LinuxQuestions.org
Visit the LQ Articles and Editorials section
Go Back   LinuxQuestions.org > Forums > Linux Forums > Linux - General
User Name
Password
Linux - General This Linux forum is for general Linux questions and discussion.
If it is Linux Related and doesn't seem to fit in any other forum then this is the place.

Notices



Reply
 
Search this Thread
Old 10-29-2003, 03:27 PM   #1
zxcv
LQ Newbie
 
Registered: Jul 2003
Posts: 2

Rep: Reputation: 0
Virtual IP weirdness


Hi, I am having some really strange weirdness with a virtual IP. I have a RHAS box with iptables on it, eth0 being external interface bound on .100, eth1 is internal bound to .101, and a virtual IP on eth0 which is .102. I am running DNS/BIND on it and it's bound on all 3 IPs (waiting for my utility machine to show up, hence why we're using .102 so that clients don't have to change their registrar information). Everything worked fine for months, until yesterday, and there have been no known changes.

From the box itself, dns answers on .100, .101, and .102 no problem. Internally, .101 answers fine. Externally, only .100 answers, and .102 gives no answer. I set iptables to log queries, and nothing is going to .102 (but .101 and .100 are fine). I turned off iptables just to verify, and as far as I have been able to determine, it's not any firewall problem. I added a new virtual IP, .123, and it was having the same problem as .102. I even cannibalized a working IP from another machine to test and make sure it wasn't a particular IP. And the gateway router routes traffic fine.

It appears as if the virtual IP only works if you're looking at it from the host machine itself. There aren't any errors. Unfortunately, I can't really take the firewall down to test traffic easily lest I get pissed-off clients. A quick test with tcpdump didn't show any traffic (tcp or udp) for .102 when dns was requested. The machine has been rebooted, still no-go.

Might it be the kernel?

Thanks for any help!
 
Old 10-31-2003, 03:53 PM   #2
clacour
Member
 
Registered: Sep 2003
Location: Dallas, Tx, USA
Distribution: Red Hat, Gentoo, Libranet
Posts: 98

Rep: Reputation: 16
I don't know that I'll be much help, but since nobody else tried to tackle it...


First thing that occurs to me (since you've rebooted the box to no effect) is to try to pin down whether this is a bind issue or a network issue.

Can you ping 102 from outside? If so, how about things like telnet, ssh, http?

If you can't even ping it, it's likely a network issue - a switch that's had a configuration change (or even a port gone bad), something like that.

Set up a virtual IP on 101. If that works, it's that much less likely that it's a problem with Linux's networking.

If it DOESN'T work, that's a pretty strong indicator there IS a problem with Linux's networking. If you have a vendor and support contract for RHAS, I'd use it at that point.

If the internal virtual address works, and especially if the external ping works, I'm inclined to say it's a problem with BIND.

I'll wait for feedback from you before I say any more.


P.S. What version of bind are you running?
 
Old 10-31-2003, 07:44 PM   #3
zxcv
LQ Newbie
 
Registered: Jul 2003
Posts: 2

Original Poster
Rep: Reputation: 0
I found something really interesting this morning, a feature with the linux 2.2-2.4 kernels. Basically, if you have more than one NIC on the same physical network, such as eth0 and eth1 on the same switch, then eth1 can answer arp for eth0 and vice versa. There is a patch for it, which I haven't done yet, since the overall plan was to move eth1 on a separate physical network than eth0 (but I couldn't due to not having equipment at the time).

http://www.linuxvirtualserver.org/Jo...p_problem.html

(go down to section 4.5.4)

This is what I was seeing on tcpdump... eth1 was answering arp to eth0 and the router would be confused and use the wrong MAC address.

I believe the reason I hadn't seen this behavior before was that I didn't have much traffic on that network at the time. We moved a few webservers with high traffic to that network (changing datacenters) and suddenly this behavior shows up.

If you'd like, I'll post my results after I'm able to co-ordinate with the router guys some static routes...
 
Old 11-01-2003, 01:02 AM   #4
clacour
Member
 
Registered: Sep 2003
Location: Dallas, Tx, USA
Distribution: Red Hat, Gentoo, Libranet
Posts: 98

Rep: Reputation: 16
Kudos to you for wading through that stuff...

You don't need to post it for my sake (although I will definitely remember that this situation is possible -- we use multiple NICs in most of our machines, and they all go pretty much to the same switch.)

If you think it will help out somebody else, please do post it. I hate Googling for a problem and finding nothing but somebody else 6 months ago who had the same problem (and no answer!).

Good luck,

CHL
 
  


Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
PPPoE weirdness matrixcubed Linux - Networking 1 10-07-2005 11:41 PM
weirdness with log2() spuzzzzzzz Programming 5 08-08-2004 06:45 PM
Synaptic Weirdness chris319 Linux - Software 2 09-06-2003 12:43 AM
network weirdness rickenbacherus Linux - Networking 3 04-02-2003 08:20 PM
DNS weirdness snowdog12 Linux - Newbie 5 02-13-2003 08:45 PM


All times are GMT -5. The time now is 07:38 AM.

Main Menu
Advertisement
My LQ
Write for LQ
LinuxQuestions.org is looking for people interested in writing Editorials, Articles, Reviews, and more. If you'd like to contribute content, let us know.
Main Menu
Syndicate
RSS1  Latest Threads
RSS1  LQ News
Twitter: @linuxquestions
identi.ca: @linuxquestions
Facebook: linuxquestions Google+: linuxquestions
Open Source Consulting | Domain Registration