Motivation is a tricky beast. There have been times where I read every single log line as it flew by, trying to understand the details, optimizing stuff. There have been times (more recently) where I simply don't look at any information of the node, aside from "is it still running". Nowadays, I just do what feels right, without any plan. If my motivation drops too low, I'll close the node. Luckily, there's a lot of automation going on behind the scenes, so my motivation would need to drop quite a lot.

I stopped opening channels to new peers a while ago, mainly because I haven't figured out how to find good peers. I experimented quite a bit over the years, sometimes with success, but not in a satisfactory and organized manner. Due to the size/popularity of my node, I get lots of channels from (for me) new peers, and I just keep those that are worth it. I need to take good care of my sats, and having them sit idle in useless channels isn't worth it - which is why the number of channels might be rather low. The channels that remain, however, are good (or new).

If I take a closer look at the top channels while including bidirection traffic (i.e. incoming traffic also counts as "good" if it leaves through a channel where I charge fees), the top nodes for the past 60 days are, sorted by profitability with the best peer listed first:

I guess you could say that larger nodes are better. Keep in mind, though, that not all large nodes are good peers. I've cut some ties, and I will cut more.

Regarding profitability: keep in mind that this involves quite a bit of risk (hot wallet, "beta" software). Furthermore, running a routing node takes quite a bit of skill. If this were an established industry, my salary would be more than what any reasonable node could earn - risk and other costs aside. It's a hobby. I'm fine with smaller nodes as peers, but ultimately, they tend to hog my sats more than larger nodes do.

I closed a few channels with bfx-lnd0 with almost all of the liquidity on their side. Having those parallel channels doesn't serve any benefit for me, I still have plenty of inbound liquidity from bfx-lnd0. This way I could get a few sats (to be used elsewhere) while reducing the risk of force closes and other channel related costs at a time with very low on-chain fees.

I'm using the rating system implemented in lnd-manageJ, see https://github.com/C-Otto/lnd-manageJ/blob/main/rating.md for details. If a peer has a very low rating (ignoring the first 45 days of a new channel), I consider closing.

Yeah. I think a lot of this is due to private node runners establishing nodes, doing some rebalancing (at a loss). I'm not sure, though.

22-23 was the 2nd wave of noobs getting into umbrels...

Performance is fine. Shortly before the new compact database format was introduced I had issues because of the database file size, but this has been resolved years ago. I think ZFR would have worked with this updated version of lnd.

As far as I know you can't switch implementations while retaining your channels and/or node ID. I'd rather not start from scratch, a lot of how LN works is reputation based.

I stopped the old server, copied the files using scp, edited the name and started the new server. During that I made extremely sure that the old copy cannot be started anymore (intentionally breaking stuff like deleting the lnd binary). Once the new server was up and running, I deleted all traces of the old server on the old host. That's it.

Yes.

See #922319