LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

RE: ip_masq_ftp nat passive

To: 'Julian Anastasov' <ja@xxxxxx>
Subject: RE: ip_masq_ftp nat passive
Cc: "'lvs-users@xxxxxxxxxxxxxxxxxxxxxx'" <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
From: Jeremy Kusnetz <JKusnetz@xxxxxxxx>
Date: Thu, 24 May 2001 12:44:57 -0400
The 10.75.32.17 is just another RIP, I have 3 of them going right now, the
results are the same no matter which RIP it hits.

I'll try using a different FTP server to see if that makes any difference.

IP masq. is checked in the kernel, is it possible that I don't have all the
correct configs in the kernel set?  The configs run on the LVS box, and
ip_masq_ftp runs on the LVS box, it doesn't matter how the realservers are
set up, correct?  From what I understand, the realservers don't even need to
be linux boxes for LVS-NAT, correct?

-----Original Message-----
From: Julian Anastasov [mailto:ja@xxxxxx]
Sent: Thursday, May 24, 2001 12:39 PM
To: Jeremy Kusnetz
Cc: 'lvs-users@xxxxxxxxxxxxxxxxxxxxxx'
Subject: RE: ip_masq_ftp nat passive



        Hello,

On Thu, 24 May 2001, Jeremy Kusnetz wrote:

> ftp> passive
> Passive mode on.
> ftp> ls
> ---> PASV
> 227 Passive mode on (10,75,32,17,18,31)
> ftp: connect: No route to host
> ftp>

        At first look it seems you are using FTP server that is not
supported from Linux 2.2 masquerade. The kernel waits for

227 Entering Passive Mode (xxx,xxx,xxx,xxx,ppp,ppp)

        I assume the check in the kernel is very restrictive and can't
detect the real server reply. Is that possible? Change the kernel and/or
the FTP server :)

        And why 10.75.32.17 is in the game while I don't see it in the
tcpdump output?

Regards

--
Julian Anastasov <ja@xxxxxx>


<Prev in Thread] Current Thread [Next in Thread>