High onchain fee environment

Notes on using the Lightning Network when miner fees are high.

Preparation

Recommendations to get ready for a high fee environment beforehand.

Channel management

  • Open channels during low fee times strategically (weekends)

  • Close inactive and unreliable channels early during low fee times

  • Minimise downtime and instability if running a routing node

  • Use private (unannounced) channels as a spending node so downtimes won't make the peer force-close

Wallet management

  • Funding from a single sig wallet is cheaper

  • Prepare a selection of good size UTXOs for minimal cost channel opens

    • consolidate (beware of privacy implications)

    • a well funded and long running JoinMarket Maker wallet will have different sizes of coinjoined outputs available

Configuration

General

  • maximise uptime

  • configure a hybrid connection if possible - note that skipping to always use a proxy will expose your IP address (use a VPN like Tunnelsats)

LND

  • check the options in the sample lnd.conf

  • Use Anchor Commitments

    • on by default on LND when both peers support anchors

    • will have 100000 sats reserved in the onchain wallet of LND to pay the closing fee with CPFP

    • raise the max-commit-fee-rate-anchors to a comfortably high level to avoid purging the transactions from the mempools.

  • Set the minchansize (eg. avoid <500k channels on a routing node)

  • set a long payments-expiration-grace-period

  • increase the CLTV delta: bitcoin.timelockdelta from the default 80

  • increase the smallest HTLC the node is willing to send out (in millisatoshi)

  • increase routing fees (applies to new channels only)

    [Application Options]
minchansize=500000
max-commit-fee-rate-anchors=100
payments-expiration-grace-period=10000h

[Bitcoin]
bitcoin.timelockdelta=144
bitcoin.minhtlcout=100000
bitcoin.basefee=1000
bitcoin.feerate=2500

[tor]
tor.skip-proxy-for-clearnet-targets=true

  • consider increasing the fees towards peers not using anchor commitments:

    lncli listchannels | jq '.channels[] | {remote_pubkey: .remote_pubkey, commitment_type: .commitment_type}'

  • set the base_fee_msat, fee_rate_ppm, min_htlc_msat and time_lock_delta on existing channels

    lncli updatechanpolicy --base_fee_msat=1000 --fee_rate_ppm=2500 --min_htlc_msat=100000 --time_lock_delta=144

CLN

  • see the possible config options: https://github.com/raspiblitz/raspiblitz/blob/dev/FAQ.cl.md#all-possible-config-options

  • cln config settings for new channels:

    min-capacity-sat=500000
cltv-final=144
fee-base=1000
fee-per-satoshi=2500
htlc-minimum-msat=100000

always-use-proxy=true

Routing fees and balances

  • All nodes:

    • the channel balance will appear to be smaller as the commitment reserve will be higher

    • offchain transaction fees will also increase (remains to be proportional to the payment amount)

    • payment failures might appear more often as liquidity dries up

  • Routing nodes:

  • Offchain traffic will increase

  • During high fee periods less channels will be opened

  • Less capital will be reallocated

  • Submarine Swaps get more expensive (requires on-chain transactions)

  • Overall channels will get out of balance quicker

Opening channels

  • Include a change output to be able to use CPFP to bump the fee of the channel open transaction

  • Batch opens

    • the biggest saving is when using 1 input to open multiple channels

    • aim for next block with the opening transaction fee to prevent the fees running away

    • can use PSBT-s (even from ean external wallet) with the command line tools available:

    • avoid opening public and private (unannounced) channels in the same batch - defeats the purpose of not announcing channels in the gossip

Do not leave a channel open pending for more than 2016 blocks (~2 weeks)

A pending channel will become "stale" after 2016 blocks - the peer will forget about the funding transaction, so the channel will never become online.

Closing channels

  • Aim for a cooperative close if must

    • can use CPFP from the receiving wallet if a pending ccooperative close is stuck in the mempool with a low fee

  • Run the channel close command again if the transaction has been removed from the mempool

    lncli closechannel FUNDING_TXID INDEX

  • Force closes are ~5x more expensive than the next block fee at the last update

    • LND updates every 10 min on an online channel

    • Long inactive channels are a liability - especially if was online last at low miner fee period

  • Avoid and prevent force closes by minimizing downtime and instability as a routing node

Watchtowers

  • If using watchtowers need to set the

    wtclient.sweep-fee-rate= in the lnd.conf

    to a sat/byte level where it can confirm within the CSV delay in case a breach transaction is broadcasted by the peer while the node is offline.

  • The CSV delay can be set to be longer with:

    lncli updatechanpolicy

Future improvements

  • Anchor commitments by default (affects only new channels and both peers need to support the feature)

  • Splicing and dual funding - extend the channel capacity in a single transaction

  • Taproot - can save on sending from the multisig (~26 byte from the min 140 bytes)

  • Taproot - funding from multisig wallets will be the same cost as from single sig wallets (slightly more expensive for single-sig)

  • ELTOO - multiparty channels and channel factories

References

PreviousConnecting to a lightning node remotelyNextCompact the LND database (channel.db)

Last updated

Was this helpful?