Proxmox Mail Gatway as a Relay in Stalwart

[Obsidiannite21]

I'm trying to configure Proxmox Mail Gateway (PMG) as a relay in Stalwart. Previously, I used SMTP2GO, so all I had to do was set a new relay host to my PMG IP on port 26. However, when I send an email, I immediately get the error:

"connection to '192.168.x.x' failed: record not found for MX"

I suspected an issue with the web GUI, so I manually added the following to the config file:

[queue.strategy]
route = [
{ if = "is_local_domain('*', rcpt_domain)", then = "'local'" },
{ else = "'proxmox_mg'" }
]

[queue.route.proxmox_mg]
type = "smtp"
host = "192.168.x.x"
port = 26
This did not resolve the issue. PMG logs show no connection attempts from Stalwart, so I considered it might be a network problem. However, running

openssl s_client -crlf -starttls smtp -connect 192.168.x.x:26
from the Stalwart container successfully connects, and I see the connection in the PMG logs. This confirms it's not a networking issue.

I read that: Stalwart’s [queue.route.proxmox_mg] relay config may need a hostname/FQDN (like proxmox.local or pmg.internal.lan) instead of a bare IP address. Some mail servers, if given an IP, will try to perform deliver-as-MX instead of just connecting directly. Is this true? Using a FQDN isn't an option presently.

[dpurnam]

Definitely upvote this enhancement request

[leadenchopsticks]

Definitely upvote this enhancement request

So, this bug is already there for more than a year? This renders the relay functionality complete useless, or am I missing something?

[Obsidiannite21]

Thank you, I just did. At least now I know its a known issue and not a configuration error on my part.

[nomadturk]

@Obsidiannite21

A roundabout way of having an FQDN would be adding an entry to your /etc/hosts file like:

192.168.x.x mx.proxmox.local
then adding mx.proxmox.local as the relay route.

[Obsidiannite21]

Thank you for the suggestion. I tried that but it's still looking for an mx record. A work around may be to set something up in my pihole config to simulate a local mx record. I haven't tried it yet though.

[leadenchopsticks]

I think in your config, type = "smtp" should be type = "relay"

Although I have that in my config, I am currently running into the same issue: have configured an IP as relay host but stalwart keeps trying to do MX lookups.

[David-Development]

Which version of Stalwart are you using? I've encountered the same issue in v0.13 where Stalwart seems to ignore all SMTP Relay Settings -> #2265

[Obsidiannite21]

I am using version 0.11.8

[leadenchopsticks]

Which version of Stalwart are you using? I've encountered the same issue in v0.13 where Stalwart seems to ignore all SMTP Relay Settings -> #2265

Using 0.13.3

[pepa65]

I am noticing some of your key names (type, host) are off in the [queue.route] section. After I set these records in [queue.route.RELAYNAME] it works:

• address (the domain name, localhost in your case)
• port (you have this)
• protocol ("smtp", you have this key name as type)
• type ("relay", not "smtp" like you had)

[Obsidiannite21]

I'll give that a try! Thanks for the suggestion.

[Obsidiannite21]

I added this to the end of the config.toml file:

[queue.strategy]
route = [
{ if = "is_local_domain('*', rcpt_domain)", then = "'local'" },
{ else = "'proxmox_mg'" }
]

[queue.route.proxmox_mg]
address = "192.168.x.x"
port = "26"
protocol = "smtp"
type = "relay"

I rebooted Stalwart, and sent an email, but got the same error. So I checked the WebUI and it still shows the old config for the Relay Host proxmox_mg. It's as if its ignoring the config file, so I removed the relay and now the log shows:

2025-10-13T16:27:49Z WARN Remote host ID not found (smtp.remote-id-not-found) queueId = 268630217004109824, from = "myemail", to = ["[email protected]"], size = 3094, total = 1, id = "proxmox_mg"

AI suggested I add, the following but it didn't help.

[smtp.client.remote.proxmox_mg]
hosts = ["192.168.x.x"]
port = 26
tls = "none"
helo-name = "mail.mydomain.com"
timeout = "30s"

[pepa65]

The stuff that the AI suggested is rubbish. Can't you remove the related config from config.toml and set everything in the webadmin?

• Settings>SMTP>Outbound>Strategies: Routing local else "proxmox_mg"
• Settings>SMTP>Outbound>Routing: RelayHost "proxmox_mg", local, mx
  Fill in Server Details and Authentication here for proxmox_mg

[Obsidiannite21]

No, if I configure the Relay Host in the Web Admin using the IP of the Proxmox Mail Gateway, I immediately get

Your message could not be delivered to the following recipients:
connection to 'ip of pmg' failed: record not found for MX

[David-Development]

I think you need a DNS record instead of the IP. I've used AdGuard Home and you wrote earlier that you have pihole so maybe it works there

[Obsidiannite21]

Yes, I plan to give that a try. I just haven't had time. I don't love the idea that if my pihole goes down, I can no longer send emails, especially considering some outages are reported by email.

[David-Development]

I think you could also try to add it to the /etc/hosts file.. But I just double checked my setup and I had to use a DNS (IP didn't work) - same error you got. - There is also an discussion about this: #2210 and an open issue: #838

[David-Development]

@pepa65 You said

set everything in the webadmin?
Settings>SMTP>Outbound>Strategies: Routing local else "proxmox_mg"
Settings>SMTP>Outbound>Routing: RelayHost "proxmox_mg", local, mx
Fill in Server Details and Authentication here for proxmox_mg

I'm using Stalwart v0.13.4 and my UI doesn't have the strategies setting. Any idea why? (Screenshot)

[Obsidiannite21]

That is what I see as well.

[pepa65]

Go to: Management>Maintenance>Update Webadmin
I am on 0.13.4 as well. The webadmin needs to be manually updated (either on the webpage or through the API).

[David-Development]

Well.. that was embarrassing.. 😅 I thought updating Stalwart (Docker Container) would automatically include the updated WebAdmin Interface that's why I never dared to click the update button.. Thank you @pepa65

[pepa65]

I don't understand why stalwart not automatically gets the appropriate version of the webadmin interface @mdecimus , but that's how it is for now...

[mikaelhg]

Thanks @pepa65, for pointing out the admin UI manual update, and the reason why delivery suddenly stopped working was that routes turned into strategies. I have Stalwart on automatic upgrade with watchtower, but if it can just suddenly stop working during a minor version upgrade, I don't think that'll be a good configuration.

[pepa65]

I do manual upgrades through a script which also takes care of the webUI upgrade, I think this is safest. It seems Stalwart is going to change in the near future, so let's wait and see.