Fix splicing hang, speed CI#8911
Open
rustyrussell wants to merge 9 commits intoElementsProject:masterfrom
Open
Fix splicing hang, speed CI#8911rustyrussell wants to merge 9 commits intoElementsProject:masterfrom
rustyrussell wants to merge 9 commits intoElementsProject:masterfrom
Conversation
rustyrussell
force-pushed
the
guilt/flakes31
branch
2 times, most recently
from
7c72d0c to
309d453
Compare
rustyrussell
force-pushed
the
guilt/flakes31
branch
from
309d453 to
bce3b30
Compare
I got a flake, but all I see is:
```
> assert suspended == set()
E AssertionError: assert {'c26485f2839c5a27'} == set()
E
E Extra items in the left set:
E 'c26485f2839c5a27'
E
E Full diff:
E - set()
E + {
E + 'c26485f2839c5a27',
E + }
tests/test_misc.py:5049: AssertionError
```
Add some more diagnostics, but it's also clear that something is being
suspended and not terminating before we finish.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This is useful for splicing: given an HTLC state, do we need to send more messages to get it into a non-pending state? Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
And remove `uncommitted_ok` flag which was always false. Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Otherwise, we can hang: we don't send commitment_signed, and they're waiting to receive it. 1. We defer fee updates, blockheight updates and master requests (adding and closing htlcs) if we're *trying* or *started* to quiesce. 2. We only stop actually sending commitment_signed if we have sent STFU. Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> Changelog-EXPERIMENTAL: Protocol: avoid an occasional hang when splicing with a pending closing HTLC.
It's timing out after 2 hours sometimes: this now make it finish in 53 minutes. Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
…f random delays. Sometimes this times out after 30 minutes. Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
rustyrussell
force-pushed
the
guilt/flakes31
branch
5 times, most recently
from
e69d5d3 to
ce19937
Compare
Usually downloading and installing takes 90 seconds. But sometimes it takes an hour! Use caching for this, to keep it consistent. Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
.github/scripts/setup.sh does this already, *and* it uses the cache now. Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
rustyrussell
force-pushed
the
guilt/flakes31
branch
4 times, most recently
from
1d42ca2 to
6b5650a
Compare
I noticed this line causing a delay; ChatGPT thinks NSS name lookup. I've removed the useradd line altogether. I also install eatmydata first, to try to speed the other installs. This drops the install step from 1m45 to about 30 seconds. Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
rustyrussell
force-pushed
the
guilt/flakes31
branch
from
6b5650a to
953097c
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
test_splice_rbf kept hanging. Turns out one side was stopping sending before a commitment_signed, resulting in a hang. I clarified the logic around when we can send STFU, and when we should defer new actions, and now it's both simpler to understand and doesn't hang.
I also ended up speeding CI further: