Ok. Where to start? Aside from the fact that I'm new here, hello =)
First off, I've hunted for this problem, but it's hard to quantify the issue into a few searchable words, so the results that i receive are either entirely useless or way too focused, and invariably off base, to be useful.
That said:
In our datacenter cabinet we have a number of servers, all on the same public address space, and everything except for the FC11 box is behaving, so I know that it's not the datacenter.
Basically, it went like this - I had Kubuntu (feisty, i believe) running on the 1850 initially, running Bugzilla, GLPI, and SVN, but i was having some serious troubles with the remote desktop controls, and the Red Hat utilities from Dell seem to be working on Fedora rather well, so i decided to switch.
I built the server here in my office, using a spare 1850 that i have for that very purpose, moved my repositories, the bugz and the GLPI databases, and made sure that it could be reached via one of our public IP's from the outside world. It worked flawlessly.
The I took it to the datacenter, swapped the drives, and changed the box's network settings to be identical to the settings of the previous Kubuntu box. That's when the strangeness started.
The box cannot see the outside world. The outside world cannot see the box. My domain points to the box's public IP, and it doesn't resolve.
However, if I RDC to one of the XServes we have running in the closet, i can access everything - the host name of the box resolves, and i can hit webmin to make changes.
So i definitely have access to the box, i just cannot for the blasted life of me figure out why something from a 64.88.xxx.xx address can reach it, and something from any other address cannot, and why the box itself, while responding to ping (so long as the pinger is 64.88.), cannot... i actually forgot where this thought was going. But it responds to everything, the bugs.mydomain.net hostname resolves (we are
not running our own DNS), i can SSH to is, but
only if i am remotely connected to another server in the closet!
(Oh, this behaviour continues if the firewall is disabled, if the box is set to a different IP - though the hostname, of course, doesn't resolve - or any number of other attempts that i no longer remember what i tried.)
- edit -
We do not use NAT or any hardware firewalls on this segment of the cabinet - these machines maintain their own software firewalls, and are assigned a direct public ip and gateway. There is no address/port forwarding, there are no strange rules or mappings there is no shaper. It's pretty basic really.
- edit the second -
A traceroute to the box shows that it's not making it to the next hop past the main router at EDC/QT (the DC) - which means that the route is not the issue... it's the box. The internal routing in the cabinet is handled by a Cisco managed 24-port switch, but it's not doing anything to it either.
I am 100% positive that this is something so unbelievably stupid that I'm going to kick myself when it's exposed, but if anyone can lead me in the right direction I would be extremely grateful.
Thanks for taking the time...
-/A\V/