Thanks Tom. I’m able to route through different pools. The issue is that the kumo-proxy seems to only send traffic out the default LAN IP so no matter what I specify for socks5_proxy_source_address it ignores that value and just uses my primary lan ip
so for example, I have a LAN IP of x.x.x.101 and x.x.x.102. Each map to a different WAN IP. But no matter what value I specify for socks5_proxy_source_address kumo-proxy seems to only send traffic through my x.x.x.101 LAN IP and therefore I can’t proxy multiple ips out the WAN
Can you show a couple of delivered log lines?
sure thing…
{
"type":"Delivery",
"id":"78102fd9e24a11ee91f40672a40910a3",
"sender":"XXXXX",
"recipient":"XXXXX",
"queue":"default-tenant@XXXXX",
"site":"XXXXX",
"size":909,
"response":{
"code":250,
"enhanced_code":null,
"content":"Mail accepted for delivery",
"command":".\r\n"
},
"peer_address":{
"name":"XXXXX",
"addr":"XXXXX"
},
"timestamp":1710451996,
"created":1710451996,
"num_attempts":0,
"bounce_classification":"Uncategorized",
"egress_pool":"pool-1",
"egress_source":"ip-2",
"feedback_report":null,
"meta":{
},
"headers":{
},
"delivery_protocol":"ESMTP",
"reception_protocol":"ESMTP",
"nodeid":"c3047569-3ede-40ae-b4e6-774ddc6cc407"
}
That example shows it’s using the ip-2 which has a source address that should map to 102 lan address - but the proxy still routes through the primary 101 lan address.
Ok I’m going to do some digging.
FWIW it seems to have the same issue with curl:
curl -x socks5://x.x.x.102:5000 ifconfig.me
# Returns the wan ip addressed that's mapped to .101 LAN IP, whereas:
curl --interface x.x.x.102 ifconfig.me
# returns the correct wan ip
Hmm. Can you try shutting down our proxy and test a different SOCS5 proxy? tsocks or similar? Just to eliminate it as a potential source of trouble.
yeah let me try tsocks
I’ve tried a couple other socks5 proxies - they bind to the correct IP but they all bind on port 0 instead of 25 so they don’t make it out the door:
BIND x.x.x.102:0
well maybe that’s not the issue but it doesn’t deliver through the other socks servers - they get hung up
Are you on GCP by chance?
aws - proxying to a datacenter
ok
Darn - that would have been an easy fix ![]()
These log lines show the two deliveries I made using that one pool to the Socks Proxy. I think the only difference is that I only have on IP address plumbed. I’ll try to add another and see if it makes a difference, but you can see here that it does alternate between the IPs in the pool.
message.txt (2.81 KB)
yeah it alternates for me just fine as well. The problem is that the kumo-proxy seems to ignore the socks5_proxy_source_address and only bind to the primary source address
Ok I tested using the same configuration but with haproxy and it works perfectly. So:
When using kumo-proxy - this doesn’t bind to the correct source address and the traffic is sent via the primary source instead of the defined .102 source
[source."ip-2"]
socks5_proxy_server = 'x.x.x.x:5000'
socks5_proxy_source_address = 'x.x.x.102'
But this config (using haproxy) works exactly as expected:
[source."ip-2"]
ha_proxy_server = 'x.x.x.x:5000'
ha_proxy_source_address = 'x.x.x.102'
So it seems like something is wrong with kumo-proxy since the exact same haproxy-version of the config works as expected.
Thank you. Are you also able to grep the journal for the proxy info?
journalctl -f -n 50 -u kumomta |grep socks5