2
0
mirror of https://github.com/lightninglabs/loop synced 2024-11-09 19:10:47 +00:00
loop/docs/autoloop.md
2023-08-11 11:02:03 -07:00

14 KiB

Autoloop

The Loop client contains functionality to dispatch Loop Out swaps automatically, according to a set of rules configured for your node's channels, within a budget of your choosing.

The Autoloop functionality is disabled by default, and can be enabled using the following command:

loop setparams --autoloop=true

Easy Autoloop

If you don't want to bother with setting up specific rules for each of your channels and peers you can use easy Autoloop. This mode of Autoloop requires for you to set only the overall channel balance that you don't wish to exceed on your Lightning node. For example, if you want to keep your node's total channel balance below 1 million satoshis you can set the following

loop setparams --autoloop=true --easyautoloop=true --localbalancesat=1000000

This will automatically start dispatching Loop Outs whenever you exceed a total channel balance of 1M sats. Keep in mind that on first time use this will use the default budget parameters. If you wish to configure a custom budget you can find more information in the Budget section.

In addition, if you want to control the maximum amount of fees you're willing to spend per swap, as a percentage of the total swap amount, you can use the feepercent parameter as explained in the Fees section.

Liquidity Rules

At present, Autoloop can be configured to either acquire incoming liquidity using Loop Out, or acquire outgoing liquidity using Loop In. It cannot support automated swaps in both directions. To set the type of swaps you would like to automatically dispatch, use:

loop setrule --type={in|out}

Autoloop will perform Loop Out swaps by default.

Swaps that are dispatched by Autoloop can be identified in the output of ListSwaps by their label field, which will contain: [reserved]: autoloop-out.

Even if you do not choose to enable Autoloop, we encourage you to experiment with setting the parameters described in this document because the client will log the actions that it would have taken to provide visibility into its functionality. Alternatively, the SuggestSwaps rpc (loop suggestswaps on the CLI) provides a set of swaps that Autoloop currently recommends, which you can use to manually execute swaps if you'd like.

Note that autoloop parameters and rules are not persisted, so must be set on restart. We recommend running loopd with --debuglevel=debug when using this feature.

Liquidity Targets

Autoloop can be configured to manage liquidity for individual channels, or for a peer as a whole. Peer-level liquidity management will examine the liquidity balance of all the channels you have with a peer. This differs from channel-level liquidity, where each channel's individual balance is checked. Note that if you set a liquidity rule for a peer, you cannot also set a specific rule for one of its channels.

Liquidity Thresholds

To setup Autoloop to dispatch swaps on your behalf, you need to set the liquidity balance you would like for each channel or peer. Desired liquidity balance is expressed using threshold incoming and outgoing percentages of capacity. The incoming threshold you specify indicates the minimum percentage of your capacity that you would like in incoming capacity. The outgoing threshold allows you to reserve a percentage of your balance for outgoing capacity, but may be set to zero if you are only concerned with incoming capacity.

Autoloop will perform swaps that push your incoming capacity to at least the incoming threshold you specify, while reserving at least the outgoing capacity threshold. Rules can be set as follows:

loop setrule {short channel id/ peer pubkey} --incoming_threshold={minimum % incoming} --outgoing_threshold={minimum % outgoing}

To remove a rule from consideration, its rule can simply be cleared:

loop setrule {short channel id/ peer pubkey} --clear

Fees

The amount of fees that an automatically dispatched swap consumes can be limited to a percentage of the swap amount using the fee percentage parameter:

loop setparams --feepercent={percentage of swap amount}

If you are not using Easy Autoloop and you would like finer grained control over swap fees, there are multiple fee related settings which can be used to tune Autoloop to your preference. The sections that follow explain these settings in detail. Note that these fees are expressed on a per-swap basis, rather than as an overall budget.

On-Chain Fees

When performing a successful Loop Out swap, the Loop client needs to sweep the on-chain HTLC sent by the server back into its own wallet.

Sweep Confirmation Target

To estimate the amount of on-chain fees that the swap will require, the client uses a confirmation target for the sweep - the number of blocks within which you would like this balance swept back to your wallet. The time to acquire your incoming liquidity is not dependent on sweep confirmation time, so we highly recommend setting a very large sweep confirmation target (up to 250 blocks), so that your sweep can go through with very low fees.

loop setparams --sweepconf={target in blocks}

Fee Market Awareness

The mempool often clears overnight, or on the weekends when fewer people are using chain space. This is an opportune time for Autoloop to dispatch a swap on your behalf while you sleep! Before dispatching a swap, Autoloop will get a fee estimate for your on-chain sweep transaction (using its sweepconftarget), and check it against the limit that has been configured. The sweeplimit parameter can be set to configure Autoloop to only dispatch in low-fee environments.

loop setparams --sweeplimit={limit in sat/vbyte}

Miner Fee

In the event where fees spike dramatically right after a swap is dispatched, it may not be worthwhile to proceed with the swap. The Loop client always uses the latest fee estimation to sweep your swap within the desired target, but to account for this edge case where fees dramatically spike for an extended period of time, a maximum miner fee can be set to cap the amount that will be paid for your sweep.

loop setparams --maxminer={limit in satoshis}

Server Fees

Swap Fee

The server charges a fee for facilitating swaps. Autoloop can be limited to a set swap fee, expressed as a percentage of the total swap amount, using the following command:

loop setparams --maxswapfee={percentage of swap volume}

No-Show Fee

In the case of a no-show, the server will charge a fee to recoup its on-chain costs. This value will only be charged if your client goes offline for a long period of time after the server has published an on-chain HTLC and never completes the swap, or if it decides to abort the swap due to high on-chain fees. Both of these cases are unlikely, but this value can still be capped in Autoloop.

loop setparams --maxprepay={limit in satoshis}

Off-Chain Fees

The Loop client dispatches two off-chain payments to the Loop server - one for the swap prepayment, and one for the swap itself. The amount that the client will pay in off-chain fees for each of these payments can be limited to a percentage of the payment amount using the following commands:

Prepayment routing fees:

loop setparams --maxprepayfee={percentage of prepay amount}

Swap routing fees:

loop setparams --maxroutingfee={percentage of swap amount}

Budget

Autoloop operates within a set budget, and will stop executing swaps when this budget is reached. This budget includes the fees paid to the swap server, on-chain sweep costs and off-chain routing fees. Note that the budget does not include the actual swap amount, as this balance is simply shifted from off-chain to on-chain, rather than used up.

The budget value is expressed in satoshis, and can be set using the setparams loop command:

loop setparams --autobudget={budget in satoshis}

Your Autoloop budget is refreshed based on a configurable interval. You can specify how often the budget is going to refresh by using the setparams loop command:

loop setparams --autobudgetrefreshperiod={duration}

If Autoloop has used up its budget, and you would like to top it up, you can do so by either increasing the overall budget amount, or by decreasing the refresh interval. For example, if you want to set Autoloop to have a budget of 100k sats every 168 hours, you could set the following:

loop setparams --autobudget=100000 --autobudgetrefreshperiod=168h

Dispatch Control

Configuration options are also exposed to allow you to control the rate at which swaps are automatically dispatched, and Autoloop's propensity to retry channels that have previously failed.

In-flight Limit

The number of swaps that Autoloop will dispatch at a time is controlled by the autoinflight parameter. The default value for this parameter is 1, and can be increased if you would like to perform more automated swaps simultaneously. If you have set a very high sweep target for your automatically dispatched swaps, you may want to increase this value, because Autoloop will wait for the swap to fully complete, including the sweep confirming, before it dispatches another swap.

loop setparams --autoinflight=2

Failure Backoff

Sometimes Loop Out swaps fail because they cannot find an off-chain route to the server. This may happen because there is a temporary lack of liquidity along the route, or because the peer that you need to perform a swap with simply does not have a route to the Loop server's node. These swap attempts cost you nothing, but we set a backoff period so that Autoloop will not continuously attempt to perform swaps through a very unbalanced channel that cannot facilitate a swap.

The default value for this parameter is 24 hours, and it can be updated as follows:

loop setparams --failurebackoff={backoff in seconds}

Swap Size

By default, Autoloop will execute a swap when the amount that needs to be rebalanced within a channel is equal to the swap server's minimum swap size. This means that it will dispatch swaps more regularly, and ensure that channels are not run down too far below their configured threshold. If you are willing to allow your liquidity to drop further than the minimum swap amount below your threshold, a custom minimum swap size can be set. If Autoloop is configured with a larger minimum swap size, it will allow channels to drop further below their target threshold, but will perform fewer swaps, potentially saving on fees.

loop setparams --minamt={amount in satoshis}

Swaps are also limited to the maximum swap amount advertised by the server. If you would like to reduce the size of swap that Autoloop created, this value can also be configured.

loop setparams --maxamt={amount in satoshis}

The server's current terms are provided by the loop terms CLI command. The values set for minimum and maximum swap amount must be within the range that the server supports.

Manual Swap Interaction

Autoloop will not dispatch swaps over channels that are already included in manually dispatched swaps - for Loop Out, this would mean the channel is specified in the outgoing channel swap, and for loop in the channel's peer is specified as the last hop for an ongoing swap. This check is put in place to prevent Autoloop from interfering with swaps you have created yourself.

Disqualified Swaps

There are various restrictions placed on the client's Autoloop functionality. If a channel is not eligible for a swap at present, or it does not need one based on the current set of liquidity rules, it will be listed in the Disqualified section of the output of the SuggestSwaps API. One of the following reasons will be displayed:

  • Budget not started: if the start date for your budget is in the future, no swaps will be executed until the start date is reached. See budget to update.
  • Budget elapsed: if Autoloop has elapsed the budget assigned to it for fees, this reason will be returned. See budget to update.
  • Sweep fees: this reason will be displayed if the estimated chain fee rate for sweeping a loop out swap is higher than the current limit. See sweep fees to update.
  • In-flight: there is a limit to the number of automatically dispatched swaps that the client allows. If this limit has been reached, no further swaps will be automatically dispatched until the in-flight swaps complete. See in-flight limit to update.
  • Budget insufficient: if there is not enough remaining budget for a swap, including the amount currently reserved for in-flight swaps, an insufficient reason will be displayed. This differs from budget elapsed because there is still budget remaining, just not enough to execute a specific swap.
  • Swap fee: there is a limit placed on the fee that the client will pay to the server for automatically dispatched swaps. The swap fee reason will be shown if the fees advertised by the server are too high. See swap fee to update.
  • Miner fee: if the estimated on-chain fees for a swap are too high, Autoloop will display a miner fee reason. See miner fee to update.
  • Prepay: if the no-show fee that the server will pay in the unlikely event that the client fails to complete a swap is too high, a prepay reason will be returned. See no show fees to update.
  • Backoff: if an automatically dispatched swap has recently failed for a channel, Autoloop will backoff for a period before retrying. See failure backoff to update.
  • Loop Out: if there is currently a Loop Out swap in-flight on a channel, it will not be used for automated swaps. This issue will resolve itself once the in-flight swap completes.
  • Loop In: if there is currently a Loop In swap in-flight for a peer, it will not be used for automated swaps. This will resolve itself once the swap is completed.
  • Liquidity ok: if a channel's current liquidity balance is within the bound set by the rule that it applies to, then a liquidity ok reason will be displayed to indicate that no action is required for that channel.
  • Fee insufficient: if the fees that a swap will cost are more than the percentage of total swap amount that we allow, this reason will be displayed. See fees to update this value.
  • Loop In unreachable: if the client node is unreachable by the server off-chain, this reason will be displayed. Try improving the connectivity of your node so that it is reachable by the loop server.

Further details for all of these reasons can be found in loopd's debug level logs.