a snoop -v of it. \r\n\r\nThe Solaris release you are using is Solaris 9 Dec of 2002 (really old). Did you patch it? If not, you may want at least install recommened patch cluster. Otherwise weird thing like this could happen. If you\'d like to post the patch info, you could show me uname -a and showrev -p output. \r\n\r\nThis problem is really weird. At this time I\'m still not convinced that it is necessary a server based problem even though it could be. It still could be a network problem, such as your network device is blocking UDP high ports. You have two network devices between your server and the internet, one is 192.168.11.1 and the other is 172.16.8.21 (or external IP 202.99.57.113). The most effective way of troubleshooting the problem is actually monitor the traffic on these network devices. Even it turns out that it is your server\'s problem, you can tell it better from the logs on these devices. Without accessing to them, we are dancing around to figure it out. I strongly recommend that you get your network admin involved. |
|