I figured out a temporary workaround using ‘ip netns exec’
It works, but things like apticron will still fail.
--
Swimmingly,
Joe
swimgeek.com/blog +27 82 562 6167 instagram.com/joe.swimgeek
"...all progress depends on the unreasonable man.”
On 16 Feb 2022, at 11:24, Joe Botha <joe@swimgeek.com> wrote:
Hi
Yes, sorry, should maybe have had title with Source _IP_address_
I think you should be able to specify a source IP when opening the tcp socket.
Would probably need to add some code for IPv4 and IPv6 cases.
--
Swimmingly,
Joe
swimgeek.com/blog +27 82 562 6167 instagram.com/joe.swimgeek
"...all progress depends on the unreasonable man.”On 16 Feb 2022, at 11:11, Julian Andres Klode <jak@debian.org> wrote:
On Wed, Feb 16, 2022 at 10:42:57AM +0200, Joe Botha wrote:Package: apt
Version: 2.2.4
Severity: wishlist
Tags: ipv6
Dear Maintainer,
*** Reporter, please consider answering these questions, where appropriate ***
I use Debian on a server with multiple upstream internet links. Some of these links have peering point IPs.
It happens fairly often that apt-get traffic uses the source IP in a peering point range and then downloads fail.
Please consider adding a feature so I can configure apt-get to specify a source IP or interface on the server.
For example: traceroute has a -s option: "Chooses an alternative source address."
I was about to say that specifying ip addresses for sources needs to be
solved at an RFC level (need to standardize URLs like http://foo[ip]/),
but um now I understand that you don't want to configure the IP of the
source, but the interface we send packets from.
Probably should be retitled with a less ambiguous title :D
Anyway, no idea how complex that feature is, I guess proof of
concept patches welcome.
--
debian developer - deb.li/jak | jak-linux.org - free software dev
ubuntu core developer i speak de, en