It looks like the # of LN nodes and # of LN channels are down 8% and 10% respectively over the last 30 days according to 1ML.
I suspect much of this can be attributed to the two LND issues from the last few weeks & Zero Fee Routing’s node shutdown announcement, but am curious to know exactly why these numbers are down.
Are node operators consciously shutting down their own nodes & channels due to these bugs? Did the recent LND bugs cause LN explorers to display old/incorrect data? Is something else going on?
Any additional context into why these numbers are down is appreciated.
I thought I will not come back / post anymore on SN, but this question must be answered.
My answer is in my old article I wrote back in May, about the precarious situation of many LN nodes run by noobs, without a base knowledge of how LN works. I raised from then some alerts to LN nodes runners, devs, users in general. Some they listen, some not.
Here are my articles and I suggest to all to read them and pay attention to the details:
TLDR:
  • many make the confusion between a routing node and a simple LN node. Are not the same.
  • from plethora of new Umbrel users that was starting in 2020, many "left the building" because they were opportunists that only want to have "passive income". Slowly they realized that is not the case. Running a good routing node is not for noobs and REQUIRE a lot of knowledge. Knowledge that these opportunists do not have it and also do not have the patience to read and learn. They just want NOW.
  • from plethora of RPi nodes, many realized that these toys ARE NOT so good for running a good routing node. This end up in many hardware failures, nodes in limbo, funds in limbo, frustrations etc. Myself I was helping many of these users to recover their nodes. So in time these users were migrating to other types of LN nodes.
reply
good points, the idea that some of the hobbyist operators looking to make a quick buck have been washed out of the LN node operator ecosystem makes sense.
do you think that some of this drop in LN nodes and channels represents this class of hobbyist node operators giving up and throwing in the towel?
also, why did you think you wouldn’t be back to post on SN?
reply
The last couple of months a lot of people point out it is better to open big channels, maybe there is something related with those numbers, a user without enough funds it's almost force by the market to open a couple of big channels.
reply
how would this trend change the number of nodes on the network?
reply
I only focus in the channel part of the question :), you can see in 1ML more Network Capacity with less channels, for the number of nodes, no idea, I don't think it's related with the latests LND bugs.
reply
interesting take, you’re right that capacity is still up… but i wonder if capacity would be up even more if it weren’t for the LND bugs.
public capacity is up 3% this month, but was up 6% the month prior. i believe public capacity also dropped briefly on the day ZFR announced the shutdown of his node.
reply
if this was the case, wouldn't the number of nodes online be relatively the same, since they are just closing channels and opening larger ones?
reply
You are right, nodes down 8%, channels 10%, the something related could be only 2% :), but we are just making correlation and correlation does not imply causation https://swizec.com/blog/pirates-downfall-causes-global-warming/ maybe with more info.
reply
reply
Great question.
reply
I was running an LND node, and in the past week ALL of my channels were force closed due to the bug. ALL of my Sats are now on-chain. I have to work on re-opening the channels which I plan to do this weekend.
Really big bummer. But I'm glad the system is being hardened by discovering these bugs early, before the masses start using Lightning.
reply
interesting, were you able to figure out why every single one of your channels was forced closed when it looks like some other LND channels were unaffected? (since 90%+ of all channels from last week survived)
agree with you on the early bug discovery!
reply
That's the part I'm confused about. LND nodes make up about half of the network at least, don't they? (too lazy to go look up the exact number right now)
If everyone experienced what I did, wouldn't this be an even bigger issue?
I do have one hypothesis. There were 2 LND upgrades this week. The first upgrade fixed the main bug, but it introduced a new bug in doing so. Something which effected syncing to the graph. I don't think there was more than a 24 hour period between the first update and the emergency second one. I happened to update during that short window, and attempt to sync my node.
I think most nodes probably skipped that middle update, going straight from 0.15.2 to 0.15.4. The middle one, 0.15.3 isn't even available anymore because of how bad it was.
So, there were probably very few of us who tried upgrading during that time-window, and we're the ones making up that 10% loss in channels this past week.
I had about 25 channels too.... all gone :(
reply
What was bad about 0.15.3? My understanding was that the issues had always existed?
reply
The main issue they were trying to fix had existed. It sounded like the release with the fix included a Pull Request which introduced a new bug that affected syncing. This second bug was easy to isolate and fix, but it caused a lot of trouble for 24 hrs. I'm sure the dev who created the Pull Request is feeling pretty embarrassed this week.
Disclaimer: I didn't verify anything I'm saying here. I'm piecing together what I heard on the Simply Bitcoin podcast with my own experience with bugs & upgrades on my node this week.
reply
yeah, i believe you’re right that a majority of nodes are on LND today. where would you look to find the exact number? has anyone built a dashboard showing the % of nodes on different implementations?
interesting hypothesis, perhaps other node operators can chime in with their experience.
reply
reply
I wonder how they collected that data… or how LN implementation market share has evolved in the last two years?
reply
also zerofeerouting is shutting down his node, so a lot of people closed channels preemptively. he was like top 3 node on the network
reply
LND being too reckless.
Time to start using C-Lightning
reply
LND BUG borked all LND nodes.
reply
what does borked mean?
are you saying that node operators who haven’t upgraded to the latest lnd are not discoverable by the network? are their nodes entirely offline?
reply
borked as in just fucked it all up. the bug prevents the node from syncing the blockchain. therefore the nodes appear offline
Extremely water downed explanation
reply
can these nodes that appear offline still route payments?
reply
no the nodes appear offline, which means the channels associated with those nodes are unavailable and do not appear for routing LN payments.
and i do think zero fee routing closing down their node most likely had a material impact on those numbers, but i do believe the majority is from the bug. considering LND is the more popular LN implementation since its packages with node implementations like umbrel mynode etc.
personally i have two nodes, planning on getting Core LN running on my second one.
reply
got it, so even if a payment was being routed through only the nodes which appear offline, the payment would still fail?
is it possible to see whether the # of nodes running CLN has changed in the last couple weeks?
reply
are you asking if my node, your node and jims node were all offline and we have channels, if we could route a tx through our offline nodes?
reply
If that is the question the answer is no.
i’m asking if those nodes appeared offline, could they still route payments?
or said differently, what is the difference between a node that appears offline vs. one that actually is offline?