Hi there,
Great site you have here, I have found it most helpful for a number of years.
I have a very strange problem that my server provider and cPanel support can't help me with.
Here's the story....
I have a handful of servers all clustered together (2 nameserver machines and the rest are hosting boxes). I have recently had one of these boxes reformatted with CentOS 5 op sys. The others are using a combo of red-hat and fedora.
I successfully clustered the new box to my main nameserver system and returned the next day (no changes were made to either box) to find that the new box returned a cluster error (no route to host).
I completely disabled iptables and apf firewall on both boxes and the problem still exists. I then contacted cPanel support to see if they could help. No luck there, cPanel pretty much said that it was a network problem and I should contact my server provider.
That I did, it's been 5 days now and they still can't find the cause of this.
Now the problem is that I cannot ping either machine from each other, the nameserver box ping command just hangs and returns nothing. The client box (new reformatted machine) return the same no route to host error.
This occurs with iptables and apf firewall stopped (service stop command). I have configured iptables and apf to allow each IP full access as well.
Since the new box was reformatted, I have changed the hostname from it original (linux8.myserver.com to linux4.myserver.com). I was getting strange email errors stating that my host name was resolving to the wrong IP. Ironically, the server host name I have used was actually used on another box I previously had in the past, resolving to that exact IP it was on (I have not had this box for over a year).
No problems I thought, went through all my bind config files and searched for the redundant IP and hostname. Strangely there were no references to these anywhere.
My server provider resolver servers I thought must have the outdated entires perhaps. So, I thought bugger this, I'll just change the host name back to linux8.myserver.com.
Still the problem exists (even with all firewalls completely stopped).
Now here is the weird part. I can still ping each of my other machines from both boxes. But when trying to ping the two from each other it is a no goer. I can even cluster another box to the new reformatted box with no problems.
I suspect a firewall somewhere in-between my two boxes are causing this, a NIC conflict should cause the pinging to not work at all right?
It has me stumped as it was all working for the first 24 hours and strangely just stopped.
Anyone have any ideas?