Oddbean new post about | logout
 The amount of people who tell me they have force close issues on LND is staggering. Starting to think there is something seriously broken with it. Try CLN, it is a much nicer experience. channels actually stay open! I had some open for years. 
 Adding me and @bitcoinbonden as having had force close issues with LND :/ 
 RIP my sats

Lost in the void forever 😆 
 We also need to consider that CLN users are more technical. 

Anyone can run an LND node on Umbrel. Experienced and noobs alike. So you end up with a lot of RasPi nodes runs by noobs like me. 

Then I would guess LND dominates the lightning space. I wonder how many there are, but a safe bet is above 60% of nodes. 

These metrics add up. 
 I bet this has a lot to do with it. I’ve run a few different LND nodes over the years and have not experienced many force closures. If you have shitty channel partners, you’re bound to get force closed on. 
 This was happening so often that I was wondering if LND was doing force closes automatically for some reason 
 It does. Most of my force closures have been initiated by my node (not me  manually but the node automatically). I haven't figured out why it's doing it yet and there doesn't seem to be a way to stop it. 
 They are learning with mistakes  
 Maybe @niftynei knows more about this. I monitor my channel closes with core and walletnotify nostr messages, i just don’t see closes that often on CLN unless they are manually done.

Does LND close channels if the other node is offline for too long or something? Stricter validation checks? Why would this happen? 
 My raspi node crashes once a month if it doesn't restart correctly best I can do is ask for a forced close to recover my funds  
 Backups?  
 That kind of backup is precarious on lightning. Most backups will result in force-closed channels. Recovering channel state is scary due to risk of having not the most recent state and thus triggering a justice TX. 
 Wen APO 
 idk, people want CTV + stuff. Everyone pushing for it is trying to thwart anti-spam initiatives though so my social heuristics are screaming at me that it's a bad idea and I haven't figured out why yet.

I guess that puts me in camp APO. 
 What's APO and CTV?  
 AnyPrevOut and CheckTempkateVerify 
 thanks  
 Raspis are fun toys, not enterprise grade machines.  The future of money deserves better.  
 I want to learn how to run a business on LN and utilize the new features but I'm still not comfortable in how it works.  
I'd invest in enterprise grade machine if I have back ups and/or watchtowers. 

The redundancy of on chain funds makes total sense, but lightning is still confusing.  I even made $40 from someone using an old channel state no way I'd risk a lot of my own funds and especially other people's funds on it. 

Am I missing something or are all the implementations still sketchy?  
 Is this is due to the way backups are saved on each implementation? 
LND is static channel backups which can only tell your counterparty to close. 
While CLN has a way to save the latest state right? 

What's the best way to have redundancy and prevent channels closing? CLN and reference a watchtower so you don't get accused of cheating with an old state?  
 I have both nodes... CLN has less force close by "far". 
 I want to try CLN, but I'm kinda pot committed with LND now. I can't afford the onchain fees to close current LND channels and re open on a CLN node. 

Wish there was a way to swap the backend. 
 Most the Force Closures on my node were from a) The other node accepting the channel but then changing minimum channel values that then force close everything under it. b) Node is offline for too long or is not being used so they force close after a certain time of inactivity c) Human error from not understanding commands. Ill be 100% honest and say running a lightning node is a terrible experience on both CLN and LND for different reasons. 
 No issues for me using LND.  Can't remember the last time I had a forced close.  Must be 8 months or better. 
 People who talk about it seem to be surprised that it’s happening. It sounds like LND is doing it automatically without their consent? Thats why I suspected its a bug. 
 I was a CLN maxi but switched to LND because I need anchor outs... 
 CLN can easily handle anchor channels.  You just add experimental-anchors to the config.  They should probably pull back on the "experimental" thing though. It's solid. 
 Totally. IMO Lightning Labs are in cahoots with miners… 
 LND definitely has an itchy trigger finger on the force close.

Eclair is also a great option for a lightning implementation. 
 CLN has been very solid for me. I had my node go offline for prolonged periods several times, multiple Raspiblitz sd card reflashes, so on and so on and my channels stayed intact. 
 CLN FTW 
 The codebase is literal garbage. 
 Force close with LND happens when there is an open HTLC and one of the nodes go offline for some time. Then in order to clear the transation force closing happens. The problem can be the internet connection or tor, but also some bug in LND that does not deal well with HTLC in some situations. 
 Interesting, I’m not going to say CLN doesn’t have issues and poor support, but I’ve been running a CLN node on mutiny net with channels for over 250 days with no issues. 
 Why do you need hundreds of channels?  
 No, maybe 10 😂 
 This is surprising to hear! 
 I import my CLN descriptors into my core node, I haven’t seen any discrepancies. 
 have you considered forking and proof your point ?