pull down to refresh
Yay indeed!!
You can continue the manual process, in case you gonna miss it.
But as soon umbrel launches v1.6, manual installs only survive until the next reboot. So the App was a bit of a necessity, too.
It's a little bug in LND (not LNDg) and fixed in the newest release.
Release notes 21 20.1.beta
On it, we'll get this sorted. ☂️ Umbrel needs a bit of extra TLC:
https://tunnelsats.com/faq#how-do-i-finalize-the-configuration-for-umbrel-05
Anything that runs debian Linux will work with it. The specs of the URL you shared look good enough. 4Core / 16GB. But if you're really serious, look for raid options. One faulty drive can ruin a long lifespan of a lightning node. This, a UPS and a VPS and some bitcoin for the hot-wallet, of course
What's wrong with it? Generally I have heard great things. It's just incredibly hard to get in between the portman containers, cause they take security super serious
Ahh okay hmmm I am trying to figure out if I need this or not
Haha yeah I figured. Try to have a go at the FAQ and best advice probably is: Have a chat with other runners who are already a customer. In a nutshell: If you try to become a routing node, or you're already there but hitting a ceiling, this is for you
How does this differ from umbrels setting of hybrid mode?
Umbrel hybrid - basically just the name of hybrid done by LND / CLN - connects to the Lightning Network via your home-router directly, with your home ISP address.
Obviously this results in your home-IP, location and network announced to everyone who cares.
We're not crying wolf, but if you want to avoid this as an attack vector, this is why you want a VPN as a masquerading option
Tailscale is for you accessing your node in a private tunnel, and your node responding back through that tunnel.
Tunnelsats is for the Lightning Network accessing your node in a private tunnel, and responding back both via the TS-Tunnel (and Tor = hybrid)
Makes sense?
We haven't found a way to get this implemented with start9.
As we have to get onto the network layer, to mark outgoing lightning packets and send them through the tunnel, the way how start9 secured their structure makes it quite brittle.
We haven't given up, but it's currently not possible, sorry.
lncli listpeers | jq '.peers[] | select(.pub_key=="<peer_pubkey>") | .features'
This one parses my complete existing channel partners. Is it really so rarely used, that out of 140 channels, 0 support STC?
lncli listpeers | jq '.peers[] | select(.features."24".is_known and .features."26".is_known) | {pub_key, features}'One thing to add about a script I find particuarly useful to dig into when trying to answer the most pressing question a node-runner has:
How much should I charge for my new channel to X?
The script fee_adjuster.py can be run with --debug, resulting in no write changes, but just giving a terminal output on
- fetching amboss fee data for the channel you added into
feeConfig.json - following a waterfall fee-chart on what the proposed fee-setting would be, if we accept that the market currently charges that channel correctly
This is just an excerpt of what the script can do, eg fee-bands, group discounts, stuck adjustments etc. To see that in practice, below is the terminal output for my channel to Kraken 🐙⚡
.venv/bin/python ~/tools/Lightning-Python-Tools/Other/fee_adjuster.py --debug
--------------------------------------------------------------------------------
--- Stuck Check Debug for Peer 02f1a8c876... ---
Configured stuck_time_period: 3 days
Configured min_local_balance_for_stuck_discount: 25.0%
API Check Window: 12 days
Last outbound forward timestamp found: 2025-06-12 03:26:38
Calculated days since last forward: 3
Aggregate Outbound Ratio for Peer: 63.1%
Calculated stuck bands down (after checks): 1
Final Peer Status determination for stuck logic: Active (Recent Outbound Forward)
--- End Stuck Check Debug ---
--------------------------------------------------------------------------------
Alias: Kraken 🐙⚡
Pubkey: 02f1a8c87607f415c8f22c00593002775941dea48869ce23096af27b0cfdcc0b69
Channel ID: 837308890980417537
Capacity: 10,000,000
Local Balance: 6,311,206 | (Outbound: 63.1%)
Old Fee Rate (LNDg): 786
--- Fee Calculation Waterfall ---
Start Fee (Amboss WEIGHTED_CORRECTED): 1,503.4 ppm
+/- Base Adj (+0.0%): 1,503.4 ppm
+/- Group Adj (-5.0%): 1,428.3 ppm
+/- Trend Adj (+0.0%): 1,428.3 ppm
* Fee Band Adj (-45.0%): 785.5 ppm
Max Cap (3500 ppm): Not Applied
Calculated Rate (Rounded): 786 ppm
---------------------------------
--- Context & Settings ---
Group: Low Discount | Stuck ✅ | Weighted-Corrected
Fee Base Used: weighted_corrected
Trend Factor Used: 0.00 (Sensitivity: 0.60)
Fee Bands: Enabled
- Discount: -45.0%, Premium: 20.0%
- Initial Band (Liquidity 63.1%): High Discount (60-80%)
- Stuck Adjustment Applied: -1 bands
- Final Band Used for Calc: Max Discount (80-100%)
- Resulting Adjustment: -45.0%
Stuck Adjustment: Enabled
- Period: 3 days
- Min Local Balance for Discount: 25.0%
- Days Stuck: 3 days
- Peer Status: Active (Recent Outbound Forward)
- Adjustment Applied: 1 bands down
--- Inbound Auto Fee Adjustment ---
Inbound Auto Fee Global Setting: Disabled
--- Amboss Peer Fee Data (Remote Perspective) ---
+-----------+---------+--------+--------+----------+-------------+
| Time | Max | Mean | Median | Weighted | W.Corrected |
+-----------+---------+--------+--------+----------+-------------+
| TODAY | 5000000 | 6475.7 | 500 | 4066.3 | 1503.4 |
| ONE_DAY | 5000000 | 6479.9 | 500 | 4061.5 | 1498.3 |
| ONE_WEEK | 5000000 | 6527.8 | 500 | 3990.7 | 1409.4 |
| ONE_MONTH | 1500000 | 3450.8 | 500 | 3557.8 | 1650.2 |
+-----------+---------+--------+--------+----------+-------------+Well, hard to say. I would say certainly more laid back into hammock maintenance possible, but some runners might take the extra mile to make the extra buck. It probably depends on
- your motivation + time-available
- your curiosity and desire for tinkering
If both of the above is not a personal preference, todays tools like Autofees and Auto-Rebalancer in LNDg offer a hands-off "hammock" approach.
Certainly less fire-drilling compared to 2-4 years ago.
Totally concur on your love for LNDg. I'm a big fan and also contributing to their code wherever I feel I can add value.
Yeah, oral would help, but imo not enough. We need to also stresstest What we're actually trying them to proof: Cognitive abilities, adapting to change, thinking on their heels, creative thinking, abstracting is the way to go.
Memorizing, summarizing, visualising stuff? Phew, I dunno.
Keeps reminding me of two things:
- The cab driver test in London was the hardest test. They throw two street names at you, you had to explain, street by street, turn by turn, how to get from A to B. Guess how many use that knowledge today?
- One internet meme I remember: I'd love to have a version of the olympics, where every athlete could go full bonkers on steroids and drugs. Let's see how high humans can really jump
The response got hidden, apologies for the late reply.
Wrt your question: WG tunnel doesn't handle the HTLC, but the node OS does this. The connection flow of - let's take LND as it's well documented and I'm familiar from prod experience - is that while the HTLC is inflight and the tunnel drops, two things happen:
So if all 3 are a no, your HTLC will fail after expiration time is hit (if incoming = critical). If one of them is yes, your HTLC will not fail, but close on the connection which is available to be established.
Does this answer your question?