Hello all!

I`m finding docs/examples about this subject. After some googling,
I only found basic scripts. For me, these simple scripts available
only works with excellent links (dedicated ones).

The load balancing part of the problem already was solved with
LARTC howto. DGD patch for linux was not a good experience
because it only detects the first hop fail and 90% of these fails
happens after the first hop (e.g. local router as gateway
connected to remote ISP).

I searched for some linux distro with this feature (good failover)
bult-in, but didnt found anything interesting.

I would appreciate pointers to docs, distros (for use as starting
point), examples or even academic research about the topic.


Thank you,
Tom Lobato

On 3 jul, 09:11, Tom <tomlob...@gmail.com> wrote:
> Hello all!
>
> I`m finding docs/examples about this subject. After some googling,
> I only found basic scripts. For me, these simple scripts available
> only works with excellent links (dedicated ones).
>
> The load balancing part of the problem already was solved with
> LARTC howto. DGD patch for linux was not a good experience
> because it only detects the first hop fail and 90% of these fails
> happens after the first hop (e.g. local router as gateway
> connected to remote ISP).
>
> I searched for some linux distro with this feature (good failover)
> bult-in, but didnt found anything interesting.
>
> I would appreciate pointers to docs, distros (for use as starting
> point), examples or even academic research about the topic.
>
> Thank you,
> Tom Lobato

From: Dana Harding <dharding@nucleus.com>:

Would Heartbeat work for your application? http://www.linux-ha.org/Heartbeat

You should be able to configure that to do some scripting magic (read:
bring up or reconfigure a secondary link) when a ping fails over the
primary link. I haven't used it myself - so I can't comment on how
well it works.

Dana Harding
dharding@nucleus.com

From: Dana Harding <dhard...@nucleus.com>:
>
> Would Heartbeat work for your application?http://www.linux-ha.org/Heartbeat
>
> You should be able to configure that to do some scripting magic (read:
> bring up or reconfigure a secondary link) when a ping fails over the
> primary link. I haven't used it myself - so I can't comment on how
> well it works.
>
> Dana Harding
> dhard...@nucleus.com

Hi Dana!

Well, I haven't used heartbeat, but I think It is not designed for
failover of ISP links, as it works testing servers on LAN (I think),so
it expects very low latencies comparing to ISPs.

If someone knows if heartbeat know how to work with ISP failover,
please tell me. Anyway I'll look its doc.


Tom Lobato

On Fri, 04 Jul 2008 07:00:57 -0700, Tom wrote:


> If someone knows if heartbeat know how to work with ISP failover, please
> tell me. Anyway I'll look its doc.

I'm not sure what you're trying to achieve. If you've two separate
upstreams, and want to route around failure of one, then the proper
solution is to admit (or "brag" {8^) that you're multihomed and step into
the BGP world. While it's missing a few features as compared to Cisco
IOS, Quagga is a nice BGP routing daemon. I've it in use on a couple of
boxes right now (neither of which gets heavy use yet, admittedly), and a
full feed of routes from a few different upstreams isn't a problem.

There's "paperwork" involved. You'll need to get a block of addresses
from one of your ISPs (unless you're larger than I suspect) that it is
willing to have announced via the other ISP. You'll have to register
this block with the appropriate route registry. But none of this is a
big deal, and one or the other of your ISPs can likely help you.

- Andrew

Hi Andrew!
Sorry for the delay.


On 6 jul, 10:53, Andrew Gideon <c172driv...@gideon.org> wrote:
> On Fri, 04 Jul 2008 07:00:57 -0700, Tom wrote:
> > If someone knows if heartbeat know how to work with ISP failover, please
> > tell me. Anyway I'll look its doc.
>
> I'm not sure what you're trying to achieve.


I have two ISP to access the internet, and want to load balancing and
link
failover on a Linux server.


> If you've two separate
> upstreams, and want to route around failure of one, then the proper
> solution is to admit (or "brag" {8^) that you're multihomed and step into
> the BGP world. While it's missing a few features as compared to Cisco
> IOS, Quagga is a nice BGP routing daemon. I've it in use on a couple of
> boxes right now (neither of which gets heavy use yet, admittedly), and a
> full feed of routes from a few different upstreams isn't a problem.
>
> There's "paperwork" involved. You'll need to get a block of addresses
> from one of your ISPs (unless you're larger than I suspect) that it is
> willing to have announced via the other ISP. You'll have to register
> this block with the appropriate route registry. But none of this is a
> big deal, and one or the other of your ISPs can likely help you.
>
> - Andrew


Well, I`ve read about BGP, but it appears this protocol needs ISP
support
(I don`t know exactly which support).
I`m not sure if they can provide me.
One of them gave me 4 static IP address, another gave me one dynamic
IP
with DHCP, both via cable modem.

Is this enough to implement BGP or the ISP has to give me explicit BGP
support?



Tom Lobato

On 12 jul, 11:11, Tom <tomlob...@gmail.com> wrote:
>
> On 6 jul, 10:53, Andrew Gideon <c172driv...@gideon.org> wrote:
> >
> > I'm not sure what you're trying to achieve.
>
> I have two ISP to access the internet, and want to load balancing and
> link failover on a Linux server.

complementing...
load balancing is not a problem, linux is great!
failover is giving me a hard time :)
ping showed to be a hard way to achieve reliable ISP link failover

I`m playing with ping + qos + icmp priority base on iproute
but not found a reliable final solution.


Tom Lobato

On Sat, 12 Jul 2008 07:19:46 -0700, Tom wrote:

> load balancing is not a problem

W/o BGP, how are you making the determination of which egress path should
be taken by a given packet?

- Andrew