[LARTC] failover routing

[LARTC] failover routing

[Jandre Olivier]

Hi Guys,

I would just like to have advice and pointers of the best way would be,
Someting like BGP or OSPF?

I have 2 internet connections at diffrent locations. let say connection
A and B

1.) router A has a fast internet connection and a seperate interface for
clients using /lan/pppoe/ipsec etc  and another ethernet interface going
to router B

2.) router B has similiar setup as router A and also a seperate ether
interface for clients and one going to router A

3.) all clients gets masqueraded as there is limited amount of internet
routable ips

Now my first thought was to write some perl/bash scripts to just ping
your internet gateway address of Router A and if its down, just change
your default route to router B and everyone and vice versa and u  can
still get access.
This way for me is not very clean though as Im the one writing the
scripts as something like zebra might do this perfectly?
just a basic idea of what my setup is. What would be my best way of
doing this.?

-- 
/*---------------------------------------------------------------------*/
                         __   _
             ---------- / /  (_)__  __ ____  __  ---------
               ------- / /__/ / _ \/ // /\ \/ / --------
                 ---- /____/_/_//_/\_,_/ /_/\_\ ------
                    localhost@localdomain.za.net
_______________________________________________
LARTC mailing list
LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc

RE: [LARTC] failover routing

[Greg Scott]

Your script could have the backup router take on the IP Address of the
primary after it loses its heartbeat.  You'll run into a problem with
ARP caches.  I saw some code floating around earlier that allowed one
box to listen to the MAC address of another and respond to its ARP
requests.  You would need to incorporate something like this in any
solution.

And this all assumes routers A and B are in parallel; all clients and
both routers are on the same LAN.  So you have a separate NIC between
routers A and B for heartbeat.  Each router has a NIC on the LAN side,
and each has a NIC connecting to the Internet.

- Greg Scott


-----Original Message-----
From: lartc-bounces@mailman.ds9a.nl
[mailto:lartc-bounces@mailman.ds9a.nl] On Behalf Of Jandre Olivier
Sent: Wednesday, February 01, 2006 5:52 AM
To: lartc@mailman.ds9a.nl
Subject: [LARTC] failover routing

Hi Guys,

I would just like to have advice and pointers of the best way would be,
Someting like BGP or OSPF?

I have 2 internet connections at diffrent locations. let say connection
A and B

1.) router A has a fast internet connection and a seperate interface for
clients using /lan/pppoe/ipsec etc  and another ethernet interface going
to router B

2.) router B has similiar setup as router A and also a seperate ether
interface for clients and one going to router A

3.) all clients gets masqueraded as there is limited amount of internet
routable ips

Now my first thought was to write some perl/bash scripts to just ping
your internet gateway address of Router A and if its down, just change
your default route to router B and everyone and vice versa and u  can
still get access.
This way for me is not very clean though as Im the one writing the
scripts as something like zebra might do this perfectly?
just a basic idea of what my setup is. What would be my best way of
doing this.?

--
/*---------------------------------------------------------------------*
/
                         __   _
             ---------- / /  (_)__  __ ____  __  ---------
               ------- / /__/ / _ \/ // /\ \/ / --------
                 ---- /____/_/_//_/\_,_/ /_/\_\ ------
                    localhost@localdomain.za.net
_______________________________________________
LARTC mailing list
LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc
_______________________________________________
LARTC mailing list
LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc

RE: [LARTC] failover routing

[Sebastian Heidl]

On Wed, 2006-02-01 at 08:40 -0600, Greg Scott wrote:
> Your script could have the backup router take on the IP Address of the
> primary after it loses its heartbeat.  You'll run into a problem with
> ARP caches.  I saw some code floating around earlier that allowed one
> box to listen to the MAC address of another and respond to its ARP
> requests.  You would need to incorporate something like this in any
> solution.

Heartbeat (http://www.linux-ha.org/) does a gratuitous ARP (sends an
ARP-reply broadcast) when it takes over the ip of the other host so the
clients _should_ know the new MAC address belonging to the router.

_sh_

> And this all assumes routers A and B are in parallel; all clients and
> both routers are on the same LAN.  So you have a separate NIC between
> routers A and B for heartbeat.  Each router has a NIC on the LAN side,
> and each has a NIC connecting to the Internet.
> 
> - Greg Scott
> 
> 
> -----Original Message-----
> From: lartc-bounces@mailman.ds9a.nl
> [mailto:lartc-bounces@mailman.ds9a.nl] On Behalf Of Jandre Olivier
> Sent: Wednesday, February 01, 2006 5:52 AM
> To: lartc@mailman.ds9a.nl
> Subject: [LARTC] failover routing
> 
> Hi Guys,
> 
> I would just like to have advice and pointers of the best way would be,
> Someting like BGP or OSPF?
> 
> I have 2 internet connections at diffrent locations. let say connection
> A and B
> 
> 1.) router A has a fast internet connection and a seperate interface for
> clients using /lan/pppoe/ipsec etc  and another ethernet interface going
> to router B
> 
> 2.) router B has similiar setup as router A and also a seperate ether
> interface for clients and one going to router A
> 
> 3.) all clients gets masqueraded as there is limited amount of internet
> routable ips
> 
> Now my first thought was to write some perl/bash scripts to just ping
> your internet gateway address of Router A and if its down, just change
> your default route to router B and everyone and vice versa and u  can
> still get access.
> This way for me is not very clean though as Im the one writing the
> scripts as something like zebra might do this perfectly?
> just a basic idea of what my setup is. What would be my best way of
> doing this.?
> 
> --
> /*---------------------------------------------------------------------*
> /
>                          __   _
>              ---------- / /  (_)__  __ ____  __  ---------
>                ------- / /__/ / _ \/ // /\ \/ / --------
>                  ---- /____/_/_//_/\_,_/ /_/\_\ ------
>                     localhost@localdomain.za.net
> _______________________________________________
> LARTC mailing list
> LARTC@mailman.ds9a.nl
> http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc
> _______________________________________________
> LARTC mailing list
> LARTC@mailman.ds9a.nl
> http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc
_______________________________________________
LARTC mailing list
LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc

RE: [LARTC] failover routing

[comp.techs]

[-- Attachment #1: Type: text/plain, Size: 206 bytes --]

--===============1453543749==
Content-class: urn:content-classes:message
Content-Type: multipart/alternative;
	boundary="----_=_NextPart_001_01C62C02.5E22B54C"

This is a multi-part message in MIME format.

[-- Attachment #2: Type: text/plain, Size: 572 bytes --]

Hi, our network has a hand full of 2k servers that use silent rip. We use 2 Linux gateways with separate isp's.
Each gateway does a ' default-originate'  to advertise its default route in ripv2 (with on having a different metric).
The main gateway has a ping script written in shell that will ping the gateway, determine if its up or down; With either result it checks its history of the last 3 intervals, and will only stop rip if or start rip if there is/was 3 concurrent ups or downs.
I have been running this for 2 years with no problems.
 
I hope this helps
 

[-- Attachment #3: Type: text/html, Size: 1162 bytes --]

Re: [LARTC] failover routing

[the sew]

[-- Attachment #1.1: Type: text/plain, Size: 1036 bytes --]

sounds good, do your run ripv2 with zebra or quagga?, your idea sounds
exactly what I have in mind. Mind explaining a bit more technical and
sharing some usefull bits in your config?

Thanks

Sew




On 2/7/06, comp.techs <comp.techs@aspenview.org> wrote:
>
>  Hi, our network has a hand full of 2k servers that use silent rip. We use
> 2 Linux gateways with separate isp's.
> Each gateway does a ' default-originate'  to advertise its default route
> in ripv2 (with on having a different metric).
> The main gateway has a ping script written in shell that will ping the
> gateway, determine if its up or down; With either result it checks its
> history of the last 3 intervals, and will only stop rip if or start rip if
> there is/was 3 concurrent ups or downs.
> I have been running this for 2 years with no problems.
>
> I hope this helps
>
>
> _______________________________________________
> LARTC mailing list
> LARTC@mailman.ds9a.nl
> http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc
>
>
>

[-- Attachment #1.2: Type: text/html, Size: 1966 bytes --]

[-- Attachment #2: Type: text/plain, Size: 143 bytes --]

_______________________________________________
LARTC mailing list
LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc

RE: [LARTC] failover routing

[comp.techs]

[-- Attachment #1: Type: text/plain, Size: 206 bytes --]

--===============1855685569==
Content-class: urn:content-classes:message
Content-Type: multipart/alternative;
	boundary="----_=_NextPart_001_01C62C2F.D91EE144"

This is a multi-part message in MIME format.

[-- Attachment #2: Type: text/plain, Size: 2101 bytes --]

Hi,
 
yes we run ripv2 and both quagga and zebra.
 
The script on the primary gateway just pings the nexthop. The return value is formated.
and stored in a log file. 
Next the script checks the last three values in the log, if current value 100 (down) or 0 (up) matches the last
three values in the log file determines the outcome of either stopping rip or starting it.
 
For example: logfile values 100 100 100  current value 100 --this would stop ripd
In order for ripd to start  we would need : 0 0 0 in the logfile and a current value 0, this would then start ripd
 
 
! this is what is in my secondary gateway ripd.conf
router rip
 default-information originate
 offset-list 1 out 4 eth0
 network eth0

access-list 1 permit 0.0.0.0


 


!this is what is in my primary gateway ripd.conf
router rip
 default-information originate
  network eth0

 
Jason
________________________________

From: the sew [mailto:sewlist@gmail.com]
Sent: Tue 2/7/2006 1:58 PM
To: comp.techs
Cc: lartc@mailman.ds9a.nl
Subject: Re: [LARTC] failover routing


sounds good, do your run ripv2 with zebra or quagga?, your idea sounds exactly what I have in mind. Mind explaining a bit more technical and sharing some usefull bits in your config?

Thanks

Sew





On 2/7/06, comp.techs <comp.techs@aspenview.org> wrote: 

	
	Hi, our network has a hand full of 2k servers that use silent rip. We use 2 Linux gateways with separate isp's.
	Each gateway does a ' default-originate'  to advertise its default route in ripv2 (with on having a different metric).
	The main gateway has a ping script written in shell that will ping the gateway, determine if its up or down; With either result it checks its history of the last 3 intervals, and will only stop rip if or start rip if there is/was 3 concurrent ups or downs.
	I have been running this for 2 years with no problems.
	 
	I hope this helps
	 

	_______________________________________________
	LARTC mailing list
	LARTC@mailman.ds9a.nl
	http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc
	
	
	



[-- Attachment #3: Type: text/html, Size: 4556 bytes --]