More force closes in the meantime, one even happened for a pending htlc while both me and my peer were online... but the node is still active, which is good!
*click* for more
Lots happened in the meantime, and lots I didn't update in here: cool new features of my script, issues here and there, force closes (oh those damn, pesky, stinky force closes, I almost decided to quit), and I'll probably go on not updating this, but feel free to holler at me on Telegram for anything!
GordianWatch service has been launched! For now free and for my channel peers only, will notify you as soon as you are seen offline by my node, ask me for details.
Gordian is now hybrid!
Gordian was chosen by clboss as a channel peer for SethForPrivacy, overall a cool experience to realize that an algo examined my node over a period of time and finally decided to throw liquidity at it.
ZeroFeeRouting is another cool operator to look for! There was an issue with a script they were using that caused a force close of our channel, even without a pending HTLC, due to me being offline for about 12hrs straight due to a power issue. I planned to re-open to them no matter what, but this makes me all the happier to do so!
Gordian surpassed the 30 channels mark, and the 2.1M capacity streak was broken by an incoming 3M channel from South Korea (very cool operator, thank you!)
I decided to enable max_htlc
for all my channels, and increase CLTV delta to reduce risk of force closures (I'll take the slightly lower chance of being selected for routes, safety above all), plus I reduced the updatechanpolicy
frequency from every 3hrs to every 6.
Second friday report in the green, costs are at 96% of total income after a net month of activity, I am still looking into extending my reach into the network to be selected in more paths.
It's like LN was a malicious lover, welcoming me by giving a lot of sudden affection (in the form of profitability in a record timeframe), only to throw me down with disdain (as I'm hardly routing anything now, while activity was way more in the past week, even at higher fees than now). My script works without sentiment, as it only knows how to scan for flow, and won't take it personally; please have patience, Gordian's peers. On a side note, there's that one peer whose node really likes to disable our channel, let's hope we find a way to solve that.
Gordian is profitable! The accounting shows 10sats of net profit over the life of the node, which has been quite a short life, as first channel dates back to 21st march, all fees were frozen at 2500ppm for the first 6 days until I got to 11 peers and started the fee autotuning script, 12 days offline starting 4th april because of filesystem failure, this means the node became profitable in less than 20 days, is it a record or what?
I need to mark the date: I just received an unsolicited inbound channel, of the just right capacity, 2.1M, by GirthMaul, this is my very first spontaneus inbound channel and I am really happy!
G-Spot, albeit arguably expensive, is ran by a very dependable operator.
First month after launch, minus 12 days of downtime, minus first 6 days of very high default fees while waiting to get critical edge mass, it has now been 12 days of "proper routing", and I am at 156% costs vs earned fees, break even here I come!
I lost a very good peer today! Goes to show that even if you are an excellent system operator, rm -rf
is a great leveler waiting for you just around the corner; I wish them to get back up ASAP.
Gordian is back up and running! just one casualty for a force close after a few days of being offline; luckily the postgresql dump I had was fine, ended up using the main instance dump even if the replica looked "newer" (also enabled synchronous replication this time): lightningnetwork/lnd#6395
Blockchain is fully synced, ETA for LN node reboot is 15th or 16th.
Blockchain sync at 70% roughly, system has been hardened even more than before, with solid zswap setting, new leaner power SSD's (Kingston A400's, slower for desktop usage, perfect for node operations), more stable (I hope) and definitely better looking SATA adapters, and new USB3 hub incoming.
Gordian is offline. Some mean errors with Raid1 setup, not exactly troubleshooted yet (most probably a silly power delivery issue to the disks, that started a chain reaction of R/W errors which didn't stop even after fixing power lines), led to system instability and freezes. New USB3 powered HUB and new USB-SATA adapters are incoming, and tests will be done right after; worst case scenario, I'll also need to swap out SSD's, but this will be clear after new components start arriving. I beg my pardon to all my current peers, please do not close channels to me, as my postgresql database is working and tested, last time I had to cold reboot the system I made sure I could connect correctly to the network, I stopped accepting forwards with BOS, and then shut down LND gracefully. I also have the postgres replica sitting in another ext4 disk that's not been touched by the failures, and I have tested and verified working a procedure to dump the postegresl and import it in another LND instance successfully on playground signet.
Reached 12 edges to 11 peers, enough to be palatable as a routing node, so my scripts have been unleashed and fee tuner + rebalancer are now running
Opened my very first reciprocal channels with a peer, too bad it couldn't be a dualfunded channel, as being without inbound I couldn't receive the initiating keysend
Fully automated lnd
node, launched on mainnet on 2022-03-21, after months of testing my own scripts on Plebnet Playground.
- https://amboss.space/node/025ea01268089a458b77330c341e8f608e9a3002fa41a75075d1e0c4111e0e98c7
- https://1ml.com/node/025ea01268089a458b77330c341e8f608e9a3002fa41a75075d1e0c4111e0e98c7
looking for as many 2.1M balanced edges with cherry-picked, like-minded and strategically-placed operators as possible (or until I allocate all available funds)starting plan was to allocate 1M per channel on my side, with minimum 2M capacity; as of now, I want 2100000 channels to become my trademark!- I probably reached the target total capacity by now, will consider new channel opens on my side without as much random dualfunded ones
Me! Holler on telegram if you are cool and friendly.
lnd.conf
is set to open at 2500ppm and 1.199base- fees will be slowly tuned down to search for flow (or back up again to shape it)
max_htlc
and9999basefee
are NOT active anymore, to examine failures for policy updates, fees are only tuned for flow, not for local balance- rebalances run frequently and automatically, as long as they are profitable
- multiple channels with same peer are transparently managed by my script as if they were one channel
For fun! And to rebel against Gvt's holding money hostage with banks' complicity! What else?! Oh, also free sprinkling magic sats coming over lightning through the intertubes.
Happening as you read this!
- you will have to be patient while my autofee script does its thing in search for flow
- script tries to set absolute minimum fees that cover open/close costs, but this will be lifted in case of very little flow
- tor only, at least for now; if and when profit is enough I will consider hybrid mode with a VPN
- internet connection is not datacenter-grade, but top-tier-home-grade-stable (see above notice about profit for 3G redundancy)
- I perfectly understand the logic behind girrrthy channels, yet I am not a whale and have my own logic which I hope will pay off, so I have to spread my sats thin!
Node has been hardened as best as I could, featuring:
Raspberry Pi 4 8GB, 2x Kingston A400 SSD's (1Tb + 256GB) + 1x cheap 128GB SSD, on an externally powered USB3 hub- i5-6300u 8GB laptop, 256GB Intel M2 + 960GB Kingston A400 Sata SSD's
- btrfs RAID1
- zswap enabled
- posgresql backend
synchronous replication - encrypted data partitions
- encrypted cloud backups
- custom on/offline notification system
DIY UPS with voltage monitoring via serial interface, stopping services when battery is getting low, and restarting them when AC is back up and battery is charged enough- laptop battery is a great integrated UPS that also preserves its life thanks to linux power management
- notification watchdogs for system load, RAM usage, CPU temperature, SSD health
- SSH login via hardware key
- I have NO watchtowers whatsoever! (wink wink)