Hi, We have several different real internet connections as well as tunnels
used for certain purposes. Some have "capped" volume and others not; the
ones that do not have different data rates; and all have varying
reliability.
I'm having trouble understanding how the weight values work for the opti
Hi Norm,
On 1/9/19 11:24 PM, Norman Henderson wrote:
> Hi, We have several different real internet connections as well as
> tunnels used for certain purposes. Some have "capped" volume and others
> not; the ones that do not have different data rates; and all have
> varying reliability.
>
> I'm ha
Thank you Tom. Can you clarify the significance of the numbers:
e.g. provider a weight 1, provider b weight 10, provider c weight 100,
which provider is preferred and what will be the proportional allocation of
new flows?
- Norm
On Thu, Jan 10, 2019 at 5:53 PM Tom Eastep wrote:
> Hi Norm,
>
> On
On 1/10/19 9:18 AM, Norman Henderson wrote:
> Thank you Tom. Can you clarify the significance of the numbers:
> e.g. provider a weight 1, provider b weight 10, provider c weight 100,
> which provider is preferred and what will be the proportional allocation
> of new flows?
New flows will be apport
Thanks. Digging into it, I read that as of kernel 4.4, load balancing is
done per-flow based on a hash over the source and destination addresses:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=07355737a8badd951e6b72aa8609a2d6eed0a7e7
And: "Multipath routes make the s
On Fri, 11 Jan 2019 16:20:53 +0100
Norman Henderson wrote:
> Trouble is, it doesn't look like it works that way - but I need to
> watch it for a while. Anyway I guess it isn't really Shorewall doing
> this, rather iproute2.
No. It's kernel which handles routing.
--
Tuomo Soini
Foobar Linux se