Strange Site Loading Problem

I don't really know where to start asking this... but I guess we have to start somewhere.

I have a dedicated server which has quite a few people on there. Since about 2 days ago, CSF started acting up on me blocking "named" service and all. After that, it seems that about 12 hours ago, something else went wrong. This time it basically went a little more drastic.

All our sites on the Shared IP is having problems loading. It would connect and all but there's like no data transferred from the server to the client and after a long wait, it just time-out. All our other sites which is on a Dedicated IP works totally fine. Connection is good, sites loading fine and all. Just only the sites which was on our Shared IP is going a little crazy.

Have you guys ever encountered anything like this before? This is really a strange problem which I've never came across before. Not too sure what is happening but even after moving 1 of the site from the Shared IP to a Dedicated IP, it is still not loading up correctly and all.

Please do help if you have seen or dealt with this problem before.

Thanks a bunch

Well you should start by going through logs. Look into /var/log/httpd/error.log and domain logs. Without errors we can only guess what went wrong.

You should have been investigating when CSF began acting, as that meant something is going on.
I suggest you to make user backups.

Here's what I have recently found out...

When I moved my sites from 1 IP to another IP, it works perfectly well. There's nothing in the httpd logs and all. Nothing is wrong at all.

The only possible explanation which I could think about is that someone on the same network is trying or is actually using my IP. Would that be a possible explanation? Since the provider does not isolate IP addresses and all(no vLAN or DHCP), I don't know how they control things but all I know is that I can't use their given IP.

Further to that, it just seems strange that if on that IP alone all the sites are not reachable at times and sometimes it loads up just fine but if you click on links or etc, they will timeout. But when I change IP's for the sites even if sharing on another IP, everything still works.

It's just totally strange.

Was just wondering if is common for IP Conflicts to have such characteristics?

aodat2 said:

The only possible explanation which I could think about is that someone on the same network is trying or is actually using my IP. Would that be a possible explanation?

Click to expand...

Yes. Which is why we and many other providers use VLANs for dedicated servers. However with the current IPv4 shortage and the impossibility of hosting companies to depend entirely on IPv6, some companies are now redefining their networks to allow multiple servers on the same networks; it saves 3 IP#s per subnet.

Since the provider does not isolate IP addresses and all(no vLAN or DHCP), I don't know how they control things but all I know is that I can't use their given IP.

Click to expand...

If that's the only issue; make it theirs; they need to fix it because you can't. But before you decide it's their issue and not yours, start by completely turning off CSF and check to see if that resolves the problem.

Further to that, it just seems strange that if on that IP alone all the sites are not reachable at times and sometimes it loads up just fine but if you click on links or etc, they will timeout.

Click to expand...

Expected behavior if there's sometimes no traffic to the other system using the same IP#; internal routers reload their tables on a (usually defined) schedule, and could assign a route to a different NIC each time depending on most immediate traffic.

But when I change IP's for the sites even if sharing on another IP, everything still works.

Click to expand...

Because that IP# is managed by your server.

Note, though, that this scenario, while possible, is by no means proven. And that most cheap internal routers don't offer a view into the routes which would definitively see the problem for what it is.

Also, I'm NOT a network engineer, and I haven't asked our datacenter network engineer; this is my own explanation. If anyone has more to share, please do so.

Jeff

Thanks for everyone's help...

Finally the host admit that it was their problem and that they are responsible for it because of a caching problem on their switch.

Stupid ain't it?

I did finally check with my network guy; whenever they reissue an IP# which someone has had before they need to clear their router cache.

Good discussion here (tcpipguide.com).

They've probably learned their lesson now .

Jeff

The funny thing is that they keep blaming me for messing things up on my server which I did not. I don't know what the hell they are doing but I guess they really need to brush up on their network stuff.