Add `ChannelPending` event emitted upon `funding_signed` by tnull · Pull Request #2098 · lightningdevkit/rust-lightning

tnull · 2023-03-10T16:32:07Z

Users currently don't have good way of being notified when channel open negotiations have succeeded and new channels are pending confirmation on chain.

To this end, we add a new ChannelPending event that is emitted when send or receive a funding_signed message, i.e., at the last moment before waiting for the confirmation period.

codecov-commenter · 2023-03-10T17:10:49Z

Codecov Report

Patch coverage: 80.95% and project coverage change: -0.03⚠️

Comparison is base (0e28bcb) 91.38% compared to head (3acc6f8) 91.36%.

❗ Current head 3acc6f8 differs from pull request most recent head 9873c7d. Consider uploading reports for the commit 9873c7d to get more accurate results

📣 This organization is not using Codecov’s GitHub App Integration. We recommend you install it so Codecov can continue to function properly for your repositories. Learn more

Additional details and impacted files

@@ Coverage Diff @@## main #2098 +/- ##
==========================================
- Coverage 91.38% 91.36% -0.03% 
==========================================
Files 102 102 Lines 49767 49844 +77 Branches 49767 49844 +77 ==========================================
+ Hits 45482 45540 +58 - Misses 4285 4304 +19

Impacted Files	Coverage Δ
lightning-background-processor/src/lib.rs	`77.80% <ø> (ø)`
lightning/src/events/mod.rs	`30.85% <0.00%> (-1.29%)`	⬇️
lightning/src/ln/functional_test_utils.rs	`92.59% <93.75%> (+<0.01%)`	⬆️
lightning/src/ln/chanmon_update_fail_tests.rs	`98.59% <100.00%> (+<0.01%)`	⬆️
lightning/src/ln/channel.rs	`90.08% <100.00%> (+0.04%)`	⬆️
lightning/src/ln/channelmanager.rs	`89.13% <100.00%> (+0.03%)`	⬆️
lightning/src/ln/functional_tests.rs	`98.22% <100.00%> (+<0.01%)`	⬆️
lightning/src/ln/priv_short_conf_tests.rs	`97.56% <100.00%> (+0.03%)`	⬆️
lightning/src/ln/reload_tests.rs	`95.57% <100.00%> (+0.01%)`	⬆️

... and 3 files with indirect coverage changes

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

tnull · 2023-03-13T20:59:51Z

Also rebased on main to take advantage of the CI fix.

tnull · 2023-03-13T21:22:08Z

... and kicked CI. Having tests that rely on bitcoind is really a gift that keeps on giving:

---- test_esplora_syncs stdout ----
thread 'test_esplora_syncs' panicked at 'called `Result::unwrap()` on an `Err` value: JsonRpc(Rpc(RpcError { code: -4, message: "Wallet file verification failed. SQLiteDatabase: Unable to obtain an exclusive lock on the database, is it being used by another instance of Bitcoin Core?\n", data: None }))', tests/integration_tests.rs:35:65
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

TheBlueMatt · 2023-03-14T18:17:35Z

Note that you'll need to squash all the commits before we land - we should avoid having commits that individually fail tests as it makes future bisection hard.

RCasatta · 2023-03-16T09:34:18Z

... and kicked CI. Having tests that rely on bitcoind is really a gift that keeps on giving:

---- test_esplora_syncs stdout ----
thread 'test_esplora_syncs' panicked at 'called `Result::unwrap()` on an `Err` value: JsonRpc(Rpc(RpcError { code: -4, message: "Wallet file verification failed. SQLiteDatabase: Unable to obtain an exclusive lock on the database, is it being used by another instance of Bitcoin Core?\n", data: None }))', tests/integration_tests.rs:35:65
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

Can't reproduce locally by trying in a while loop.

But I stopped early because the test implementation is leaky, I think the drop on ElectrsD and BitcoinD is not called because in OnceCell? Since it is not called the process cannot be stopped/killed and unpleasantly stick around.

The main point of bitcoind and electrsd is to have isolated process, I suggest to create one instance in every test using it instead of having a global &'static

tnull · 2023-03-16T13:12:31Z

Can't reproduce locally by trying in a while loop.
But I stopped early because the test implementation is leaky, I think the drop on ElectrsD and BitcoinD is not called because in OnceCell? Since it is not called the process cannot be stopped/killed and unpleasantly stick around.
The main point of bitcoind and electrsd is to have isolated process, I suggest to create one instance in every test using it instead of having a global &'static

I think spinning up new dedicated BitcoinD/EsploraD instance per test case is infeasible, especially when we need to premine 101 blocks every time before we can get started. However, it seems you're right and once_cell indeed won't call drop and hence doesn't allow us to clean up properly upon shutdown, which is kind of unfortunate. Will have to further look into how to mitigate this.

RCasatta · 2023-03-16T16:05:16Z

I think spinning up new dedicated BitcoinD/EsploraD instance per test case is infeasible, especially when we need to premine 101 blocks every time before we can get started

How many tests with electrs do you have? It takes a couple of seconds to spin up bitcoind and electrs synced with 101 blocks, and by default N tests are launched in parallel where N is the number of your cores.

tnull · 2023-03-27T11:45:00Z

How many tests with electrs do you have? It takes a couple of seconds to spin up bitcoind and electrs synced with 101 blocks, and by default N tests are launched in parallel where N is the number of your cores.

You're right, should be fine for now, although we might want to reconsider if the number of tests grows. Fixed in #2132

TheBlueMatt · 2023-03-29T22:10:08Z

+	/// Used to indicate that a channel with the given `channel_id` is being opened and pending
+	/// confirmation on-chain.
+	///
+	/// This event is emitted when the funding transaction has been signed and is broadcast to the


Should we have a note that, in rare cases, this event may be missed if a channel is ultimately confirmed quickly prior to the event being processed and then we restart? Its a stupid rare case but its possible I think?

I'm confused, which scenario are you describing exactly? Wouldn't rather be an edge case that we emit the event twice when we decide to rebroadcast on restart?

I think we can: (a) fund the channel, with an async monitor update, (b) not have that complete, but persist the manager, then (c) complete the monitor update and broadcast the funding tx (d) then restart, replay the chain and see the funding tx confirm, then (e) re-complete the monitor update completed in c, then we wont get the event again but the user will never have used it.
I think with "full" async monitor updates this becomes more realistic.

Should we have a note that, in rare cases, this event may be missed if a channel is ultimately confirmed quickly prior to the event being processed and then we restart?
Since we plan to leverage it for #2137, seems we should always regenerate it in cases like this? The channel will actually be ready, but as long as it's queued before the ChannelReady event, it should be fine.

Yea, that's my biggest concern here.

I think we can: (a) fund the channel, with an async monitor update, (b) not have that complete, but persist the manager, then (c) complete the monitor update and broadcast the funding tx (d) then restart, replay the chain and see the funding tx confirm, then (e) re-complete the monitor update completed in c, then we wont get the event again but the user will never have used it.
Ugh., I see. In that scenario it would persist the event queue when the event hasn't been emitted yet and wouldn't re-emit it after restart. Mh, instead of leaving a note I went ahead and am now tracking the emission of ChannelPending in the same way as ChannelReady and removed it from internal_funding_created entirely.
This means that we're now only emitting it after CMU completion or reestablish, i.e., would re-emit it if the "emitted" state wasn't persisted before.

tnull · 2023-03-31T11:37:53Z

Rebased on main to fix failing CI.

TheBlueMatt · 2023-03-31T19:12:46Z

LGTM, I think, feel free to squash the fixups.

tnull · 2023-04-01T10:08:26Z

LGTM, I think, feel free to squash the fixups.

Squashed without further changes.

wpaulino

LGTM, just two minor comments

wpaulino · 2023-04-03T16:47:26Z

 } else { panic!(); }

 node_b.handle_funding_created(&node_a.get_our_node_id(), &get_event_msg!(node_a_holder, MessageSendEvent::SendFundingCreated, node_b.get_our_node_id()));
+ let events_b = node_b.get_and_clear_pending_events();


Nit: Use expect_channel_pending_event here and below

Mh, yeah, unfortunatly I can't easily as expected_channel_pending_event takes a functional_test_utils::Node as parameter, and this sets up the ChannelManager manually. I'd rather not rewrite this entire test or all other tests just to avoid writing the match statement out once.

Currently, users don't have good way of being notified when channel open negotiations have succeeded and new channels are pending confirmation on chain. To this end, we add a new `ChannelPending` event that is emitted when send or receive a `funding_signed` message, i.e., at the last moment before waiting for the confirmation period. We track whether the event had previously been emitted in `Channel` and remove it from `internal_funding_created` entirely. Hence, we now only emit the event after ChannelMonitorUpdate completion, or upon channel reestablish. This mitigates a race condition where where we wouldn't persist the event *and* wouldn't regenerate it on restart, therefore potentially losing it, if async CMU wouldn't complete before ChannelManager persistence.

ghost · 2023-04-03T17:06:15Z

Squashed with the fixup:

diff --git a/lightning/src/events/mod.rs b/lightning/src/events/mod.rs
index f20d6667..2d576d33 100644
--- a/lightning/src/events/mod.rs+++ b/lightning/src/events/mod.rs@@ -631,5 +631,5 @@ pub enum Event {
/// The `temporary_channel_id` this channel used to be known by during channel establishment.
///
- /// May be `None` for channels created prior to LDK version 0.0.115.+ /// Will be `None` for channels created prior to LDK version 0.0.115.
former_temporary_channel_id: Option<[u8; 32]>,
/// The `node_id` of the channel counterparty.

ghost · 2023-04-03T16:55:59Z

+ write_tlv_fields!(writer, {
+ (0, channel_id, required),
+ (2, user_channel_id, required),
+ (4, former_temporary_channel_id, required),


nit: option is a bit more straightforward/saves a few bytes on the wire. It doesn't matter, though, just for future reference.

Yes, it's a tiny bit more efficient but to be honest I don't liking mixing implementation type and serialization type semantics if it can be avoided. In my mind option -> field is optional, not field is Option.

ghost

LGTM. I believe the CI failure is unrelated but can't actually see the issue.

ghost · 2023-04-03T18:21:53Z

Tracked CI failure as #2144

tnull force-pushed the 2023-03-add-channel-pending-event branch from 84ea88f to 226dfdeCompare March 10, 2023 16:34

wpaulino reviewed Mar 13, 2023
View reviewed changes

Comment threadlightning/src/util/events.rs
Comment threadlightning/src/ln/functional_test_utils.rs Outdated
Comment threadlightning/src/ln/channelmanager.rs Outdated

TheBlueMatt reviewed Mar 13, 2023
View reviewed changes

Comment threadlightning/src/util/events.rs Outdated
Comment threadlightning/src/util/events.rs

tnull force-pushed the 2023-03-add-channel-pending-event branch from 5bd01c1 to 37ebcdaCompare March 13, 2023 20:49

tnull force-pushed the 2023-03-add-channel-pending-event branch from 37ebcda to 970fa70Compare March 13, 2023 21:20

tnull force-pushed the 2023-03-add-channel-pending-event branch from 970fa70 to d2af89bCompare March 14, 2023 11:21

tnull mentioned this pull request Mar 14, 2023
Address the race condition rust-bitcoin/bitcoind#101
Open

wpaulino reviewed Mar 14, 2023
View reviewed changes

Comment threadlightning/src/util/events.rs
Comment threadlightning/src/ln/channel.rs Outdated

tnull force-pushed the 2023-03-add-channel-pending-event branch from d2af89b to 58ea755Compare March 14, 2023 15:25

TheBlueMatt reviewed Mar 14, 2023
View reviewed changes

Comment threadlightning/src/ln/channelmanager.rs Outdated
Comment threadlightning/src/util/events.rs Outdated

tnull force-pushed the 2023-03-add-channel-pending-event branch 2 times, most recently from 5526228 to 6a36d31Compare March 29, 2023 14:11

TheBlueMatt mentioned this pull request Mar 29, 2023
Support manual signaling of channel readiness #2137
Closed

TheBlueMatt reviewed Mar 29, 2023
View reviewed changes

tnull mentioned this pull request Mar 30, 2023
Return funding txid on connect_open_channellightningdevkit/ldk-node#53
Closed

tnull force-pushed the 2023-03-add-channel-pending-event branch from 01a545c to 3acc6f8Compare March 31, 2023 11:37

tnull force-pushed the 2023-03-add-channel-pending-event branch from 3acc6f8 to 8475376Compare April 1, 2023 10:07

wpaulino reviewed Apr 3, 2023
View reviewed changes

tnull force-pushed the 2023-03-add-channel-pending-event branch from 8475376 to 9873c7dCompare April 3, 2023 17:04

wpaulino approved these changes Apr 3, 2023
View reviewed changes

TheBlueMatt approved these changes Apr 3, 2023
View reviewed changes

dunxen approved these changes Apr 3, 2023
View reviewed changes

TheBlueMatt merged commit 7ca3709 into lightningdevkit:mainApr 3, 2023

tnull mentioned this pull request Apr 4, 2023
Add ChannelPending event lightningdevkit/ldk-node#60
Merged

Uh oh!

Conversation

tnull commented Mar 10, 2023• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

codecov-commenter commented Mar 10, 2023• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

tnull commented Mar 13, 2023

Uh oh!

tnull commented Mar 13, 2023• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

TheBlueMatt commented Mar 14, 2023

Uh oh!

RCasatta commented Mar 16, 2023• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tnull commented Mar 16, 2023• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

RCasatta commented Mar 16, 2023

Uh oh!

tnull commented Mar 27, 2023

Uh oh!

TheBlueMattMar 29, 2023

Choose a reason for hiding this comment

Uh oh!

tnullMar 30, 2023• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

TheBlueMattMar 30, 2023

Choose a reason for hiding this comment

Uh oh!

wpaulinoMar 30, 2023

Choose a reason for hiding this comment

Uh oh!

TheBlueMattMar 30, 2023

Choose a reason for hiding this comment

Uh oh!

tnullMar 31, 2023

Choose a reason for hiding this comment

Uh oh!

tnull commented Mar 31, 2023

Uh oh!

TheBlueMatt commented Mar 31, 2023

Uh oh!

tnull commented Apr 1, 2023

Uh oh!

wpaulino left a comment

Choose a reason for hiding this comment

Uh oh!

wpaulinoApr 3, 2023

Choose a reason for hiding this comment

Uh oh!

tnullApr 3, 2023

Choose a reason for hiding this comment

Uh oh!

Uh oh!

ghost commented Apr 3, 2023

Uh oh!

ghostApr 3, 2023

Choose a reason for hiding this comment

Uh oh!

ghostApr 3, 2023

Choose a reason for hiding this comment

Uh oh!

ghost left a comment

Choose a reason for hiding this comment

tnull commented Mar 10, 2023•
edited
Loading

codecov-commenter commented Mar 10, 2023•
edited
Loading

tnull commented Mar 13, 2023•
edited
Loading

RCasatta commented Mar 16, 2023•
edited
Loading

tnull commented Mar 16, 2023•
edited
Loading

tnullMar 30, 2023•
edited
Loading