We just went to replace the bridge/firewall services one one server with the same on another. It's pretty simple, and I literally cloned (w/ rsync) a third server that does this onto the one that will be the new one. Then copied the /etc/sysconfig/iptables from the one being replaced, and brought it up this morning.
Nope. We had to put everything back the way it was.
The new one sees the two or three servers behind the firewall, and we can ping them, from the new box. On one, we see IPP broadcasts; in fact, we see lots of broadcast packets using tcpdump. From outside, though, you can't see the servers. Trying to ping them, they see nothing. It seems to be the case that tcp and icmp packets are blocked, and we can't figure out why.
CentOS 5.6.
ifcfg-eth0
DEVICE=eth0 BRIDGE=br3 BOOTPROTO=dhcp HWADDR=aa:bb:cc:dd:ee:ff ONBOOT=yes
ifcfg-eth1
DEVICE=eth1 BRIDGE=br3 HWADDR=aa:bb:cc:dd:ee:gg ONBOOT=yes
ifcfg-br3
DEVICE=br3 ONBOOT=yes TYPE=Bridge BOOTPROTO=static IPADDR=<our ip> NETMASK=255.255.254.0 NETWORK=<our nw> GATEWAY=<our gw>
Any ideas?
mark
I thought all we were going to is remove the IA_REMOTE Banner for the BYG-1 Display applications.
-----Original Message----- From: centos-bounces@centos.org [mailto:centos-bounces@centos.org] On Behalf Of m.roth@5-cent.us Sent: Monday, June 13, 2011 2:02 PM To: CentOS mailing list Subject: EXTERNAL: [CentOS] A bridge problem
We just went to replace the bridge/firewall services one one server with the same on another. It's pretty simple, and I literally cloned (w/ rsync) a third server that does this onto the one that will be the new one. Then copied the /etc/sysconfig/iptables from the one being replaced, and brought it up this morning.
Nope. We had to put everything back the way it was.
The new one sees the two or three servers behind the firewall, and we can ping them, from the new box. On one, we see IPP broadcasts; in fact, we see lots of broadcast packets using tcpdump. From outside, though, you can't see the servers. Trying to ping them, they see nothing. It seems to be the case that tcp and icmp packets are blocked, and we can't figure out why.
CentOS 5.6.
ifcfg-eth0
DEVICE=eth0 BRIDGE=br3 BOOTPROTO=dhcp HWADDR=aa:bb:cc:dd:ee:ff ONBOOT=yes
ifcfg-eth1
DEVICE=eth1 BRIDGE=br3 HWADDR=aa:bb:cc:dd:ee:gg ONBOOT=yes
ifcfg-br3
DEVICE=br3 ONBOOT=yes TYPE=Bridge BOOTPROTO=static IPADDR=<our ip> NETMASK=255.255.254.0 NETWORK=<our nw> GATEWAY=<our gw>
Any ideas?
mark
_______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
On Monday 13 June 2011 14:02, the following was written:
We just went to replace the bridge/firewall services one one server with the same on another. It's pretty simple, and I literally cloned (w/ rsync) a third server that does this onto the one that will be the new one. Then copied the /etc/sysconfig/iptables from the one being replaced, and brought it up this morning.
Nope. We had to put everything back the way it was.
The new one sees the two or three servers behind the firewall, and we can ping them, from the new box. On one, we see IPP broadcasts; in fact, we see lots of broadcast packets using tcpdump. From outside, though, you can't see the servers. Trying to ping them, they see nothing. It seems to be the case that tcp and icmp packets are blocked, and we can't figure out why.
Is the firewall IP or port based or a combo of both? Is the firewall setup on the bridge interface or on each individual server interface i.e., eth0, eth1 etc..
What does ifconfig show you? Are all the interfaces started? Do the DHCP interfaces receive a DHCP address?
Robert Spangler wrote:
On Monday 13 June 2011 14:02, the following was written:
We just went to replace the bridge/firewall services one one server with the same on another. It's pretty simple, and I literally cloned (w/ rsync) a third server that does this onto the one that will be the new one.Then copied the /etc/sysconfig/iptables from the one being replaced, and brought it up this morning.
Nope. We had to put everything back the way it was.
The new one sees the two or three servers behind the firewall, and we can ping them, from the new box. On one, we see IPP broadcasts; in fact, we see lots of broadcast packets using tcpdump. From outside, though, you can't see the servers. Trying to ping them, they see nothing. It seems to be the case that tcp and icmp packets are blocked, and we can't figure out why.
Is the firewall IP or port based or a combo of both? Is the firewall setup on the bridge interface or on each individual server interface i.e., eth0, eth1 etc..
Not sure how to answer that. I'd say it's on the external interface.
What does ifconfig show you? Are all the interfaces started? Do the DHCP interfaces receive a DHCP address?
Yep. And route shows *only* br3, and when I restart the network br3 brings up eth0 and eth1.
mark
On 6/13/2011 1:02 PM, m.roth@5-cent.us wrote:
We just went to replace the bridge/firewall services one one server with the same on another. It's pretty simple, and I literally cloned (w/ rsync) a third server that does this onto the one that will be the new one. Then copied the /etc/sysconfig/iptables from the one being replaced, and brought it up this morning.
Nope. We had to put everything back the way it was.
The new one sees the two or three servers behind the firewall, and we can ping them, from the new box. On one, we see IPP broadcasts; in fact, we see lots of broadcast packets using tcpdump. From outside, though, you can't see the servers. Trying to ping them, they see nothing. It seems to be the case that tcp and icmp packets are blocked, and we can't figure out why.
CentOS 5.6.
ifcfg-eth0
DEVICE=eth0 BRIDGE=br3 BOOTPROTO=dhcp HWADDR=aa:bb:cc:dd:ee:ff ONBOOT=yes
ifcfg-eth1
DEVICE=eth1 BRIDGE=br3 HWADDR=aa:bb:cc:dd:ee:gg ONBOOT=yes
ifcfg-br3
DEVICE=br3 ONBOOT=yes TYPE=Bridge BOOTPROTO=static IPADDR=<our ip> NETMASK=255.255.254.0 NETWORK=<our nw> GATEWAY=<our gw>
Any ideas?
Are the HWADDR= entries fixed up to match the actual hardware after the copy? And does ifconfig show that your config actually set up what you expected? CentOS isn't very predictable in terms of which NIC gets which interface name.
Les Mikesell wrote:
On 6/13/2011 1:02 PM, m.roth@5-cent.us wrote:
We just went to replace the bridge/firewall services one one server with the same on another. It's pretty simple, and I literally cloned (w/ rsync) a third server that does this onto the one that will be the new one.Then copied the /etc/sysconfig/iptables from the one being replaced, and brought it up this morning.
Nope. We had to put everything back the way it was.
The new one sees the two or three servers behind the firewall, and we can ping them, from the new box. On one, we see IPP broadcasts; in fact, we see lots of broadcast packets using tcpdump. From outside, though, you can't see the servers. Trying to ping them, they see nothing. It seems to be the case that tcp and icmp packets are blocked, and we can't figure out why.
<snip>
Are the HWADDR= entries fixed up to match the actual hardware after the copy? And does ifconfig show that your config actually set up what you expected? CentOS isn't very predictable in terms of which NIC gets which interface name.
Yes. And I made sure of that, before we started this excersize. (And my manager asked the same question - he's one of us, you see, *not* a PHB)
mark
mark
m.roth@5-cent.us wrote:
Les Mikesell wrote:
On 6/13/2011 1:02 PM, m.roth@5-cent.us wrote:
We just went to replace the bridge/firewall services one one server with the same on another. It's pretty simple, and I literally cloned (w/ rsync) a third server that does this onto the one that will be the new one.Then copied the /etc/sysconfig/iptables from the one being replaced, and brought it up this morning.
Nope. We had to put everything back the way it was.
The new one sees the two or three servers behind the firewall, and we can ping them, from the new box. On one, we see IPP broadcasts; in fact, we see lots of broadcast packets using tcpdump. From outside, though, you can't see the servers. Trying to ping them, they see nothing. It seems to be the case that tcp and icmp packets are blocked, and we can't figure out why.
<snip> > Are the HWADDR= entries fixed up to match the actual hardware after the > copy? And does ifconfig show that your config actually set up what you > expected? CentOS isn't very predictable in terms of which NIC gets > which interface name.
Yes. And I made sure of that, before we started this excersize. (And my manager asked the same question - he's one of us, you see, *not* a PHB)
mark
Without knowing more about your current server there is not much we can help you. I am fluent in networking, I am 7 years WISP and 4-5 years network/wireless consultant.
Are you using that new unit (router/gateway is what they are called, not servers, you will just confuse things) as a pass through bridge with added IP firewalling (only 2 interfaces)? Or are you supposed to route (one outgoing interface eth2 and br3 as local LAN)?
Why do you have bootproto=dhcp on eth0? Is NETMASK=255.255.254.0 supposed to be .254.0 or is it an typo?
Have you removed ARP entries from ARP cache of neighboring units (servers, upstream routers) etc?
Have you enabled ip_forwarding ?
If you have pass through bridge with only two interfaces, have you considered that maybe you should reverse/switch LAN cables plugged in eth0 and eth1 since firewall script is probably setup as one direction only, and if you reverse the flow firewall might block all. Test with firewall disabled/stopped.
Using combination of bridge and firewall is not wise at all, I would say it's quite a mess. It is always best to use routing.
Ljubomir
On Mon, Jun 13, 2011 at 19:30, Ljubomir Ljubojevic office@plnet.rs wrote:
m.roth@5-cent.us wrote:
Les Mikesell wrote:
On 6/13/2011 1:02 PM, m.roth@5-cent.us wrote:
We just went to replace the bridge/firewall services one one server with the same on another. It's pretty simple, and I literally cloned (w/ rsync) a third server that does this onto the one that will be the new one.Then copied the /etc/sysconfig/iptables from the one being replaced, and brought it up this morning.
Nope. We had to put everything back the way it was.
The new one sees the two or three servers behind the firewall, and we can ping them, from the new box. On one, we see IPP broadcasts; in fact, we see lots of broadcast packets using tcpdump. From outside, though, you can't see the servers. Trying to ping them, they see nothing. It seems to be the case that tcp and icmp packets are blocked, and we can't figure out why.
Maybe some router or switch has your old mac address forced (or kind of static). Can you reboot them? (or contact your ISP to know for sure?)
-- Marcelo
"¿No será acaso que ésta vida moderna está teniendo más de moderna que de vida?" (Mafalda)
On 6/13/2011 3:01 PM, m.roth@5-cent.us wrote:
Les Mikesell wrote:
On 6/13/2011 1:02 PM, m.roth@5-cent.us wrote:
We just went to replace the bridge/firewall services one one server with the same on another. It's pretty simple, and I literally cloned (w/ rsync) a third server that does this onto the one that will be the new one.Then copied the /etc/sysconfig/iptables from the one being replaced, and brought it up this morning.
Nope. We had to put everything back the way it was.
The new one sees the two or three servers behind the firewall, and we can ping them, from the new box. On one, we see IPP broadcasts; in fact, we see lots of broadcast packets using tcpdump. From outside, though, you can't see the servers. Trying to ping them, they see nothing. It seems to be the case that tcp and icmp packets are blocked, and we can't figure out why.
<snip> > Are the HWADDR= entries fixed up to match the actual hardware after the > copy? And does ifconfig show that your config actually set up what you > expected? CentOS isn't very predictable in terms of which NIC gets > which interface name.
Yes. And I made sure of that, before we started this excersize. (And my manager asked the same question - he's one of us, you see, *not* a PHB)
I missed that 'from outside' part before. If that means on the other side of a router, note that routers generally have a 20 minute arp cache so when you move the IP to a different MAC address you either have to wait a long time or log into the router and 'clear arp' before things will work again. There's probably a way to make the interface send a gratuitous arp that the router will catch, but I don't know it off the top of my head.
On 06/13/2011 11:02 AM, m.roth@5-cent.us wrote:
We just went to replace the bridge/firewall services one one server with the same on another. It's pretty simple, and I literally cloned (w/ rsync) a third server that does this onto the one that will be the new one. Then copied the /etc/sysconfig/iptables from the one being replaced, and brought it up this morning.
Specifically what did you rsync? If you copied the ifcfg files, you probably need to adjust the HWADDR in each. If you didn't get all of /etc, you might need sysctl.conf. I'm guessing that's the case, given the symptoms and the fact that you had to also copy the iptables file.
ifcfg-eth0
DEVICE=eth0 BRIDGE=br3 BOOTPROTO=dhcp HWADDR=aa:bb:cc:dd:ee:ff ONBOOT=yes
There should not be a BOOTPROTO in this file.
ifcfg-br3
DEVICE=br3 ONBOOT=yes TYPE=Bridge BOOTPROTO=static IPADDR=<our ip> NETMASK=255.255.254.0 NETWORK=<our nw> GATEWAY=<our gw>
You don't need NETWORK here.
It would also be helpful to see the contents of /etc/sysctl.conf or the output of:
# cat /proc/sys/net/ipv4/ip_forward # cat /proc/sys/net/bridge/bridge-nf-call-* # brctl show
-
Gordon Messmer -
Les Mikesell -
Ljubomir Ljubojevic -
m.roth＠5-cent.us -
Marcelo Roccasalva -
Massey, Ricky -
Robert Spangler