the proxy is the same machine, just 2 different outbound ips
the second question is socks5_proxy_source_address required when using socks? for our current configuration, if the proxy chose its default interface, it would actually choose the IP correctly no matter which address family
Ohhhh. I think I see what you are asking now. Unless something has changed presently, KumoProxy will not automatically tunnel from IPv4 to IPv6.
The way you have it configured, you have the server on Ip 4 and the source on IPv6.
I would not mix the two. I can look into it deeper, but my initial thought is to have 2 separate proxies, one for v4 and one for v6.
i see
so you’re saying instead of
|SMTP SERVER| - V4 → |SMTP OUTBOUND PROXY| → V4,V6
it should be
|SMTP SERVER| - V4 → |SMTP OUTBOUND PROXY| → V4
|SMTP SERVER| - V6 → |SMTP OUTBOUND PROXY| → V6
honestly the right way is just to outbound using the smtp server - i have byoipv6 setup on azure and my network announced over there except my current smtp network is different from the one announced in azure so i’d have to go through a warm-up and i don’t really want to do that yet
especially since there’s a warm-up feature on the kumo backlog 
tom - thank you for the help; i tried the two separate proxies, one for v4 and one for v6 and it worked a treat. is it possibly my issue was using the same proxy instance for both uses instead of using a different proxy instance on a different port?
@faithful-ostrich after running more tests - i thought it was all good but it’s not; it was just happenstance that both tests went off without a hitch
still getting the transient failures when it sends v6 traffic to the v4 proxy and v4 traffic to the v6 proxy
this is is the error that i log, it’s categorized as a TransientFailure. KumoMTA internal: failed to connect to any candidate hosts: connect to 52.xx.xx.xx:25 and read initial banner: Skipping ip-2 because socks5_proxy_source_address 2602:xx:xx:xx::118 address family does not match the destination address family 52.xx.xx.xx:25, connect to 52.36.144.28:25 and read initial banner: Skipping ip-2 because socks5_proxy_source_address 2602:xx:xx:xx::118 address family does not match the destination address family 52.36.144.28:25
Can you please share the exact full error you are receiving?
ya
let me get my sources.toml too rq
[source."ip-1"]
ehlo_domain='smtp-out01.domain.com'
socks5_proxy_server = '172.xx.xx.xx:2525'
socks5_proxy_source_address = '199.xx.xx.xx'
[source."ip-2"]
ehlo_domain='smtp-out01.domain.com'
socks5_proxy_server = "[2602:xx:xx:xx::118]:3535"
socks5_proxy_source_address = "2602:xx:xx:xx::118"
[pool."pool-1"]
[pool."pool-1"."ip-1"]
[pool."pool-1"."ip-2"]
im kind of looking over kumod\src\egress_source.rs and it looks like there’s a good bit of logging around this area would temporarily turning the log level to tracing while i do a few tests help to pinpoint what the deal is?
it just feels like some kind of quirk with the round-robin nature of the ip sources, i have the trace log that i think shows the hotspot but it will take me a little while to anonymize it
@faithful-ostrich you can see the inject happen and then on line 11 it’s decided on ip-2 for that domain and the error from ready_queue happens on line 26
And did you share the full configs? The sources file only tells part of the story.