You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Using port 8080:8080 for NextCloud blocks setup page #560
Thanks for the report — I dug into this today. A few verified facts that should clear this up:
The example targets the right port. The official nextcloud:latest image (35.0.2 as of today) runs Apache inside the container on port 80 — its Dockerfile is a plain php:8.5-apache base with no port overrides (nextcloud/docker → 35/apache/Dockerfile), and the image README says the same ("exposes port 80"). So tsdproxy.port.1: "443/https:80/http" in the example is correct for a brand-new image.
tsdproxy doesn't need any published ports on nextcloud. It reaches the container directly over the Docker network. The example intentionally has no ports: entry on the nextcloud service — if you added 8080:8080 there, that's the problem: host port 8080 is tsdproxy's dashboard (the 8080:8080 mapping on the tsdproxy service). When two containers try to bind the same host port, the second one can't bind — with a restart policy it may look "running" while no webserver ever comes up, which matches your 502 (tsdproxy can't connect to the target port).
If you want to open Nextcloud directly from the host as well, publish a free host port instead, e.g. ports: ["8081:80"] — the container-internal side stays 80 and the tsdproxy label doesn't change.
One caveat: if by "first time setup on 8080" you meant Nextcloud AIO — its setup interface really does live on port 8080, but that's a different image (nextcloud/all-in-one) that manages its own containers; the docs example is for the plain nextcloud image and doesn't apply to AIO as-is.
If it still 502s after removing any 8080:8080 mapping from the nextcloud service, could you share your docker-compose file and the output of docker ps for both containers?
Ah, yes, I'm talking about NextCloud AIO, here's my docker-compose.yaml :
name: nextcloud-aio # Add the container to the same compose project to which all the sibling containers are added automatically
services:
tsdproxy:
image: almeidapaulopt/tsdproxy:2
volumes:
- /var/run/docker.sock:/var/run/docker.sock
- tsdproxy-data:/data
- ./config:/config
ports:
- "8008:8008"
extra_hosts:
- "host.docker.internal:host-gateway"
restart: unless-stopped
nextcloud-aio-mastercontainer:
image: ghcr.io/nextcloud-releases/all-in-one:latest # This is the container image used. You can switch to ghcr.io/nextcloud-releases/all-in-one:beta if you want to help testing new releases. See https:/>
init: true # This setting makes sure that signals from main process inside the container are correctly forwarded to children. See https://docs.docker.com/reference/compose-file/services/#init
restart: always # This makes sure that the container starts always together with the host OS. See https://docs.docker.com/reference/compose-file/services/#restart
container_name: nextcloud-aio-mastercontainer # This line is not allowed to be changed as otherwise AIO will not work correctly
volumes:
- nextcloud_aio_mastercontainer:/mnt/docker-aio-config # This line is not allowed to be changed as otherwise the built-in backup solution will not work
- /var/run/docker.sock:/var/run/docker.sock:ro # May be changed on macOS, Windows or docker rootless. See the applicable documentation. If adjusting, don't forget to also set 'WATCHTOWER_DOCKER_SOCKET>
labels:
tsdproxy.enable: "true"
tsdproxy.authkey: "tskey-auth-kf8RmsS5Jy11CNTRL-3xn86aPnHGJk6HEZeqz7GJWX2GK91cQJ"
tsdproxy.name: "nube"
tsdproxy.dash.label: "Nextcloud"
tsdproxy.dash.icon: "si/nextcloud"
# HTTPS on 443 -> container port 80 (HTTP)
tsdproxy.port.1: "443/https:80/http, tailscale_funnel"
tsdproxy.port.2: "80/http->https://nube.<tailnet-name>.ts.net"
environment: # This line is needed (has to be uncommented) when using any of the options below
- PUID=1000
- PGID=1000
- OVERWRITEPROTOCOL=https
volumes: # If you want to store the data on a different drive, see https://github.com/nextcloud/all-in-one#how-to-store-the-filesinstallation-on-a-separate-drive
tsdproxy-data:
nextcloud_aio_mastercontainer:
name: nextcloud_aio_mastercontainer # This line is not allowed to be changed as otherwise the built-in backup solution will not work
It's not possible to change it using port 8080 for setup afaik.
I'm trying to do the setup without this plugin but I've been having so many issues... I was hoping this would solve it for me.
Thanks — the compose settles it: you're on Nextcloud AIO, which works differently from the docs example.
Port 8080 is AIO's own interface — the passphrase/setup page served by the mastercontainer (self-signed cert). AIO upstream explicitly says to open it by IP address, not a domain (HSTS will block domain access to it later), so that page isn't meant to go through tsdproxy or any hostname: do the initial setup directly at https://<server-ip>:8080. Your 8008:8008 mapping for tsdproxy's dashboard is fine and unrelated.
The example you linked targets the plain official nextcloud image (Apache listening on internal port 80) — that's where its 443/https:80/http label comes from. AIO doesn't match that shape.
To expose AIO's actual Nextcloud through tsdproxy: follow AIO's own reverse-proxy flow — start the mastercontainer with APACHE_PORT=<port> (their sample uses 11000). Nextcloud is then served inside the Docker network by the nextcloud-aio-apache container on that port, so label that container:
tsdproxy.port.1: "443/https:11000/http"
tsdproxy reaches it over the Docker network — no extra published port needed.
The 502 fits this picture: the proxy was up but the target port (the example's 80) had nothing answering on the AIO stack, so there was nothing to forward to.
An AIO-specific example in our docs would be a good addition — noted on our side.
Quick follow-up: the AIO example mentioned above is now up as #564 — a dedicated docs page for this exact setup: tsdproxy labels on the nextcloud-aio-mastercontainer, Nextcloud exposed via APACHE_PORT (443/https:11000/http), the AIO setup UI kept out of tsdproxy (upstream guidance is to open it by IP, since HSTS blocks domain access later), plus a manual-install variant where the nextcloud-aio-apache container can be labeled directly.
Since you have an AIO stack running: if you get a chance to try the example, feedback over on #564 would be very welcome.
Ooh thank you, I decided to stick with Caddy and port forwarding since I realized Tailscale would severely limit the download speed and this was an unsuccessful attempt at trying to get over a bump that had me stuck for many days - turns out my issue was that I had a VPN client installed in my server while the VPN server was on my router, which was redundant and creating a bit of a loop when trying to validate my domain. So I won't be using tsdproxy at the moment, but hopefully this comes useful to other newbs out there like me that give it a shot.
I couldn't find again the page that told me about the download speed limit on Tailscale, did I get that right? I'd be using the free tier. It was something like 17mbp/s if I remember right but looking it up now I can't find such thing.
Thanks for the follow-up — glad you found the real culprit (the redundant VPN client creating a loop when validating the domain).
On the speed question: the free Personal plan has no bandwidth cap. Over a direct WireGuard connection you get essentially full line speed on every plan — plan limits are about users, devices, ACL groups and tagged resources, not throughput. Speed only drops in two situations:
Relayed connections. When two devices can't establish a direct link, traffic is relayed through a DERP server, and those limit throughput for fairness between users (per Tailscale's docs, relays "may offer lower maximum throughput"). There's no published per-plan number, so a figure like "17 mb/s" isn't an official limit — you may have seen a community benchmark of a relayed connection.
Tailscale Funnel. Exposing a service to the general public internet (ports 443/8443/10000) is subject to non-configurable bandwidth limits — that's the documented limit, and likely the page you half-remember.
Plain tailnet access (the typical tsdproxy use case) is direct-connection speed. You can check whether a given connection is direct or relayed with tailscale ping <machine> or tailscale status.
Hope Caddy + port forwarding serves you well — and tsdproxy will be here if you ever circle back.
Activity
almeidapaulopt commentedon Nov 13, 2026
Thanks for the report — I dug into this today. A few verified facts that should clear this up:
The example targets the right port. The official
nextcloud:latestimage (35.0.2 as of today) runs Apache inside the container on port 80 — its Dockerfile is a plainphp:8.5-apachebase with no port overrides (nextcloud/docker → 35/apache/Dockerfile), and the image README says the same ("exposes port 80"). Sotsdproxy.port.1: "443/https:80/http"in the example is correct for a brand-new image.tsdproxy doesn't need any published ports on
nextcloud. It reaches the container directly over the Docker network. The example intentionally has noports:entry on the nextcloud service — if you added8080:8080there, that's the problem: host port 8080 is tsdproxy's dashboard (the8080:8080mapping on the tsdproxy service). When two containers try to bind the same host port, the second one can't bind — with a restart policy it may look "running" while no webserver ever comes up, which matches your 502 (tsdproxy can't connect to the target port).If you want to open Nextcloud directly from the host as well, publish a free host port instead, e.g.
ports: ["8081:80"]— the container-internal side stays 80 and the tsdproxy label doesn't change.One caveat: if by "first time setup on 8080" you meant Nextcloud AIO — its setup interface really does live on port 8080, but that's a different image (
nextcloud/all-in-one) that manages its own containers; the docs example is for the plainnextcloudimage and doesn't apply to AIO as-is.If it still 502s after removing any
8080:8080mapping from the nextcloud service, could you share your docker-compose file and the output ofdocker psfor both containers?opponent019 commentedon Nov 13, 2026
Ah, yes, I'm talking about NextCloud AIO, here's my docker-compose.yaml :
It's not possible to change it using port 8080 for setup afaik.
I'm trying to do the setup without this plugin but I've been having so many issues... I was hoping this would solve it for me.
almeidapaulopt commentedon Nov 13, 2026
Thanks — the compose settles it: you're on Nextcloud AIO, which works differently from the docs example.
Port 8080 is AIO's own interface — the passphrase/setup page served by the
mastercontainer(self-signed cert). AIO upstream explicitly says to open it by IP address, not a domain (HSTS will block domain access to it later), so that page isn't meant to go through tsdproxy or any hostname: do the initial setup directly athttps://<server-ip>:8080. Your8008:8008mapping for tsdproxy's dashboard is fine and unrelated.The example you linked targets the plain official
nextcloudimage (Apache listening on internal port 80) — that's where its443/https:80/httplabel comes from. AIO doesn't match that shape.To expose AIO's actual Nextcloud through tsdproxy: follow AIO's own reverse-proxy flow — start the mastercontainer with
APACHE_PORT=<port>(their sample uses 11000). Nextcloud is then served inside the Docker network by thenextcloud-aio-apachecontainer on that port, so label that container:tsdproxy reaches it over the Docker network — no extra published port needed.
An AIO-specific example in our docs would be a good addition — noted on our side.
almeidapaulopt commentedon Nov 15, 2026
Quick follow-up: the AIO example mentioned above is now up as #564 — a dedicated docs page for this exact setup: tsdproxy labels on the
nextcloud-aio-mastercontainer, Nextcloud exposed viaAPACHE_PORT(443/https:11000/http), the AIO setup UI kept out of tsdproxy (upstream guidance is to open it by IP, since HSTS blocks domain access later), plus a manual-install variant where thenextcloud-aio-apachecontainer can be labeled directly.Since you have an AIO stack running: if you get a chance to try the example, feedback over on #564 would be very welcome.
opponent019 commentedon Nov 16, 2026
Ooh thank you, I decided to stick with Caddy and port forwarding since I realized Tailscale would severely limit the download speed and this was an unsuccessful attempt at trying to get over a bump that had me stuck for many days - turns out my issue was that I had a VPN client installed in my server while the VPN server was on my router, which was redundant and creating a bit of a loop when trying to validate my domain. So I won't be using tsdproxy at the moment, but hopefully this comes useful to other newbs out there like me that give it a shot.
I couldn't find again the page that told me about the download speed limit on Tailscale, did I get that right? I'd be using the free tier. It was something like 17mbp/s if I remember right but looking it up now I can't find such thing.
almeidapaulopt commentedon Nov 16, 2026
Thanks for the follow-up — glad you found the real culprit (the redundant VPN client creating a loop when validating the domain).
On the speed question: the free Personal plan has no bandwidth cap. Over a direct WireGuard connection you get essentially full line speed on every plan — plan limits are about users, devices, ACL groups and tagged resources, not throughput. Speed only drops in two situations:
Plain tailnet access (the typical tsdproxy use case) is direct-connection speed. You can check whether a given connection is direct or relayed with
tailscale ping <machine>ortailscale status.Hope Caddy + port forwarding serves you well — and tsdproxy will be here if you ever circle back.