The port definition in docker-compose.yml has double inputs, so please delete one in rows 453-455 ports: - "${HTTPS_BIND:-}:${HTTPS_PORT:-443}:${HTTPS_PORT:-443}" - "${HTTP_BIND:-}:${HTTP_PORT:-80}:${HTTP_PORT:-80}" This throws an error with ports already bound When you put it like this, it works: ports: - "${HTTPS_BIND:-}:${HTTPS_PORT:-443}" - "${HTTP_BIND:-}:${HTTP_PORT:-80}"

docker-compose.yml with Update March 2026 with error

iup6004

The port definition in docker-compose.yml has double inputs, so please delete one in rows 453-455
ports:

“${HTTPS_BIND:-}:${HTTPS_PORT:-443}:${HTTPS_PORT:-443}”
“${HTTP_BIND:-}:${HTTP_PORT:-80}:${HTTP_PORT:-80}”
This throws an error with ports already bound

When you put it like this, it works:
ports:

“${HTTPS_BIND:-}:${HTTPS_PORT:-443}”
“${HTTP_BIND:-}:${HTTP_PORT:-80}”

DocFraggle

iup6004 No, you are wrong. Those are not “double inputs”, that’s normal docker compose style

[host_ip:]host_port:container_port

So if you have set your variables like

HTTPS_BIND=127.0.0.1
HTTPS_PORT=8443

it’s translated to

127.0.0.1:8443:8443

iup6004

Strange?
When a add the “double” entry again, I get still the following error:

Error response from daemon: failed to set up container networking: driver failed programming external connectivity on endpoint mailcowdockerized-nginx-mailcow-1 (17b5a89a39b929acbd8f9134e5b3069af015e51da3fdc2081ef2e5cb326ac5b1): Bind for 0.0.0.0:80 failed: port is already allocated

As soon as a delete the second :${HTTPS_PORT:-443} and :${HTTP_PORT:-80}, all containers are starting (as well the mailcowdockerized-nginx-mailcow-1 container) again.

According to https://goneuland.de/mailcow-e-mail-komplettsytem-mit-antivirus-spam-filer-webmail-webfrontend-installieren-mittels-docker-und-traefik/#comment-6783 it was a configuration missing in my docker-compose.override.yml with “ports: !reset” for the container nginx-mailcow.

All good now!

DerGunni

iup6004 thanks. Adding the ports-reset-ection below to docker-compose.override.yml fixed it here aswell:
nginx-mailcow: ports: !reset

MaxPain

I installed the last two updates today (yes, I know I’m a bit late — I only did it today, 21 days after 2026-03a and 2026-03b were released).

I just wanted to confirm that this bug still exists and that I also had to fix the docker-compose.override.yml file by adding this extra entry:

nginx-mailcow:
ports: !reset

It’s working again now. I was briefly in panic mode because I hadn’t made a backup before the update… definitely not best practice, but we learn from our mistakes.

Kevo

MaxPain I just wanted to confirm that this bug still exists and that I also had to fix the docker-compose.override.yml file by adding this extra entry:

nginx-mailcow:
ports: !reset

I think that indicates a problem with the installation somewhere. It’s my understanding that the !reset tag just sets a default value or null value. The fact that it works makes me think there is something else trying to use those ports somewhere. If it were me I’d want to track that down and fix it rather than relying on the !reset option.

DocFraggle

I agree. To track it down, stop the stack with

docker compose down

Reboot your system so that all other services start with their default ports and check which ports are used with

netstat -tulpen

I bet there is another service using port 80 (and maybe port 443) per default which you have to disable