====== Your own streaming server ======
Four steps. One command does the install, and it asks you nothing. About ten minutes, and most of that is waiting.
This builds the machine that catches your bonded phone connection and passes it on to YouTube, Twitch, Kick or anywhere else. You do not need to understand servers to do it. The receiver arrives as **one ready-built Docker container** called ''%%StreamHelpers-Receiver%%'' — nothing is compiled on your machine, and there is no Docker Compose to install.
**Restreamer is not Restream.io.** This receiver includes datarhei Restreamer. It can publish directly to your platforms, or publish to the separate Restream.io service for distribution. You do not need a Restream.io subscription just to use the receiver or publish to several platforms. See [[dashboard:restreamer-vs-restream-io|Restreamer vs Restream.io: do I need both?]] before choosing your path.
If your receiver was installed before September 2026 and lives in ''%%/opt/srtla-receiver%%'', it is the older Compose stack: everything from the dashboard's side is the same, but the commands are different — see [[#older-receiver-the-compose-stack|Older receiver (the Compose stack)]] at the end of this page.
===== 1. Pick a machine =====
Anything that runs Docker on an ordinary Intel/AMD processor or a 64-bit ARM processor will do, including Apple Silicon Macs. There are two sensible answers:
**Use something you already own.** A NAS, a home server, a mini PC, an old laptop left switched on — if it is already running all day, it costs you nothing extra. Two things a machine at home needs that a rented one does not: it has to be **reachable from outside** (one UDP port forwarded on your router, and a dynamic DNS name if your home address changes), and your home //upload// speed becomes the ceiling for your stream, because the receiver does not just catch your video, it sends it on from your house. Bonding four modems into 15 Mbit/s achieves nothing if your broadband uploads at 8. On Unraid, read [[receiver:unraid|Running the receiver on Unraid]] first.
**Or rent one.** Worth doing if your home upload is weak, if your ISP blocks incoming connections, or if you would rather your stream did not depend on your house having power. A rented server costs somewhere between $5 and $12 a month. Read [[receiver:hosting|Renting a server]] for current plan names and prices before you click — the cheapest safe choice changes, and the cheapest-looking plan is often ARM.
**Set a budget before you create the server.** In the provider's billing settings, add a budget or spending alert a little above the plan's monthly price, and use a hard spending cap or prepaid credit where the provider offers one. A budget is an alarm, not a fuse: it tells you early when a forgotten server, a bigger plan picked by mistake or traffic over the plan's allowance starts running up the bill. Do the same for a [[receiver:failover-server|failover server]]. When you no longer need a server, destroy it — most providers keep billing for a server that is only stopped.
**Intel/AMD and ARM64 are both supported.** The installer asks Docker which processor it runs on and downloads the matching native image automatically. Apple Silicon, Hetzner's ''%%CAX%%'' line, Oracle's Ampere shapes and Amazon's Graviton instances use ARM64. Existing Intel/AMD receivers keep using the x86-64 image. Older 32-bit ARM machines are not supported.
**Choose the location carefully.** This is the one decision on the page you cannot easily undo. Put the receiver near **the platform you stream to** — not near where you live, and not in the region closest to where you will be filming. SRTLA is built to get your video from a moving vehicle to the server across bad connections; the leg it cannot help with is server-to-platform, so give that one the short, stable path.
Many providers work — DigitalOcean, Vultr, Linode, OVH, Hetzner and others. Pick **Ubuntu** when it asks for an operating system, and make sure the machine comes with a **public IPv4 address** (Hetzner charges a little extra for one; an IPv6-only plan cannot be reached by your phones). Two shared CPU cores and **2 GB of memory** is plenty. The receiver passes video through without re-encoding it, so the processor sits almost idle and a bigger plan buys nothing.
**A Windows PC?** Use Docker Desktop with its WSL 2 backend and run the receiver command in an Ubuntu terminal. The Windows steps below explain how to set that up; Docker Desktop does not install Ubuntu for you.
**A Mac?** Install Docker Desktop and complete its first-run setup. Then paste the receiver command into macOS Terminal, not Docker Desktop's Logs or Exec tab. The installer finds Docker Desktop even if its command-line shortcuts were not installed, starts it when needed and waits for the engine. No manual PATH changes are required. Intel and Apple Silicon Macs use their respective native images. Keep the Mac awake while it is receiving.
===== 2. Run one command =====
Open a terminal on that machine and paste this in. On a rented server, the provider's control panel has a console button that opens one in your browser — you do not need to install anything to get there. On Unraid, open the terminal from the top-right of its web page; you are already root there, so the command runs as-is.
wget -qO- https://streamhelpers.com/receiver | sh
This is the same command the dashboard's **Set up your stream** wizard shows. Older dashboards, notes and bookmarks show the longer address ''%%https://streamhelpers.com/irl-dashboard/receiver/install.sh%%'' — that is the same installer, and it still works.
==== macOS: install from Terminal ====
Install [[https://docs.docker.com/desktop/setup/install/mac-install/|Docker Desktop for Mac]], choosing **Apple Silicon** or **Intel** to match your Mac. Open Docker from Applications and complete its first-run setup and any permission dialogs.
Press **Command + Space**, type **Terminal** and press Return. Use the Mac's own Terminal app, **not** a container's Logs or Exec tab in Docker Desktop. Paste this command and press Return:
curl -fsSL https://streamhelpers.com/receiver | sh
''%%curl%%'' is built into macOS; ''%%wget%%'' might not be. You do not need Homebrew, Docker Compose or a compiler. The installer finds Docker Desktop's command-line tool automatically even if it is not on your PATH, starts Desktop if needed and waits for the engine. Intel and Apple Silicon Macs get their matching native image.
Leave Terminal open until installation finishes. Open [[http://localhost:8088|http://localhost:8088]] in your Mac's browser, then follow **Add a Tailscale key on the settings page** below. Keep Docker Desktop running and the Mac awake while receiving.
==== Windows: install from an Ubuntu terminal (WSL 2) ====
Use a supported Windows 10 or Windows 11 machine with hardware virtualisation enabled. The receiver is a Linux container: **do not paste the receiver command directly into PowerShell or Command Prompt**, and do not use a container's Exec tab.
First install Ubuntu under WSL. Right-click **Start**, choose **Terminal (Admin)** or **Windows PowerShell (Admin)**, and run:
wsl --install -d Ubuntu
Restart Windows if prompted. Open **Ubuntu** from Start and finish its first-run setup, choosing a Linux username and password. Password entry shows no characters; that is normal. If Ubuntu is already installed, check it from PowerShell with ''%%wsl --list --verbose%%''. Its VERSION must be **2**; if it says 1, run ''%%wsl --set-version Ubuntu 2%%'' (use the exact distribution name shown in the list).
Install [[https://docs.docker.com/desktop/setup/install/windows-install/|Docker Desktop for Windows]], open it and complete its first-run setup. In Docker Desktop:
* Under **Settings → General**, enable **Use the WSL 2 based engine**.
* Under **Settings → Resources → WSL Integration**, enable your **Ubuntu** distribution and apply the changes.
* Use **Linux containers**, not Windows containers. If WSL Integration is missing, check Docker Desktop's tray menu for **Switch to Linux containers**.
Close and reopen **Ubuntu** from Start. Alternatively, open Windows Terminal and choose **Ubuntu** from its tab dropdown. Check Docker is reachable from that Ubuntu terminal:
docker info
If it says ''%%docker: command not found%%'' or cannot connect to the daemon, start Docker Desktop and check Ubuntu's WSL Integration setting above. **Do not install a second Docker Engine inside Ubuntu.**
Once ''%%docker info%%'' works, paste this into the **same Ubuntu terminal** and press Enter:
curl -fsSL https://streamhelpers.com/receiver | sh
Leave the terminal open until installation finishes. Open [[http://localhost:8088|http://localhost:8088]] in your normal Windows browser, then follow **Add a Tailscale key on the settings page** below. Keep Docker Desktop running and prevent Windows from sleeping while receiving.
==== Windows: PowerShell alternative and common setup problems ====
You can launch the Linux installer from **PowerShell** without opening Ubuntu separately. This command downloads a fresh copy, shows download progress and only runs it if the download succeeds:
wsl -d Ubuntu -- sh -c 'curl -fL --connect-timeout 15 --max-time 60 https://streamhelpers.com/receiver -o /tmp/streamhelpers-receiver-install.sh && sh /tmp/streamhelpers-receiver-install.sh'
Use your exact distribution name if it is not ''%%Ubuntu%%''. Copy only the command inside the code block, not ''%%PS C:\...>%%'', output, ''%%> >%%'' markers or an entire troubleshooting transcript. The ''%%&&%%'' above belongs to the Linux shell inside the quotes, not to PowerShell.
**Ubuntu already exists, but Docker says “this WSL 1 distro”.** ''%%ERROR_ALREADY_EXISTS%%'' from ''%%wsl --install%%'' means Ubuntu is already installed, not that you need to delete it. In PowerShell, check the distribution's generation:
wsl --list --verbose
The **VERSION** beside Ubuntu must be **2**. A recent ''%%wsl --version%%'' (for example, WSL runtime 3.x) does not prove that: the runtime and the distribution generation are different. ''%%wsl --update%%'' updates the runtime but does not convert Ubuntu. If Ubuntu says VERSION 1, run:
wsl --set-version Ubuntu 2
Let conversion finish, then enable Ubuntu under Docker Desktop's **Settings → Resources → WSL Integration**. Do not unregister/delete Ubuntu or reset Docker Desktop to fix this.
**Docker works in PowerShell but not Ubuntu.** Docker Desktop must be running, using Linux containers, and integrated with this specific distribution. Enable Ubuntu's integration and **Apply & restart**; if already enabled, toggle it off and apply, then back on and apply. Reopen Ubuntu, or check from PowerShell:
wsl -d Ubuntu -- docker info
Look for a **Server** section without an error, not just client version information. If it still fails, keep the full error for troubleshooting; a custom ''%%DOCKER_HOST%%'' or Docker context can point to a different engine. Do not install another Docker Engine inside Ubuntu or use ''%%systemctl%%'' to repair Docker Desktop integration.
**The installer says healthy/Done, but there is no copy-and-paste code.** Installation succeeded; first-time Tailscale setup is still required. Open [[http://localhost:8088|http://localhost:8088]] in the Windows browser, add the Tailscale auth key and save. The **Paste this into your dashboard** code appears once Tailscale connects. No reinstall is needed.
**The printed settings address starts with 172.x.** It may be Ubuntu's private virtual-network address rather than your Windows computer's LAN address. For local Docker Desktop, use ''%%http://localhost:8088%%'' on the same Windows computer first, even if the installer prints a different address. Do not use the WSL address as your router's port-forward destination; forward UDP to the Windows computer's actual LAN address. A custom/remote Docker context is different: the ports belong to that Docker host, not automatically to your local PC.
==== DNS failures or a download that appears to do nothing ====
''%%curl: (6) Could not resolve host%%'' or ''%%Resolving timed out%%'' happens before the installer can run if Ubuntu cannot resolve the download site. A silent ''%%curl -s%%'' command can look like a blinking cursor while it waits; use the progress/timeout PowerShell command above to see the failure promptly.
Test Windows and Ubuntu separately from PowerShell:
Resolve-DnsName streamhelpers.com
wsl -d Ubuntu -- getent ahostsv4 streamhelpers.com
wsl -d Ubuntu -- curl -I -L --connect-timeout 10 --max-time 20 https://streamhelpers.com/receiver
If Windows resolves the name but Ubuntu returns no addresses, the site is not necessarily down: WSL's DNS path is failing. Home DNS such as AdGuard Home/Unbound, router DNS, VPNs, Tailscale DNS and firewall policies are possible things to check, **not proof that any one caused it**. Check WSL 1 versus WSL 2 first. Updating the WSL runtime alone does not convert a WSL 1 distribution.
For diagnostics, inspect Ubuntu's generated configuration and Windows' configured IPv4 DNS servers:
wsl -d Ubuntu -- cat /etc/resolv.conf
Get-DnsClientServerAddress -AddressFamily IPv4 | Where-Object { $_.ServerAddresses.Count -gt 0 } | Format-Table InterfaceAlias,ServerAddresses -AutoSize
Simply replacing the nameserver with your router or ''%%8.8.8.8%%'' is not a guaranteed fix: direct DNS can be blocked or unreachable from WSL even when HTTPS works. Do not repeatedly overwrite ''%%/etc/resolv.conf%%'' or permanently disable generated DNS without understanding the network configuration.
To distinguish DNS from HTTPS connectivity, use a **current IPv4 address returned by Windows** with curl's ''%%--resolve%%'' option. Replace the placeholder below; do not paste it literally:
wsl -d Ubuntu -- curl -I -L --connect-timeout 10 --max-time 20 --resolve streamhelpers.com:443:CURRENT_IPV4_FROM_WINDOWS https://streamhelpers.com/receiver
If this succeeds while the normal request fails, HTTPS to that address works and DNS remains the problem. This bypass is only a diagnostic; it does not repair DNS for Docker, Tailscale or other downloads. Cloudflare addresses can change, so do not copy an old address from a forum or leave a permanent hosts-file pin. Keep HTTPS certificate verification enabled.
If earlier troubleshooting changed DNS, retain the original backup and restore the intended WSL/network configuration once the cause is understood. Remove any temporary ''%%streamhelpers.com%%'' hosts-file entry marked ''%%receiver-install-temporary%%'' after normal lookup works. Avoid deleting unrelated hosts entries or blindly restoring a backup over later intentional changes.
**Suspect an old installer?** Re-download rather than running a saved ''%%/tmp/streamhelpers-receiver-install.sh%%''. The newer script prints **Checking operating system**, Docker's server version and **Checking installer tools**; a WSL 2-specific integration message is also evidence you received the update, not the old generic Docker error. Download progress alone, or a smaller redirected response before the script, is not evidence of stale caching. If comparing a direct installer URL with a cache-busting query, use ''%%https://streamhelpers.com/irl-dashboard/receiver/install.sh?check=UNIQUE_VALUE%%''; a query on the short redirect may not carry through to the target. Never disable TLS verification to work around caching.
**Receiving video from outside your home:** on either Mac or Windows, the localhost address is only the settings page on that computer. You still need **UDP 25554** forwarded to the computer, as described in step 3. On Windows, also allow that incoming UDP port through Windows Firewall; opening only a TCP port does not work.
==== Update or reinstall on Mac or Windows ====
Use **macOS Terminal** on a Mac, or the **Ubuntu terminal** on Windows:
curl -fsSL https://streamhelpers.com/receiver | sh -s -- --reinstall
This keeps the receiver's data volume, settings, recordings, Tailscale login and existing bonded-stream port. **Run it while you are not streaming:** download and loading happen before the old receiver is removed, but replacing the container briefly interrupts reception. Do not delete the receiver's Docker volume or reset Docker Desktop to update it.
==== What the installer does ====
**What it needs.** Docker Engine **20.10 or newer**, running **Linux containers**, a POSIX shell and ''%%curl%%'' or ''%%wget%%''. No Docker Compose or build tools. The installer checks the shell utilities it uses before creating the receiver volume. If Docker is missing on a cloud server, it prints the Linux installation command (''%%curl -fsSL https://get.docker.com | sh%%''); on a NAS, enable or update Docker in the UI.
**Windows prerequisite errors now explain the fix.** WSL 1 is rejected before Docker is used, with ''%%wsl --list --verbose%%'' and the PowerShell conversion command. Updating WSL itself does not convert an existing Ubuntu distribution. WSL 2 without Docker access gets Docker Desktop integration instructions, not advice to start a separate Linux Docker service. If you saved an old installer under ''%%/tmp%%'', download it again to get these checks.
**Download failures:** image downloads show progress, and curl bounds connection/stalled-transfer waits. Its error distinguishes DNS lookup, connection and certificate problems. Windows DNS working does not prove Ubuntu's DNS works; check inside Ubuntu before retrying. An installer cannot diagnose a failure downloading the installer itself, because its code has not started yet. Do not disable HTTPS certificate verification or reset Docker Desktop to fix DNS.
On macOS, missing Docker gets Docker Desktop instructions instead of the Linux installation command. The installer never installs Docker Desktop silently or bypasses its first-run permission dialogs. With local Docker Desktop, the settings address printed at the end is ''%%http://localhost:8088%%'' on that Mac; outside video senders still need the UDP port forwarded to the Mac.
**What it does.** It asks you nothing. It checks the processor, then checks that Docker can actually //write// to its storage — it creates the receiver's data volume before downloading anything, so a full or read-only Docker disk (a classic Unraid fault) produces one clear sentence instead of a confusing failure part-way through the download. Then it downloads a ready-built image of about 250 MB, loads it into Docker and starts one container called ''%%StreamHelpers-Receiver%%''. Expect a minute or two of downloading and a short wait while it starts — the first start takes a little longer because Restreamer initialises its database. The container is set to restart itself, so it comes back after a reboot.
Inside that one container are the four things a receiver needs: ''%%srtla_rec%%'', the bonding receiver from BELABOX; the official ''%%datarhei/restreamer%%'', which publishes to your platforms; a backup-video controller that swaps to a video of your choice when the bond drops; and Tailscale. The dashboard's [[dashboard:go-live#automatic-backup-stream|Backup stream]] card is where you upload that video and tune the switching. Everything the receiver stores — its settings, Tailscale's identity, Restreamer's configuration and recordings — lives in one Docker volume, ''%%streamhelpers-receiver-data%%'', which survives updates.
**When every bonded connection drops at once** — a tunnel, a dead zone — the receiver holds your stream open for up to ten seconds, so it carries on the moment the signal is back instead of ending and starting again. A receiver installed before 4 October 2026 gets this when you update it with ''%%--reinstall%%'' ([[#if-something-looks-wrong|below]]).
**Voice isolation** (the dashboard's Restreamer tab) uses DeepFilterNet3 on receivers installed or updated from 12 October 2026. Older receivers cannot run it: the card shows a fault until you update them with ''%%--reinstall%%''.
**When it finishes** it prints the address of the receiver's own **settings page** on port ''%%8088%%'', such as ''%%http://192.168.0.50:8088%%''. Open that in a browser on the same network as the machine. It is your next stop.
==== Add a Tailscale key on the settings page ====
The settings page has a box marked **Tailscale auth key — required**, and both the installer and the page link straight to [[https://streamhelpers.com/tailscale-keys|streamhelpers.com/tailscale-keys]], which walks you through creating one. Sign in to Tailscale, click **Generate auth key**, change nothing, copy the key (it starts with ''%%tskey-%%''), paste it into the box and press **Save changes**.
Tailscale is a free private network joining only your own devices — this server, your router, your laptop, your phone. Here it is not optional, for the same reason on a rented server and on a machine at home: **the receiver never puts Restreamer or its settings page on the open internet.** Restreamer's own page (port ''%%8080%%'') is reachable only over the tailnet, the settings page refuses connections from public addresses, and until the receiver has a tailnet address it has **no pairing code to give you** — the page says //This server cannot be paired with your dashboard yet//. Video still arrives in the meantime; it is only the dashboard's control connection that waits.
Tailscale is also what lets you open both pages from a phone on mobile data, in another city, without exposing anything. That matters most for a machine at home, whose pages otherwise only open at home — the one place you will not be when you need them.
Free for personal use, and the free plan is far larger than anything here will come close to needing — [[https://tailscale.com/|tailscale.com]]. When you save the key, only the Tailscale service inside the container restarts; your video is unaffected. **Then turn off key expiry** for the new machine in Tailscale's admin console (the machine list, the machine's menu, //Disable key expiry//). If you skip that, Tailscale drops the machine months later and the dashboard silently loses its control connection.
If you are already on a tailnet, nothing changes — the receiver simply appears in it under the name in **Tailscale machine name** (''%%streamhelpers-receiver%%'' unless you change it). You do not need Unraid's or Synology's own Tailscale plugin; the receiver brings its own, inside the container, and it can happily sit alongside a plugin that puts the NAS itself on your tailnet. Video never travels through Tailscale.
**Cannot reach port 8088 from where you are?** Until Tailscale connects, the settings page only opens from the machine's own network. From elsewhere, use an SSH tunnel — ''%%ssh -L 8088:localhost:8088 user@your-server%%'', then open ''%%http://localhost:8088%%'' — rather than forwarding ''%%8088%%'' on the firewall, which the page would refuse anyway. Once Tailscale is up, the page is at ''%%http://:8088%%'' from anywhere.
**Something go wrong, or want to start over?** Run the same command with ''%%| sh -s -- --reinstall%%'' on the end. It replaces the container and keeps the data volume, so settings, recordings and your Tailscale login all survive — that is also how you update. If you chose a different ''%%--srt-port%%'' when you installed, give it again here. Running the plain command again on a machine that already has a receiver does **not** reinstall anything: it finds the receiver, leaves it running and prints your pairing code and settings-page address again, so it is safe to do mid-stream.
===== 3. Open one port =====
Your phones need to reach the server, so one port has to be let through: **UDP 25554**. The settings page lists it under **Ports to forward**, and the installer prints it at the end.
|If you rented a server |Open it in the provider's firewall — Hetzner calls it Firewalls, AWS calls it Security Groups, DigitalOcean calls it Cloud Firewalls. Protocol **UDP**, port **25554**, source **Anywhere** or **0.0.0.0/0**. This is separate from any firewall on the machine itself.|
|If you used your own machine|Forward it on your router — the setting is usually under Port Forwarding or NAT. Send **UDP 25554** to the machine's local address, which the installer prints for you. |
**It must be UDP, not TCP.** This is the single most common reason a setup that looks finished never receives anything. Bonding is UDP. If the firewall only offers you one box and it says TCP, look for a protocol dropdown.
Do not restrict the source to your phone's IP address. Mobile networks hand out a different one constantly, and the SIM that gets a new address simply stops contributing to the bond.
That is the only port. Nothing else is forwarded, even with Tailscale running — and the bonded video cannot go //through// Tailscale either. Bonding works by sending across several connections at once and continuously measuring which is healthiest; a tunnel collapses all of them onto one path, which defeats the whole mechanism while still looking like it works. Restreamer's port ''%%8080%%'' and the settings page on ''%%8088%%'' stay closed to the internet on purpose; the dashboard reaches them over the tailnet.
If you need a different port, choose it when you install: ''%%wget -qO- https://streamhelpers.com/receiver | sh -s -- --srt-port 25555%%''. The port is fixed when the container is created — Docker only forwards that one port into it — so changing **Bonded stream port** on the settings page cannot move it; newer receivers show the field read-only and say so. To move an existing receiver, run ''%%--reinstall --srt-port 25555%%'' (every other setting is kept), and give the same ''%%--srt-port%%'' again whenever you ''%%--reinstall%%'' later. Whatever you choose, forward that.
===== 4. Put the values in the dashboard =====
The settings page has a section headed **Paste this into your dashboard** with one long line starting ''%%IRLCFG-%%'' and a **Copy code** button. The line appears the moment Tailscale connects. You do not have to read it or understand it — copy the whole line, open the dashboard, go to **Set up your stream** (a new dashboard opens there by itself; otherwise **Settings → Guided setup**), paste it into the box on **The machine at the far end**, and press **Apply**.
That fills in the receiver address, the receiver port, your Restreamer address and your Restreamer login in one go, and tells you exactly what it changed. It merges into what you already have rather than wiping anything. Restreamer's login is already set — the receiver generated a password on its first start, the user is ''%%admin%%'', and there is no first-run setup screen to go and find. **New password** on the settings page makes a fresh one; copy the code again afterwards and paste it into the dashboard again, because the old code carries the old password.
**That line contains your Restreamer password.** It is only base64, not encryption. It goes no further than your own dashboard, but do not paste it into a Discord, a forum or a bug report when you are asking for help — and the same goes for screenshots of the settings page, which shows the password in clear.
Lost it? Open the settings page again, or run the install command again on the server — on a machine that already has a receiver it changes nothing and just prints the code.
Would rather type the values yourself? The settings page shows them in plain text. Open **Settings**, go to the **Bonding** tab, and enter:
|Receiver host|The server's **public** address — never a tailnet address, because bonding cannot run through a tunnel|
|Receiver port|''%%25554%%'', or whatever you set as the bonded stream port |
Then, under **Settings → Restreamer**, the Restreamer address is ''%%http://:8080%%'' — the **Tailscale** address this time, because that is the only one Restreamer answers on — with the login from the settings page. The two addresses are opposite on purpose: video goes to the public address, control goes over the tailnet.
**If you used a machine at home, set up a dynamic DNS name.**
Home connections are given a new public address from time to time, and when yours changes the dashboard will simply stop reaching the server. A dynamic DNS name is free, points at your connection permanently, and updates itself. Most routers have DuckDNS or No-IP built in — look under Dynamic DNS. Use that name as the Receiver host instead of the number.
Rented servers keep their address for the life of the server, so this does not apply to them. The receiver works out its own public address; if it gets that wrong — an unusual NAT, or a host with several addresses — fill in **Public address override** on the settings page and the pairing code updates.
phones —SRTLA/UDP 25554→ your server —SRT, inside the container→ Restreamer —RTMP→ YouTube / Twitch / Kick
**Very advanced, optional: a failover server.** Once this server works, you can add a second, cheap one that sits idle and can take over if this one stops answering mid-stream. It is set up from the dashboard's Restreamer tab; see [[receiver:failover-server|Failover server]].
===== If something looks wrong =====
The settings page's **Status** section shows each service — restreamer, srtla, standby, tailscale — with its state and any note it wants you to read. That is the first place to look.
^What you see ^Almost always ^
|Dashboard says no answer from the receiver |**UDP 25554 is not open** — in the provider's firewall on a rented server, or as a port forward on your router at home. Or the address is typed wrong. |
|It worked for weeks, then the receiver stopped answering after you stopped and started the server |**The public address changed.** Rented servers keep theirs for the life of the server, so this points at a home connection whose ISP address moved, or a provider that does not give static addresses. Copy the pairing code from the settings page again and apply it, and set up a dynamic-DNS name so it cannot happen again.|
|Bonding works from Wi-Fi but one SIM never contributes |The firewall rule on **UDP 25554** has a restricted source. Mobile addresses are carrier-assigned and change constantly — set the source to **Anywhere / 0.0.0.0/0**. |
|Bonding connects but only uses one connection |You gave it a Tailscale address. Use the public IPv4. |
|The settings page says //This server cannot be paired with your dashboard yet// |No Tailscale key yet, or the key was refused — auth keys expire and most are single-use. Generate a fresh one, paste it in, **Save changes**. Video is still being received while you do this. |
|The settings page says //This server cannot receive a stream yet// |The machine has no public address, or its address is a carrier-grade NAT one shared with other customers. Port forwarding cannot fix that. Ask your ISP for a public IP, or rent a cloud server instead. Tailscale does not solve this — bonding cannot run through a tunnel. |
|Restreamer's page will not load |You are using the public address. Restreamer only answers over Tailscale — use ''%%http://:8080%%'', or press **Open Restreamer →** on the settings page. |
|Restreamer's page stopped loading over Tailscale months later |Tailscale expired the machine's key. Paste a fresh key into the settings page and save — only the Tailscale service restarts. Then turn off key expiry for the machine in your Tailscale machine list so it cannot happen again. **Your stream keeps being received the whole time** — video never travelled over Tailscale.|
|Dashboard says the Restreamer login is wrong, but it is right |The address is missing **''%%:8080%%''**. Or you pressed **New password** on the settings page and did not paste the new code into the dashboard. |
|Stream is arriving but the preview is black, and the player says //“no available working or supported playlists”//|The channel is publishing H.265 only, which browsers cannot decode. Open the dashboard's Restreamer tab and use the fix button — it repackages so a browser-playable rendition exists alongside the original, without re-encoding your video or sending anything extra up the bond. |
|Firewall rule added and the stream still never arrives |You allowed **TCP**. Bonding is **UDP 25554**. |
|It worked, then stopped after a reboot |It should not — the container restarts itself. Check ''%%docker ps%%''; if it is missing, ''%%docker start StreamHelpers-Receiver%%''. If it keeps stopping, ''%%docker logs StreamHelpers-Receiver%%'' says why. |
|The installer says //Docker cannot create a volume// |Docker's storage is full or has gone read-only. This is a Docker fault, not a receiver one, and reinstalling containers will not fix it. On Unraid, read [[receiver:unraid|Running the receiver on Unraid]]. |
|The installer says //Port 25554 is already used by the container …// |Something else on the machine — usually an older receiver — holds the port. Stop that container, or pass ''%%--srt-port%%'' to use another one and forward that instead. |
|The installer says it is using native ''%%linux/arm64%%'' or ''%%linux/amd64%%'' |This is normal: both architectures are supported. No need to change your server plan. Older 32-bit ARM is not supported.|
|Windows install finishes healthy, but prints no pairing code |Open ''%%http://localhost:8088%%'' on that Windows PC, add a Tailscale auth key and save. The code appears after Tailscale connects; do not reinstall.|
|WSL says Docker is present but cannot reach its engine |Start Docker Desktop, use Linux containers and enable this Ubuntu distribution under WSL Integration. Verify ''%%wsl -d Ubuntu -- docker info%%'' shows a working Server section.|
|Windows resolves the site, but Ubuntu says “Could not resolve host” |Troubleshoot WSL DNS separately from Windows DNS; see the DNS section above. A forced public nameserver may also be unreachable.|
Logs, all of it at once: ''%%docker logs -f StreamHelpers-Receiver%%''\\
Restart it: ''%%docker restart StreamHelpers-Receiver%%''\\
Update it, keeping everything: ''%%wget -qO- https://streamhelpers.com/receiver | sh -s -- --reinstall%%''
Still stuck? Ask in ''#help'' on [[https://streamhelpers.com/support/|Discord]] — with the log, and without the pairing code.
===== The short version =====
Know what you are doing? This is for you.
wget -qO- https://streamhelpers.com/receiver | sh
A POSIX shell installer that needs Docker Engine 20.10+ with Linux containers, standard shell utilities and curl or wget — no Compose, and it never prompts. Windows requires a WSL 2 distribution with Docker Desktop integration. It checks Docker's architecture before downloading, creates the volume ''%%streamhelpers-receiver-data%%'' first as a storage-writability check, downloads the matching prebuilt image from ''%%streamhelpers.com/receiver-image%%'' (x86-64) or ''%%streamhelpers.com/receiver-image-arm64%%'' (ARM64), loads it with ''%%docker load%%'' and runs, in effect:
docker run --name StreamHelpers-Receiver --detach --restart unless-stopped \
--volume streamhelpers-receiver-data:/data \
--publish 25554:25554/udp --publish 8088:8088/tcp \
streamhelpers-receiver:-
Inside: the official ''%%datarhei/restreamer%%'' image, pinned by digest, plus ''%%srtla_rec%%'' built from BELABOX source (AGPL — licence and source are shipped in the image under ''%%/usr/share/licenses/srtla/%%''), ''%%tailscaled%%'', a backup-video controller driven by Restreamer's API, and a small Go supervisor that restarts anything that stops, serves the settings page on ''%%8088%%'', answers Docker's healthcheck and runs the SRT keeper. Everything talks over ''%%127.0.0.1%%'' inside the container: ''%%srtla_rec%%'' listens on udp/25554 and hands SRT to the keeper on 6001, which passes every packet on to Restreamer on 6000. When a sender goes silent, the keeper keeps its session alive with the sender's own keepalive for up to 15 s, so Restreamer's 2-second idle limit does not end it; and when Restreamer has dropped a session (a restart, say), it tells the sender at once, so the sender dials again instead of waiting out its own timeout. Restreamer's API and UI on 8080 are **not published** — the dashboard reaches them over the tailnet, which is what keeps the login off the internet. State is under ''%%/data%%'' in the volume: ''%%config.json%%'', ''%%restreamer/%%'', ''%%tailscale/%%''.
Flags: ''%%--ts-key KEY%%'' for unattended installs (cloud-init; the key is otherwise entered on the settings page), ''%%--srt-port N%%'', ''%%--name NAME%%'', ''%%--reinstall%%'' (replace the container, keep the volume), ''%%--failover%%'' (a [[receiver:failover-server|failover server]]: bonding receiver and Restreamer only, pass-through), ''%%--help%%''. ''%%--yes%%'' is accepted and does nothing, because nothing asks. ''%%IRL_IMAGE_URL%%'' overrides where the image is fetched from, for a mirror on your own network.
The settings page is served only to loopback, private and tailnet addresses. Changing the Tailscale key restarts only ''%%tailscaled%%''; changing the password restarts Restreamer and the backup controller; toggling the backup restarts only the controller. The bonded stream port is published by ''%%docker run%%'' and cannot be moved from the page — newer receivers show it read-only and refuse a change — so set it with ''%%--srt-port%%'' at install, and move it with ''%%--reinstall --srt-port N%%''. The pairing code is ''%%IRLCFG-%%'' followed by base64 JSON — receiver host and port, Restreamer's base URL on the tailnet address, and its login — and the dashboard merges it into its settings rather than replacing them. No tailnet address, no code.
router —SRTLA/UDP 25554→ srtla_rec —SRT, 127.0.0.1:6001→ SRT keeper —127.0.0.1:6000→ restreamer —RTMP→ YouTube / Twitch / Kick
===== Older receiver (the Compose stack) =====
Receivers installed before September 2026 — dashboard releases up to 2026.09.12 — are a Docker Compose stack in ''%%/opt/srtla-receiver%%'' rather than one container. They keep working, and the dashboard talks to them in exactly the same way, but the commands are different. This section is for those installs; nothing in it applies to the ''%%StreamHelpers-Receiver%%'' container above.
The stack is several containers sharing one network namespace, held by a do-nothing container named ''%%streamhelpers.com%%'': ''%%restreamer%%'' (the official image, unmodified), ''%%srtla_rec%%'' (built from BELABOX source on your machine — the long stretch where nothing appears to happen during that install is the compile, and Docker's //legacy builder is deprecated// notice is harmless), ''%%sh_standby%%'' for the backup video, and optionally ''%%sh_tailscale%%''. On a machine that already ran Restreamer it used a second compose file instead, leaving that container alone. The installer wrote its choices into ''%%/opt/srtla-receiver/.env%%'' (''%%SRTLA_PORT%%'', ''%%SRT_PORT%%'', ''%%CORE_PORT%%'', ''%%NET%%'', ''%%TS_AUTHKEY%%'', ''%%WITH_TAILSCALE%%''), saved the Restreamer login in ''%%/opt/srtla-receiver/restreamer-login.txt%%'' (readable only by root), and kept a copy of itself at ''%%/opt/srtla-receiver/install.sh%%''. Its Tailscale offer came in the terminal, and on Unraid it pointed you at the **Apps tab** for Unraid's own Tailscale plugin.
Everyday commands for the old stack:
cd /opt/srtla-receiver && docker compose up -d # bring everything back after a stop or reboot
cd /opt/srtla-receiver && docker compose logs -f # all logs at once
docker compose logs -f srtla # just the bonding
docker compose logs -f standby # just the failover
sh /opt/srtla-receiver/install.sh --check # report what is working, change nothing; prints the pairing code again
Use the saved copy for the old flags. The public address now serves the new installer, which has no ''%%--check%%'' or ''%%--uninstall%%'' and answers them with //Unknown option//. The old installer's ''%%--check%%'' reports and changes nothing, ''%%--reinstall%%'' rebuilds with a fresh source clone, and ''%%--uninstall%%'' removes only what it installed; none of them delete Restreamer's volumes, so recordings and channels survive all three. Never start a container in the shared namespace on its own — ''%%docker compose up -d%%'' is the way back up.
If Restreamer stopped loading over Tailscale months later, Tailscale expired the machine's key: put a fresh key in ''%%TS_AUTHKEY%%'' in ''%%/opt/srtla-receiver/.env%%'', run ''%%cd /opt/srtla-receiver && docker compose up -d tailscale%%'', then turn off key expiry for the machine in your Tailscale machine list. Video was never travelling over Tailscale, so the stream kept arriving throughout.
**Moving to the new receiver.** There is no automatic migration, and the new installer stops if the old stack still holds UDP 25554 — it names the container that has it. On the same machine: ''%%cd /opt/srtla-receiver && docker compose down%%'' (no ''%%-v%%'', so the old volumes stay on disk), then run the one command at the top of this page and follow steps 2 to 4. Your Restreamer channels, recordings and backup video are not carried across — set the destinations up again in the dashboard's Restreamer tab and upload the backup video again on the Backup stream card. On a fresh machine, simply follow the four steps and repoint the dashboard with the new pairing code.