Troubleshooting Remote Control
Most answers are on the tab itself, under Settings › Remote Control on the machine. Start with the line under Reach this machine from anywhere — it says what the tunnel is doing right now — and with Paired devices, which says what the machine last heard from each phone. While you pair, the square where the code goes says whether the roaming address has been proven to reach this machine.
The address is not reachable #
Read the status line under Reach this machine from anywhere:
| It says | What it means |
|---|---|
| Connected. This machine is online and reachable at … | The tunnel is up. If Show pairing code still says Checking the address reaches this machine…, the address is not answering from this machine yet — see The pairing code does not appear. Otherwise the problem is on the phone's side: check it has a connection at all, and that it is opening the exact address shown |
| Connecting… | Wait a moment. The address is not live until this changes |
| The connection stopped. It will be retried. | The machine lost its outbound connection. It reconnects on its own; check the machine's own internet access |
| Not connected. Only addresses on your own network work today. | No tunnel is running. Remote Control was just turned on, or the request for an address was refused. Save the tab to ask again immediately |
| Automatic setup is not available from your licence service. | Your licensing service does not issue addresses. Run your own tunnel under Advanced |
| Could not download the tunnel program: … or Could not download the tunnel program. Check this machine's internet access. | C5 could not fetch Cloudflare's connector, cloudflared, which every tunnel needs. The rest of the line says why. It retries on its own. See The tunnel program will not download |
| This deployment's network posture is offline, so C5 does not download the tunnel program. | The egress posture is Offline, so C5 will not download the tunnel program, and the tunnel cannot start without it. See The tunnel program will not download |
| The tunnel program downloaded with the wrong contents, so it was not run — a proxy or security product on this network may be altering it. Once that is fixed, save Remote Control's settings to try again. | The download arrived, but was not the exact file C5 pins. C5 does not run it, and does not retry on its own. See The download did not match |
| The tunnel program would not start on this machine. | The connector was downloaded but would not run. C5 retries; if it keeps failing, check that security software is not blocking ~/.growther/config/rc-tunnel/cloudflared |
| Tunnels are not available on … | Cloudflare publishes no connector for this platform. Use your own network, growther connect, or your own address |
Two things outside the tab:
- The machine has to be awake. A laptop lid closed, or a desktop asleep, is not reachable by any path. Adjust the machine's sleep settings if it needs to be reachable while unattended.
- The portal says No address reported. The install has not told your account where it is. Open C5 on that computer and check Remote Control is on and saved; the address reaches the portal on the next licence check-in.
If nothing helps, growther doctor on the machine and growther remote status say what the
install believes its state to be.
The pairing code does not appear #
After Show pairing code, C5 waits until the roaming address — the issued address,
https://<label>.rc.growther.si — is proven to reach this machine before it shows a code, so
the phone you pair can use it from anywhere. No code exists while it waits, so none expires.
The square where the code goes says what it is waiting for:
| It says | What it means |
|---|---|
| Asking Growther for this machine's roaming address… | The machine is asking Growther.si for its address and tunnel. A reason under it — Growther's licence service did not answer. Trying again., or an HTTP status — means the request is failing. While the screen is open, C5 asks again with growing gaps, up to half an hour apart |
| Connecting this machine to Cloudflare. The first time, this can take a minute or two. | The connector is starting. The first time, it is downloaded first |
| Trying to connect this machine to Cloudflare again… | An attempt failed and is being retried. The reason is under it — see The address is not reachable for what each one means |
| Checking the address reaches this machine… | The tunnel is up, and C5 is checking that the address answers from this machine and not from something else. A reason under it says what it found instead: Something other than this machine is answering at this address. points at a DNS record or a Cloudflare Worker route claiming the name; This address reaches a different C5 installation, not this one. points at two machines sharing one install's identity |
| A reason, and no animation | The address cannot come without someone changing something: a proxy or security product blocking the connector's download, an Offline network posture, a licensing service that does not issue addresses, or Cloudflare refusing the tunnel. Fix what it names, then press Show pairing code again |
This is taking longer than usual. appears after ninety seconds of waiting; the reason, if there is one, is already under the square.
While the square waits, Pair over this Wi-Fi instead pairs against the machine's address on your own network. It works only while the phone is on the same network as the machine, the phone warns about the certificate, and the phone keeps that address — to use it away from your network later, pair it again. The button is not offered when the machine has no secure address on your network; the screen says the secure port may not have opened.
The install keeps one pairing code per person. Pressing Show pairing code in a second window retires the first window's code; in the same browser, the first window says A newer pairing code is showing in another window. and offers a new one there.
The tunnel program will not download #
Every tunnel — the issued address or your own — runs Cloudflare's connector, cloudflared,
which C5 downloads the first time a tunnel starts. It fetches one pinned version from
Cloudflare's releases on GitHub and checks its size and sha256 before running it. The
download goes through the proxy and certificate authorities configured for C5, to
github.com, which redirects to GitHub's file-download host,
release-assets.githubusercontent.com (GitHub has also used
objects.githubusercontent.com for this).
When the download fails, the status line under Reach this machine from anywhere says why:
| The status line says | What to do |
|---|---|
| Could not download the tunnel program: the proxy asked for credentials (407). Add them to C5's proxy URL; C5 speaks Basic proxy authentication only. | Put the user name and password in the proxy URL (http://user:password@host:port). NTLM and Kerberos are not supported |
| Could not download the tunnel program: the proxy refused it (403). Ask for github.com and release-assets.githubusercontent.com to be allowed. | Ask your network team to allow both hosts. Allowing github.com alone fails at the redirect |
| Could not download the tunnel program: the proxy refused to open a tunnel (…). | The proxy answered with another status, given in brackets. A 502, 503 or 504 means the proxy could not reach GitHub itself |
| Could not download the tunnel program: the configured proxy cannot be used, and C5 does not connect round it. Fix or remove C5's proxy URL. | The proxy setting is not something C5 can use, such as a socks5:// address. growther doctor says what is wrong with it. See An unusable proxy stops traffic |
| Could not download the tunnel program: this network's certificate is not trusted. If a proxy inspects HTTPS, add its root certificate to C5's CA bundle. | Your proxy re-signs HTTPS with its own authority. Give C5 its root certificate — see Behind a TLS-inspecting proxy |
| This deployment's network posture is offline, so C5 does not download the tunnel program. | The egress posture is Offline (Settings › Enterprise › Network), so C5 does not download cloudflared, and the tunnel cannot start until the program is on this machine. (A tunnel whose program is already here does start while Offline — turn Remote Control off to stop it.) Your own network addresses and growther connect still work |
| Could not download the tunnel program. Check this machine's internet access. | Anything else: no connection, a name that does not resolve, a proxy that does not answer, a dropped download. Check that the machine is online and that github.com and release-assets.githubusercontent.com are allowed. Test connectivity under Settings › Enterprise › Network asks both |
The proxy URL and the CA bundle are delivered by device policy (Group Policy, macOS managed
preferences or /etc/growther), by GROWTHER_PROXY_URL and GROWTHER_CA_BUNDLE, or in
c5.yaml; in a shipped build the Network card only shows them. A signed policy document
cannot set them. See Corporate proxy.
Neither the status line nor C5's error log ever contains the proxy's address, because
anyone signed in to C5, and every paired phone, can read both. They say the proxy, and
what it answered. To see which proxy C5 is using, run growther doctor on the machine
(its Proxy reachability row), or open Settings › Enterprise › Network as an
administrator. See
The proxy address stays out of logs and alerts.
C5 writes the reason to its error log, with a line ending the tunnel program did not download — retrying in …s, and keeps trying on its own: soon at first, then once a minute.
While the posture is Offline those retries contact nothing. Once you fix the cause, or
switch back to Online, the tunnel comes up without anything more from you.
The download did not match #
When the download arrives but is not the exact file C5 pins — its fingerprint differs, the server sent more than the published size, the archive does not hold the program, or the program fails C5's own check — C5 throws it away and never runs it. That nearly always means something on the network altered it: a proxy, or a security product that inspects downloads.
Retrying would fetch the same wrong file again, so C5 does not: the tunnel stays off, and
the status line reads The tunnel program downloaded with the wrong contents, so it was not
run — a proxy or security product on this network may be altering it. Once that is fixed,
save Remote Control's settings to try again. The error log has a line reading
not retrying the cloudflared download automatically: …, followed by the exact reason.
Ask your network team to let the download from github.com and
release-assets.githubusercontent.com through unaltered, then press Save on the Remote
Control tab (or turn Remote Control off and on) to try again.
The phone warns that the certificate is not trusted #
On your own network, this is expected. The machine issued that certificate itself, and nothing outside it vouches for it, so your phone is right to say so. Accept it once per device — and only on a network you trust. On a network you do not trust, a warning like this is exactly the one you should not click through.
To be rid of the warning: pair against the issued address instead, which carries a certificate your phone already trusts, or supply your own certificate under Advanced. See Setup.
Two machines keep taking the address from each other #
The symptom: the line under Reach this machine from anywhere says Connected for a while, then drops to The connection stopped. It will be retried. and stays there — and on another machine the same address has just come up.
The address Growther.si issues is derived from your install's identity, and a copy of an install shares that identity: a second copy restored from a backup, a cloned virtual machine, or the old machine still running after you moved to a new one. When a copy asks for its address, Growther.si treats it as the same install asking again: the tunnel the other copy was using is deleted, a fresh one is created under the same name, and the address now points at the copy that asked last. The other copy loses it — its connector keeps retrying a tunnel that no longer exists — and takes it back the next time Remote Control is turned off and on there.
Two copies cannot share one address, and growther activate does not separate them: on a
machine that already has an identity, it resumes as that same install, with the same
address. To settle it:
- Stop C5 on the copy you are retiring (
growther stop), and do not start it again. If it starts on its own, also rungrowther service uninstallthere. - On the machine you keep, turn Remote Control off and on to take the address back. Turning it off unpairs every device, so pair your devices again afterwards.
Do not turn Remote Control off on the copy you are retiring while the other one is in use,
and do not run growther uninstall there, which turns it off too. The two share one
identity, so that would release the address the other machine is using.
If both machines must stay in use, contact support; see
Getting help.
Occasionally the request for the address is refused with public name already in use. The
Save itself succeeds: the address is asked for in the background afterwards. The tab stays at
Not connected…, and the error log (logs/errors.log in your C5 folder) has a line such as
could not get an address (HTTP 409): public name already in use. This will be retried in 60s.
Most of the time it is this install racing itself — two asks for the address in flight at
once, such as a Save landing while the tunnel was still coming up — and it clears on its
own: C5 asks again twice, a minute apart, and saving the Remote Control tab asks again at
once. Only a refusal that persists across saves means the name derived for this install is
held by a different install on Growther.si's side — a coincidence, not a copy — and that
needs support to resolve. See
Getting help.
The name works on mobile data but not on home Wi-Fi #
If you gave the machine a name of your own that points at a private address (one starting
192.168., 10., or 172.16. to 172.31.), many home routers refuse to resolve it. The
feature is called DNS rebinding protection, and it is doing its job: a public name resolving
to a private address is also what an attacker would set up.
Your options:
- add the name to your router's rebinding allow-list, if it has one;
- pair against the network address chip instead of the name while at home;
- use the issued address, which does not resolve to a private address at all.
C5 does not fall back to plain http:// on a network address: pairing needs a secure
(https) address.
Chat replies arrive late, or in larger pieces, through the tunnel #
Through the issued address, replies can arrive later or in larger pieces than at the
machine, because the path is longer and is not under Growther.si's control. The reply itself is
the same. On your own network, or over growther connect, it streams as it does at the
machine.
Nothing on the machine changes this. If it matters, use a direct path.
"Too many attempts" while pairing through the tunnel #
Through the tunnel, every caller reaches the machine from the same address, so the machine cannot tell one stranger's connection attempts from yours until a device is paired. Paired devices each get their own allowance; unpaired attempts — a pairing that has not finished, a refused sign-in — share one. If the address is being probed, you can hit that limit while pairing.
Wait a few minutes, or pair against a network address chip while you are on the same Wi-Fi. Once paired, the device is no longer affected.
The lease expired #
When a device's sign-in lease runs out, the phone is signed out and lands on a screen saying that the lease for this device has ended, and that someone at the machine can extend it — or the phone can be paired again. It does not get back in on its own: leases are extended only from the machine, so a phone whose lease has run out stays out until someone at the machine lets it back in.
At the machine, under Paired devices, the device shows Expired, and the pencil next to it offers to Extend the lease. Extending starts a fresh window from now, and the phone gets back in with the key it already holds — open C5 on it again; there is no code to scan and nothing to type. Use the same pencil to set a different length at the same time.
If extending is not offered, or the phone still cannot get in, one of these is true:
- the device was revoked at the machine. Pair it again;
- Remote Control was turned off at the machine, which revokes every device. Turn it on and pair again;
- the phone's stored key is gone — a cleared browser, a reinstalled app. Pair again.
The phone says Device No Longer Paired #
Someone at the machine revoked this device, turned Remote Control off, or pressed Disconnect everything. It happens the moment they do it, not the next time the phone asks: the phone's live connection is closed, and the app is replaced by this screen even if it was in use — nothing it had on screen stays usable. A phone that was offline at the time sees the same screen the next time it reaches the machine.
There is nothing to retry from the phone, and refreshing the page will not reconnect it. If it should still have access, pair it again from Settings › Remote Control in C5 on the machine, as the screen says; the key it held before is not reused. A lease that has run out shows a different screen, This Device's Lease Has Ended — see The lease expired; there, extending the lease at the machine is enough.
If the phone shows Server disconnected — retrying… instead, that is a different thing: the machine is unreachable, not the pairing gone. See The address is not reachable.
Sign-in at a remote-facing name is refused #
While Remote Control is on, every name that faces the outside accepts only a paired device: the issued address, a Public address you configured under Advanced, and the names on a certificate you supplied. A password, a PIN, a passkey, or a directory sign-in typed at one of those names is refused with a message saying to use a device you have paired, or to pair this one from C5 on the machine. This is deliberate: a shared secret that can be typed anywhere is exactly the kind of credential a public name should not take.
The same sign-in works as it always has at the machine itself and at its plain network
addresses (https://192.168…, for example).
Where to go next #
- Managing devices — revoke, leases, Disconnect everything.
- Common issues — for problems with C5 itself.
- Getting help
