Beat [3/4]: prepare resolvers to handle the `blockbeat` #9276

yyforyongyu · 2024-11-18T06:07:42Z

Depends on

Beat [2/4]: implement blockbeat #8894

NOTE: itest is fixed in the final PR

Beat itest [3/3]: fix all itest flakes #9260

Turns out mounting blockbeat in ChannelArbitrator can be quite challenging (unit tests, itest, etc). This PR attempts to implement it in hopefully the least disruptive way - only chainWatcher implements Consumer, and the contract resolvers are kept stateless (in terms of blocks). The main changes are,

contract resolvers are refactored - the original Resolve method is broken into two steps: 1) Launch the resolver, which handles sending the sweep request, and 2) Resolve the resolver, which handles monitoring the spending of the output.
chainWatcher implements Consumer in the following PR.

Alternatives

The original attempt is to make the resolvers subscribe to a blockbeat chan, as implemented in #8717. The second attempt is to make the resolvers also blockbeat Consumer, as implemented here.

This third approach is chosen as 1) it greatly limits the scope otherwise a bigger refactor of channel arbitrator may be needed, and 2) the resolvers can be made stateless in terms of blocks, and be fully managed by the channel arbitrator. In other words, when a new block arrives, the channel arbitrator decides whether to launch the resolvers or not, so the resolvers themselves don't need this block info.

In fact, there are only two resolvers that subscribe to blocks, the incoming contest resolver, which uses block height to decide whether to give up resolving an expired incoming htlc; and the outgoing contest resolver, which uses the block height to choose to transform itself into a timeout resolver. IMO if we can remove the inheritance pattern used in contest resolver -> time/success resolver and manage to transform resolvers in channel arbitrator, we can further remove those two block subscriptions. As for now, we can leave them there as they have little impact on the block consumption order enforced by the blockbeat.

coderabbitai · 2024-11-18T06:07:49Z

Important

Review skipped

Auto reviews are limited to specific labels.

🏷️ Labels to auto review (1)

llm-review

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Thank you for using CodeRabbit. We offer it for free to the OSS community and would appreciate your support in helping us grow. If you find it useful, would you consider giving us a shout-out on your favorite social media?

❤️ Share

🪧 Tips

Chat

There are 3 ways to chat with CodeRabbit:

Review comments: Directly reply to a review comment made by CodeRabbit. Example:
- I pushed a fix in commit <commit_id>, please review it.
- Generate unit testing code for this file.
- Open a follow-up GitHub issue for this discussion.
Files and specific lines of code (under the "Files changed" tab): Tag @coderabbitai in a new review comment at the desired location with your query. Examples:
- @coderabbitai generate unit testing code for this file.
- @coderabbitai modularize this function.
PR comments: Tag @coderabbitai in a new PR comment to ask questions about the PR branch. For the best results, please provide a very specific query, as very limited context is provided in this mode. Examples:
- @coderabbitai gather interesting stats about this repository and render them as a table. Additionally, render a pie chart showing the language distribution in the codebase.
- @coderabbitai read src/utils.ts and generate unit testing code.
- @coderabbitai read the files in the src/scheduler package and generate a class diagram using mermaid and a README in the markdown format.
- @coderabbitai help me debug CodeRabbit configuration file.

Note: Be mindful of the bot's finite context window. It's strongly recommended to break down tasks such as reading entire modules into smaller chunks. For a focused discussion, use review comments to chat about specific files and their changes, instead of using the PR comments.

CodeRabbit Commands (Invoked using PR comments)

@coderabbitai pause to pause the reviews on a PR.
@coderabbitai resume to resume the paused reviews.
@coderabbitai review to trigger an incremental review. This is useful when automatic reviews are disabled for the repository.
@coderabbitai full review to do a full review from scratch and review all the files again.
@coderabbitai summary to regenerate the summary of the PR.
@coderabbitai resolve resolve all the CodeRabbit review comments.
@coderabbitai configuration to show the current CodeRabbit configuration for the repository.
@coderabbitai help to get help.

Other keywords and placeholders

Add @coderabbitai ignore anywhere in the PR description to prevent this PR from being reviewed.
Add @coderabbitai summary to generate the high-level summary at a specific location in the PR description.
Add @coderabbitai anywhere in the PR title to generate the title automatically.

CodeRabbit Configuration File (`.coderabbit.yaml`)

You can programmatically configure CodeRabbit by adding a .coderabbit.yaml file to the root of your repository.
Please see the configuration documentation for more information.
If your editor has YAML language server enabled, you can add the path at the top of this file to enable auto-completion and validation: # yaml-language-server: $schema=https://coderabbit.ai/integrations/schema.v2.json

Documentation and Community

Visit our Documentation for detailed information on how to use CodeRabbit.
Join our Discord Community to get help, request features, and share feedback.
Follow us on X/Twitter for updates and announcements.

github-actions · 2024-11-18T06:22:22Z

Pull reviewers stats

Stats of the last 30 days for lnd:

User	Total reviews	Time to review	Total comments
guggero 🥇	25 ▀▀▀▀	15h 44m	30 ▀▀
yyforyongyu 🥈	12 ▀▀	1d 8h 24m ▀	33 ▀▀▀
ellemouton 🥉	8 ▀	16h 51m	21 ▀▀
ziggie1984	5 ▀	16h 15m	9 ▀
ProofOfKeags	4 ▀	5d 4h 23m ▀▀	14 ▀
bhandras	3	1d 14h 22m ▀	2
dstadulis	3	1h 26m	5
Roasbeef	2	4d 13h 56m ▀▀	9 ▀
ViktorTigerstrom	2	2d 14h 26m ▀	5
alexbosworth	1	4d 11h 58m ▀▀	1
ffranr	1	2h 9m	0
bitromortac	1	16h 16m	0
saubyk	1	17h 10m	0

Roasbeef · 2024-11-22T00:21:38Z

IMO if we can remove the inheritance pattern used in contest resolver -> time/success resolver and manage to transform resolvers in channel arbitrator, we can further remove those two block subscriptions. As for now, we can leave them there as they have little impact on the block consumption order enforced by the blockbeat.

Minor point, it it isn't inheritance, it's composition, with the inner resolver replacing the outer one. Go technically doesn't have inheritance at all.

Roasbeef

I didn't review the first two PRs, so one thing I'm missing context on is: why dees Resolve need to split up in order to be incorporated into the new blockbeat architecture?

Is that that since we don't know which ones will be launched ahead of time, we can't set up the blockbeat consumer DAG upfront?

Roasbeef · 2024-11-22T00:22:46Z

contractcourt/contract_resolver.go

+		prefix)
+
+	// If the ShortChanID is empty, we use the ChannelPoint instead.
+	if r.ShortChanID.IsDefault() {


But the scid isn't used above?

yikes yeah removed since SCID isn't that useful for lookups

contractcourt/htlc_timeout_resolver.go

Roasbeef · 2024-11-22T00:41:48Z

contractcourt/htlc_success_resolver.go

@@ -304,23 +247,14 @@ func (h *htlcSuccessResolver) resolveRemoteCommitOutput() (
 		return nil, err
 	}

-	// Once the transaction has received a sufficient number of
-	// confirmations, we'll mark ourselves as fully resolved and exit.
-	h.resolved = true


Is this attribute no longer used? IIRC, it has a role for the subset of outputs that are still sent to the nursery.

yeah it's moved into checkpointClaim

Roasbeef · 2024-11-22T00:42:26Z

contractcourt/htlc_success_resolver.go

@@ -373,6 +307,7 @@ func (h *htlcSuccessResolver) checkpointClaim(spendTx *chainhash.Hash,
 	}

 	// Finally, we checkpoint the resolver with our report(s).
+	h.resolved = true


Ah nvm I see it here restored. Though IIRC, with the current control flow, we shouldn't do this unconditionally.

yeah the slight difference is we put it in checkpointClaim, after we've done NotifyFinalHtlcEvent, and move it closer to Checkpoint as it's where we write to disk.

Roasbeef · 2024-11-22T00:44:52Z

contractcourt/htlc_timeout_resolver.go

-		return err
-	}
-
-	h.outputIncubating = true


Same comment here re this state value. I do think it's possible to make the resolvers 100% stateless, by relying on chain notifications to figure out if something has been mined or not. The nursery might need to be queried distinctly though .

Yes! I also realized this after reviewing #9062. What's more, because the sweeper now handles re-offered inputs by first checking the db and mempool to grab the existing tx, it means it's safe to re-offer the same inputs during startup. And since deep down the nursery also uses the sweeper, it means we can easily make all the resolvers stateless.

Atm another usage of the outputIncubating is to decide whether we should resolve the stage one or stage two HTLC, which can also be changed to rely on chain notifications. So the changes would be something like,

subscribe the htlc tx spending at the very beginning when initializing the resolver

when Resolve it, if the htlc tx is already spent, subscribe to the output of the htlc tx

if the output is spent, we are done here

otherwise, wait for the spend to resolve the output spent.

Basically we just outsource the info provided by outputIncubating to chain notification, further simplify the resolvers here. The only part I'm not sure is for neutrino backend, since there's no mempool, it can only rely on the sweeper's store to decide whether an input is new or not, which means we may need to store more info there (currently only the txid).

I actually plan to do this after the blockbeat is in to limit the scope, rough idea is 1) refactor a bit to get rid of outputIncubating; 2) do a db migration to remove the outputIncubating (maybe not needed?); 3) completely remove the nursery! WDYT?

Roasbeef · 2024-11-22T01:10:04Z

contractcourt/commit_sweep_resolver.go

-//
-// NOTE: Part of the ContractResolver interface.
-func (c *commitSweepResolver) IsResolved() bool {
-	return c.resolved


I think we can move this method to the contract kit, that way we hide the underlying atomic details.

Roasbeef · 2024-11-22T01:10:32Z

contractcourt/htlc_incoming_contest_resolver.go

@@ -204,7 +204,7 @@ func (h *htlcIncomingContestResolver) Resolve() (ContractResolver, error) {
 		log.Infof("%T(%v): HTLC has timed out (expiry=%v, height=%v), "+
 			"abandoning", h, h.htlcResolution.ClaimOutpoint,
 			h.htlcExpiry, currentHeight)
-		h.resolved = true
+		h.resolved.Store(true)


MarkResolved?

Roasbeef · 2024-11-22T01:11:07Z

contractcourt/htlc_timeout_resolver.go

@@ -1294,13 +1294,13 @@ func (h *htlcTimeoutResolver) resolveTimeoutTxOutput(op wire.OutPoint) error {
 // Launch creates an input based on the details of the outgoing htlc resolution
 // and offers it to the sweeper.
 func (h *htlcTimeoutResolver) Launch() error {
-	if h.launched {
+	if h.launched.Load() {


Same here re read/write methods to encapsulate.

yeah I like it, done!

Roasbeef · 2024-11-22T01:14:21Z

contractcourt/channel_arbitrator.go

 	c.activeResolversLock.Lock()
 	c.activeResolvers = resolvers
 	c.activeResolversLock.Unlock()

+	// Launch all resolvers.
+	c.launchResolvers()


In a recent PR, we eliminated a circular waiting condition related to tap channels (in the litd context). With this method now blocking to launch resolvers, we'll need to make sure we don't reintroduce such a livelock.

do we know why it was blocking? In this version the Launch only sends the input to the sweeper, and the waiting for spend is handled in the Resolve.

Roasbeef · 2024-11-22T01:16:39Z

contractcourt/channel_arbitrator.go

 	var reports []*ContractReport
-	for _, resolver := range c.activeResolvers {
+	for _, resolver := range c.resolvers() {


Where was the race? At a glance, everything looks to be mutex proteted.

hmm looks like it no longer existed due to the fix of resolved and launched, now removed.

yyforyongyu · 2024-11-22T12:07:29Z

why dees Resolve need to split up in order to be incorporated into the new blockbeat architecture?

So we split it into two steps, step one is Launch, which offers the inputs to the sweeper once triggered by block X, and before that block X is sent to the sweeper. This way we can ensure if block X closes a channel, its contract resolutions will offer these inputs to the sweeper, before the sweeper thinks there is nothing to be swept in block X.
Step two is then then waiting for the sweeping tx to be confirmed.

Is that that since we don't know which ones will be launched ahead of time, we can't set up the blockbeat consumer DAG upfront?

Yeah this part too - so these sub-consumers, like chain watcher and channel arbitrator, are ephemeral and are not managed by the blockbeat dispatcher. Some previous discussion can be found here. Basically it's easier to let the chainArb to manage the channelArb and chainArb here.

We now put the outpoint in the resolvers's logging so it's easier to debug. Also use the short channel ID whenever possible to shorten the log line length.

This commit adds a few helper methods to decide how the htlc output should be spent.

This commit is a pure refactor in which moves the sweep handling logic into the new methods.

This commit refactors the `Resolve` method by adding two resolver handlers to handle waiting for spending confirmations.

This commit adds new methods to handle making sweep requests based on the spending path used by the outgoing htlc output.

This commit adds checkpoint methods in `htlcTimeoutResolver`, which are similar to those used in `htlcSuccessResolver`.

This commit adds more methods to handle resolving the spending of the output based on different spending paths.

We will use this and its following commits to break the original `Resolve` methods into two parts - the first part is moved to a new method `Launch`, which handles sending a sweep request to the sweeper. The second part remains in `Resolve`, which is mainly waiting for a spending tx. Breach resolver currently doesn't do anything in its `Launch` since the sweeping of justice outputs are not handled by the sweeper yet.

This commit breaks the `Resolve` into two parts - the first part is moved into a `Launch` method that handles sending sweep requests, and the second part remains in `Resolve` which handles waiting for the spend. Since we are using both utxo nursery and sweeper at the same time, to make sure this change doesn't break the existing behavior, we implement the `Launch` as following, - zero-fee htlc - handled by the sweeper - direct output from the remote commit - handled by the sweeper - legacy htlc - handled by the utxo nursery

When calling `NotifyExitHopHtlc` it is allowed to pass a chan to subscribe to the HTLC's resolution when it's settled. However, this method will also return immediately if there's already a resolution, which means it behaves like a notifier and a getter. If the caller decides to only use the getter to do a non-blocking lookup, it can pass a nil subscriber chan to bypass the notification.

A minor refactor is done to support implementing `Launch`.

This commit makes `resolved` an atomic bool to avoid data race. This field is now defined in `contractResolverKit` to avoid code duplication.

In this commit, we break the old `launchResolvers` into two steps - step one is to launch the resolvers synchronously, and step two is to actually waiting for the resolvers to be resolved. This is critical as in the following commit we will require the resolvers to be launched at the same blockbeat when a force close event is sent by the chain watcher.

We need to offer the outgoing htlc one block earlier to make sure when the expiry height hits, the sweeper will not miss sweeping it in the same block. This also means the outgoing contest resolver now only does one thing - watch for preimage spend till height expiry-1, which can easily be moved into the timeout resolver instead in the future.

yyforyongyu self-assigned this Nov 18, 2024

yyforyongyu added this to the v0.19.0 milestone Nov 18, 2024

yyforyongyu mentioned this pull request Nov 18, 2024

Beat [4/4]: implement Consumer in chainWatcher #9277

Open

1 task

yyforyongyu added the no-itest label Nov 18, 2024

yyforyongyu force-pushed the yy-blockbeat branch from 763bfc7 to d7b0d2f Compare November 18, 2024 11:34

yyforyongyu force-pushed the yy-blockbeat-finalize branch from ac8ce82 to a16b2d4 Compare November 18, 2024 11:34

saubyk requested review from ziggie1984 and ellemouton November 18, 2024 14:53

yyforyongyu force-pushed the yy-blockbeat branch 4 times, most recently from 02cebbb to e40d7aa Compare November 19, 2024 13:28

yyforyongyu force-pushed the yy-blockbeat-finalize branch from a16b2d4 to ae69b98 Compare November 19, 2024 13:52

yyforyongyu force-pushed the yy-blockbeat branch from e40d7aa to 736edda Compare November 19, 2024 23:27

yyforyongyu force-pushed the yy-blockbeat-finalize branch 3 times, most recently from 3456f2e to 3d08b74 Compare November 20, 2024 06:21

yyforyongyu force-pushed the yy-blockbeat branch 2 times, most recently from 1e17c5c to 9ab2e53 Compare November 21, 2024 13:00

yyforyongyu force-pushed the yy-blockbeat-finalize branch from 3d08b74 to e69cd41 Compare November 21, 2024 13:16

yyforyongyu added the no-changelog label Nov 21, 2024

Base automatically changed from yy-blockbeat to yy-feature-blockbeat November 21, 2024 14:41

yyforyongyu force-pushed the yy-blockbeat-finalize branch from e69cd41 to cdd805f Compare November 21, 2024 14:45

yyforyongyu force-pushed the yy-feature-blockbeat branch from 9ab2e53 to 9ed2522 Compare November 21, 2024 14:49

Roasbeef requested changes Nov 22, 2024

View reviewed changes

yyforyongyu force-pushed the yy-blockbeat-finalize branch 2 times, most recently from 1e58126 to abbee3b Compare November 22, 2024 11:58

yyforyongyu requested a review from Roasbeef November 22, 2024 12:08

yyforyongyu force-pushed the yy-feature-blockbeat branch from 9ed2522 to 5b3b60c Compare November 25, 2024 05:49

yyforyongyu added 18 commits November 25, 2024 13:49

contractcourt: add verbose logging in resolvers

660fa82

We now put the outpoint in the resolvers's logging so it's easier to debug. Also use the short channel ID whenever possible to shorten the log line length.

contractcourt: add spend path helpers in timeout/success resolver

cd1a184

This commit adds a few helper methods to decide how the htlc output should be spent.

contractcourt: add sweep senders in htlcSuccessResolver

91cd342

This commit is a pure refactor in which moves the sweep handling logic into the new methods.

contractcourt: add resolver handlers in htlcSuccessResolver

a5c05ae

This commit refactors the `Resolve` method by adding two resolver handlers to handle waiting for spending confirmations.

contractcourt: remove redundant return value in claimCleanUp

f1a9e6d

contractcourt: add sweep senders in htlcTimeoutResolver

982470c

This commit adds new methods to handle making sweep requests based on the spending path used by the outgoing htlc output.

contractcourt: add methods to checkpoint states

dda8208

This commit adds checkpoint methods in `htlcTimeoutResolver`, which are similar to those used in `htlcSuccessResolver`.

contractcourt: add resolve handlers in htlcTimeoutResolver

fab3acf

This commit adds more methods to handle resolving the spending of the output based on different spending paths.

contractcourt: add Launch method to commit resolver

e75f396

contractcourt: add Launch method to incoming contest resolver

566eb80

A minor refactor is done to support implementing `Launch`.

contractcourt: add Launch method to outgoing contest resolver

f4e8e75

contractcourt: fix concurrent access to resolved

600732e

This commit makes `resolved` an atomic bool to avoid data race. This field is now defined in `contractResolverKit` to avoid code duplication.

contractcourt: fix concurrent access to launched

0d1e81b

yyforyongyu force-pushed the yy-blockbeat-finalize branch from abbee3b to aeafa63 Compare November 25, 2024 05:49

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Beat [3/4]: prepare resolvers to handle the `blockbeat` #9276

Beat [3/4]: prepare resolvers to handle the `blockbeat` #9276

yyforyongyu commented Nov 18, 2024 •

edited

Loading

coderabbitai bot commented Nov 18, 2024

Review skipped

Chat

CodeRabbit Commands (Invoked using PR comments)

Other keywords and placeholders

CodeRabbit Configuration File (`.coderabbit.yaml`)

Documentation and Community

github-actions bot commented Nov 18, 2024

Roasbeef commented Nov 22, 2024

Roasbeef left a comment

Roasbeef Nov 22, 2024

yyforyongyu Nov 22, 2024

Roasbeef Nov 22, 2024

yyforyongyu Nov 22, 2024

Roasbeef Nov 22, 2024

yyforyongyu Nov 22, 2024

Roasbeef Nov 22, 2024

yyforyongyu Nov 22, 2024

Roasbeef Nov 22, 2024

yyforyongyu Nov 22, 2024

Roasbeef Nov 22, 2024

yyforyongyu Nov 22, 2024

Roasbeef Nov 22, 2024

yyforyongyu Nov 22, 2024

Roasbeef Nov 22, 2024

yyforyongyu Nov 22, 2024

Roasbeef Nov 22, 2024

yyforyongyu Nov 22, 2024

yyforyongyu commented Nov 22, 2024

Beat [3/4]: prepare resolvers to handle the blockbeat #9276

Are you sure you want to change the base?

Beat [3/4]: prepare resolvers to handle the blockbeat #9276

Conversation

yyforyongyu commented Nov 18, 2024 • edited Loading

Alternatives

coderabbitai bot commented Nov 18, 2024

Review skipped

Chat

CodeRabbit Commands (Invoked using PR comments)

Other keywords and placeholders

CodeRabbit Configuration File (.coderabbit.yaml)

Documentation and Community

github-actions bot commented Nov 18, 2024

Pull reviewers stats

Roasbeef commented Nov 22, 2024

Roasbeef left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

yyforyongyu commented Nov 22, 2024

Beat [3/4]: prepare resolvers to handle the `blockbeat` #9276

Beat [3/4]: prepare resolvers to handle the `blockbeat` #9276

yyforyongyu commented Nov 18, 2024 •

edited

Loading

CodeRabbit Configuration File (`.coderabbit.yaml`)