Allow async events processing without holding `total_consistency_lock` #2199

tnull · 2023-04-18T13:16:29Z

Fixes #2003.

Unfortunately, the RAII types used by RwLock are not Send, which is why they can't be held over await boundaries. In order to allow asynchronous events processing in multi-threaded environments, we here allow to process events without holding the total_consistency_lock. We do so by cloning the events and only draining and persisting the queue after they have successfully been processed.

The first commit reverts a prior commit of #2177, as we now want the behavior of the two process_event methods to diverge, i.e., want to avoid cloning in the sync case.

I tried to be minimally invasive as the event processing will receive a general overhaul with #2167 and follow-ups and any more substantial changes would likely only make sense after they have landed.

tnull · 2023-04-18T13:20:21Z

Currently fails due to a previously-silent panic in BP tests that, due to the behavior of the tokio runtime, wasn't surfaced and caught before. Looking into that.

codecov-commenter · 2023-04-18T14:26:26Z

Codecov Report

Patch coverage: 94.11% and project coverage change: +1.04 🎉

Comparison is base (2ebbe6f) 91.34% compared to head (a5358d0) 92.38%.

❗ Current head a5358d0 differs from pull request most recent head f2453b7. Consider uploading reports for the commit f2453b7 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    #2199      +/-   ##
==========================================
+ Coverage   91.34%   92.38%   +1.04%     
==========================================
  Files         102      104       +2     
  Lines       50470    61358   +10888     
  Branches    50470    61358   +10888     
==========================================
+ Hits        46103    56688   +10585     
- Misses       4367     4670     +303

Impacted Files	Coverage Δ
lightning/src/ln/channelmanager.rs	`91.65% <75.00%> (+2.48%)`	⬆️
lightning-background-processor/src/lib.rs	`83.51% <100.00%> (+6.40%)`	⬆️
lightning-net-tokio/src/lib.rs	`78.41% <100.00%> (ø)`

... and 64 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 · 2023-04-18T14:38:10Z

Currently fails due to a previously-silent panic in BP tests that, due to the behavior of the tokio runtime, wasn't surfaced and caught before. Looking into that.

Correction: after fixing the CI script, this should now really fail until we fix the bug..

lightning/src/ln/channelmanager.rs

Just two trivial compiler warnings that are unrelated to the changes made here.

Currently the BP `futures` tests rely on `std`. In order to actually have them run, we should enable `std`, i.e., remove `--no-default-features`.

TheBlueMatt

Sorry can you not break out the macro? Not because it's wrong here but because there's a lot more complexity coming in a followup PR in there and we'll just have to add it again.

tnull · 2023-04-20T10:44:12Z

Sorry can you not break out the macro? Not because it's wrong here but because there's a lot more complexity coming in a followup PR in there and we'll just have to add it again.

Alright, dropped the revert commit and now also cloning in the sync case.

lightning/src/ln/channelmanager.rs

TheBlueMatt

LGTM.

lightning/src/ln/channelmanager.rs

TheBlueMatt · 2023-04-21T15:57:23Z

lightning/src/ln/channelmanager.rs

+				let mut pending_events = $self.pending_events.lock().unwrap();
+				pending_events.drain(..num_events);
+				processed_all_events = pending_events.is_empty();
+				$self.pending_events_processor.store(false, Ordering::Release);


Should this only happen if !processed_all_events? Not a big deal either way, I think.

You mean if we processed all events? Yeah, I think I'd leave it as is.

Oh, no, I mean literally just move the setter here into a check for if we're about to go around again.

Ah, yes, had understood as much, but we def. need to reset in the case we leave the method. We could have moved the compare_exchange out of the loop and only reset the flag on exit, but given that it's a rare edge case anyways I thought it made sense to leave as is.

Unfortunately, the RAII types used by `RwLock` are not `Send`, which is why they can't be held over `await` boundaries. In order to allow asynchronous events processing in multi-threaded environments, we here allow to process events without holding the `total_consistency_lock`.

TheBlueMatt · 2023-04-21T17:22:19Z

lightning-background-processor/src/lib.rs

-					sender.send(()).unwrap();
+					match sender.send(()) {
+						Ok(()) => {},
+						Err(std::sync::mpsc::SendError(())) => println!("Persister failed to notify as receiver went away."),


Because we're shutting the other task down after the first send. However, we also persist again on shutdown, which triggers a second send, which would panic as the receiver is already gone at that point.

I think a comment for why this is ok would be helpful

alecchendev

LGTM I think! Making sure I'm getting this right, do types need to implement Send across await boundaries because in a multi-threaded environments, a task waiting on a future to complete may be moved to execute on another thread?

alecchendev · 2023-04-21T21:55:31Z

lightning/src/ln/channelmanager.rs

-		// we can be sure no other persists happen while processing events.
-		let _read_guard = $self.total_consistency_lock.read().unwrap();
+		let mut processed_all_events = false;
+		while !processed_all_events {


How come this is all run in a while loop? IIUC there may be other events added to pending_events by other async tasks while handling the events, which is how we end up not having processed all events, but why do we keep processing until pending_events is empty as opposed to just processing the events that were present when we first call this function? I guess does it make much of a difference or is it more just that we might as well do it while we're here

Because we no longer allow multiple processors to run at the same time - if one process_events call starts, and makes some progress, then an event is generated, causing a second process_events call to happen, the second call might return early, but there's some events there the user expects to have processed. Thus, we need to make sure the first process_events goes around again and processes the remaining events.

lightning-background-processor/src/lib.rs

valentinewallace · 2023-04-21T23:30:31Z

lightning-background-processor/src/lib.rs

-					sender.send(()).unwrap();
+					match sender.send(()) {
+						Ok(()) => {},
+						Err(std::sync::mpsc::SendError(())) => println!("Persister failed to notify as receiver went away."),


I think a comment for why this is ok would be helpful

lightning/src/ln/channelmanager.rs

tnull force-pushed the 2023-04-fix-async-event-processing branch 3 times, most recently from 9d6077b to 5467f97 Compare April 18, 2023 14:36

TheBlueMatt reviewed Apr 18, 2023

View reviewed changes

lightning/src/ln/channelmanager.rs Outdated Show resolved Hide resolved

lightning/src/ln/channelmanager.rs Outdated Show resolved Hide resolved

wpaulino reviewed Apr 18, 2023

View reviewed changes

lightning/src/ln/channelmanager.rs Outdated Show resolved Hide resolved

TheBlueMatt added this to the 0.0.115 milestone Apr 18, 2023

tnull added 2 commits April 19, 2023 10:25

Fix unrelated warnings

0d59417

Just two trivial compiler warnings that are unrelated to the changes made here.

Actually test futures builds

36bf817

Currently the BP `futures` tests rely on `std`. In order to actually have them run, we should enable `std`, i.e., remove `--no-default-features`.

tnull force-pushed the 2023-04-fix-async-event-processing branch from 5467f97 to c9cfd20 Compare April 19, 2023 09:13

TheBlueMatt reviewed Apr 19, 2023

View reviewed changes

tnull force-pushed the 2023-04-fix-async-event-processing branch from c9cfd20 to dd48d55 Compare April 20, 2023 10:43

tnull force-pushed the 2023-04-fix-async-event-processing branch from dd48d55 to d7de357 Compare April 20, 2023 10:46

TheBlueMatt reviewed Apr 20, 2023

View reviewed changes

lightning/src/ln/channelmanager.rs Outdated Show resolved Hide resolved

TheBlueMatt reviewed Apr 20, 2023

View reviewed changes

lightning/src/ln/channelmanager.rs Outdated Show resolved Hide resolved

Fix BP prune timer and don't panic on persistence notification failure

b546822

tnull force-pushed the 2023-04-fix-async-event-processing branch from d9c14d4 to a5358d0 Compare April 21, 2023 09:44

This was referenced Apr 21, 2023

Switch to async background processing and cleanup runtime handling. lightningdevkit/ldk-node#68

Merged

Add Tokio example to process_events_async docs #2004

Merged

TheBlueMatt reviewed Apr 21, 2023

View reviewed changes

tnull force-pushed the 2023-04-fix-async-event-processing branch from a5358d0 to f2453b7 Compare April 21, 2023 16:05

TheBlueMatt reviewed Apr 21, 2023

View reviewed changes

TheBlueMatt approved these changes Apr 21, 2023

View reviewed changes

alecchendev reviewed Apr 21, 2023

View reviewed changes

valentinewallace approved these changes Apr 22, 2023

View reviewed changes

TheBlueMatt merged commit 5f96d13 into lightningdevkit:main Apr 22, 2023
14 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Allow async events processing without holding `total_consistency_lock` #2199

Allow async events processing without holding `total_consistency_lock` #2199

tnull commented Apr 18, 2023

tnull commented Apr 18, 2023 •

edited

codecov-commenter commented Apr 18, 2023 •

edited

tnull commented Apr 18, 2023

TheBlueMatt left a comment

tnull commented Apr 20, 2023

TheBlueMatt left a comment

TheBlueMatt Apr 21, 2023

tnull Apr 21, 2023

TheBlueMatt Apr 22, 2023

tnull Apr 24, 2023

TheBlueMatt Apr 21, 2023

tnull Apr 21, 2023 •

edited

valentinewallace Apr 21, 2023

alecchendev left a comment

alecchendev Apr 21, 2023 •

edited

TheBlueMatt Apr 22, 2023

valentinewallace Apr 21, 2023

Allow async events processing without holding total_consistency_lock #2199

Allow async events processing without holding total_consistency_lock #2199

Conversation

tnull commented Apr 18, 2023

tnull commented Apr 18, 2023 • edited

codecov-commenter commented Apr 18, 2023 • edited

Codecov Report

tnull commented Apr 18, 2023

TheBlueMatt left a comment

Choose a reason for hiding this comment

tnull commented Apr 20, 2023

TheBlueMatt 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

tnull Apr 21, 2023 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

alecchendev left a comment

Choose a reason for hiding this comment

alecchendev Apr 21, 2023 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Allow async events processing without holding `total_consistency_lock` #2199

Allow async events processing without holding `total_consistency_lock` #2199

tnull commented Apr 18, 2023 •

edited

codecov-commenter commented Apr 18, 2023 •

edited

tnull Apr 21, 2023 •

edited

alecchendev Apr 21, 2023 •

edited