Core Lightning Connection Failure

Bitcoin
1

Sorry Paul , this is out of the blue and is far from the current forum discussions.

I received information from OCEAN to upgrade to the current version of Core Lightning as they said that is why payments were delayed.

I upgraded to ver. 25.12.1~2. Now the UI is reporting “Could not establish a connection to node”

The log report:

2026-09-01T08:57:33-06:00 [ERROR - 2026-09-01T14:57:33.422Z]: WebSocket error: {}
2026-09-01T08:57:33-06:00 [ERROR - 2026-09-01T14:57:33.423Z]: WebSocket is closed at 2026-09-01T14:57:33.423Z
2026-09-01T08:57:33-06:00 {“level”:“error”,“message”:“\tCommando lightning error from listnodes command”,“timestamp”:“2026-09-01 14:57:33:5733”}
2026-09-01T08:57:33-06:00 {“level”:“error”,“message”:“\t{"code":2,"message":"Could not establish a connection to node"}”,“timestamp”:“2026-09-01 14:57:33:5733”}

LND still shows the channels and balances (but shows channels offline) but the Core Lightning UI wont open.

I looked for info on this error but I don’t see anything in particular regarding the current version.

1- Is there a cli I can run to diagnose this issue?
2- Should I try and restart the server?

Option 2 is problematic. At times when the Server One is rebooted the OS (3.5.1) sometimes disappears and has to be recovered. I do have a backup of the server from June 11, 2026 BUT this was before BIP-110 and OCEAN delayed payments because of Dual Tides. As of this morning it appears OCEAN has paid out and there are no reported errors from their payout.

I went to the CLN github but I didn’t see any discussion about this error. There was an error similar back in 24? I think but nothing current. Start9 forums has nothing either.

If it wasn’t 490,000 sats I could take the loss but I’m stressing just a bit.

Do you have any ideas? If I have to run command lines on the node how would I change into the node proper from the Mac terminal?

Sorry for being so ignorant

2

I’m no expert, but I did note that you mentioned the channels showed but were not Active. I’ve experienced this myself and was able to make them active by reconnecting the relevant peers under the Lightning-Channels/Peers-Peers and selecting Reconnect from the Actions using RTL (you can connect to both LND and CL implementations). Sorry if I’ve misunderstood your issue, but I figured it was worth the mention.

1 Like
3

I had updated all of my StartOS 0.3.5.1 servers to 0.4.0.1, so scrambling to find a server I can tinker with to get to the cli for Core Lightning. I’ve installed one on VirtualBox, but I’ll have to wait for it to finish the IBD. Will see if I can get those commands once it is finished, if you don’t solve the problem before then.

4

I just saw this:

and this is the email I received from OCEAN:

Hi,

Dual-TIDES has ended and lightning payouts just got enabled again today!

The reason why we had an extra delay for enabling lightning payouts is because some critical vulnerabilities were found in core lightning and we had to do a proper backend upgrade to avoid disruption later.

If you are running a core lighting node, it’s strongly advised to update to the latest version.

Apologies for the inconvenience that interrupted lightning payouts has caused.

Let me know if you have more questions,

Léo.

I did not see (follow the link provided) this…I simply updated via the marketplace. Current version installed is 25.12.1~2

I suppose I should just shut down the node until the release is in the market place? And I take it I cant retro grade to the previous version?

5

That announcement is completely tone deaf. They release native binaries but no source code, to ensure everyone updates before AI can see the code. But they don’t release Start9 packages or Umbrel store updates? Do they not realize that the vast majority of their users are running their software on Start9 and Umbrel, not natively?

6

Yep. Been a bad week for me evidentially. Fell for a scam and now this. Enough to make an ole man have a heart attack.

I do have a backup but I don’t think I can restore a single service?

And yes…Im about done with OCEAN. Luke has left and I just do not have great confidence in them now.

7

They are a single point of failure for the Datum protocol anyway, with Datum Prime being closed source. I don’t know if Luke’s new pool will open souce it (I’ll check when I have some time). If not, that is one of the big projects I have upcoming (to create an open source alternative to Datum Prime, so more pools can use it and become a fallback if Ocean fails).

1 Like
8

Ok, so you own both ends of these channels, correct? That gives us more options (because you can control the closure from the LND side if it is still working). Then we would just need to get the keys to the CLN base layer wallet to move the sats (those should be in the backup). Before you do anything, let it sit for now, and I’ll recreate the same scenario on my end once the IBD finishes, and practice accessing the keys (either from the cli or from the backup itself).

9

Yep. Own both ends. I’ll just go into a holding pattern as suggested.

Most appreciated Paul.

10

When I upgraded to 0.4.0 on my Pure Server all services where deleted. There where 3 options when I booted from the USB. I must have selected a new install? Not a huge deal with that server because it was just a mining server.

With the Server One that is still on 0.3.5.1 that is where all my lightning channels reside.

Its my understanding I can leave all critical services installed when upgrading to 0.4.0.

I can take steps to upgrade the server one…so question…which option should I chose when upgrading? I cannot remember exactly what the 3 options were when updating. Hate like hell to lose the 1.9million sats on those lighting service.

It is my understanding making a backup of the server one before upgrading will be only for 0.3.5.1 so that is little help for me.

So…I have no problem upgrading the Server One…but like I say…losing the 1.9m sats would be a gut punch. A month of mining down the drain basically.

I asked because I don’t want to waste your time working on the issue as you have more pressing issues.

11

The safest bet is to move all the sats off the nodes before upgrading (we just need to get access to those keys for Core Lightning). You could upgrade, but if the upgrade fails (which is a higher risk since one of your services appears to be in a bad state), you would need to reinstall 0.3.5.1 and restore them from backup, which is doable but rather nerve racking.

BTW, solving this type of problem is what I’m hoping to make into my main source of income in the future, so it isn’t a waste of my time at all.

1 Like
12

And I don’t need that.

OK. I’ll refrain until you nail down the process. Nothing that needs to be done right-away as everything seems secure for the moment. I’m in standby mode.

Talk soon. :+1:

1 Like
13

I was able to reproduce this problem and identify the cause. This particular Core Lightning package accidentally starts the node in offline mode.

That causes both symptoms you are seeing:

  • The Core Lightning dashboard cannot connect to the node and shows the “Could not establish a connection to node” / WebSocket error.
  • Your Lightning channels appear offline because the node is refusing network connections.

This appears to be a bug in this Core Lightning package version, rather than corruption or damage to your node.

Step 1

Connect to your Start9 from your Mac

Open the Terminal application on your Mac.
If you have never used SSH with your Start9 before, start with this article to set it up:
https://docs.start9.com/0.3.5.x/user-manual/ssh.html

Then connect with:

ssh start9@YOUR-SERVER-NAME.local

Replace YOUR-SERVER-NAME with the name of your Start9 server.

Step 2

Confirm that Core Lightning and your funds are still there

Run:

sudo podman exec c-lightning.embassy lightning-cli getinfo

Then:

sudo podman exec c-lightning.embassy lightning-cli listfunds

And:

sudo podman exec c-lightning.embassy lightning-cli listpeerchannels

These let you confirm that:

  • Core Lightning is running.
  • Your wallet and channels are still present.
  • Your expected funds are still accounted for.

Step 3

Confirm that you have the affected version

Run:

sudo podman images | grep lightning

You should see the image:

start9/c-lightning/main 25.12.1.2

Now check for the bug:

sudo podman run --rm --entrypoint sh docker.io/start9/c-lightning/main:25.12.1.2 \
  -c 'grep "lightningd --database-upgrade" /usr/local/bin/docker_entrypoint.sh'

The line it prints should end with:

–offline

That --offline setting is the problem.

Step 4

Make a backup of the original package image

Before changing anything, save a copy of the original:

sudo podman tag docker.io/start9/c-lightning/main:25.12.1.2 localhost/c-lightning-orig:25.12.1.2

This gives us an easy way to undo the patch if necessary.

Step 5

Remove the incorrect --offline setting

Run these commands:

mkdir -p /tmp/clnfix && cd /tmp/clnfix

Then:

printf 'FROM docker.io/start9/c-lightning/main:25.12.1.2\nRUN sed -i "s/ --offline / /g" /usr/local/bin/docker_entrypoint.sh\n' > Containerfile

Then build the corrected image:

sudo podman build -t docker.io/start9/c-lightning/main:25.12.1.2 .

This keeps the same package name/version, but removes the accidental --offline option.

Step 6

Restart Core Lightning

Restart only Core Lightning:

sudo start-cli package restart c-lightning

Give it about a minute to start.

Then check that its network ports are listening:

sudo podman exec c-lightning.embassy sh -c 'ss -tlnp | grep -E "4269|9735"'

You can also confirm that Core Lightning is no longer running with --offline:

sudo podman exec c-lightning.embassy sh -c 'cat /proc/$(pgrep -f "lightningd --database" | head -1)/cmdline | tr "\0" " "; echo'

The resulting lightningd command should NOT contain:

–offline

Step 7

Refresh the dashboard.

Refresh the Core Lightning page in StartOS.

The dashboard should now connect normally.

Your Lightning peers and channels may take a few minutes to reconnect and return to online/active status.

You may still see a log message similar to:

APP_CONNECT is defaulting to 'COMMANDO'

That message is harmless and is not the problem we are fixing.

NOTE: If you ever need to undo the patch

To undo the patch:

sudo podman tag localhost/c-lightning-orig:25.12.1.2 docker.io/start9/c-lightning/main:25.12.1.2

Then:

sudo start-cli package restart c-lightning

That restores the original package image.