There are intermittent failures on current master (https://github.com/bitcoin/bitcoin/pull/24421#issuecomment-1067451667) as the witness txs occasionally make it into other node's mempools when they shouldn't.
Fix this by removing the test as suggested earlier in #24421 (review).
<!--e57a25ab6845829454e8d69fc972939a-->
The following sections might be updated with supplementary metadata relevant to reviewers and maintainers.
<!--174a7506f384e20aa4161008e828411d-->
Reviewers, this pull request conflicts with the following ones:
If you consider this pull request important, please also help to review the conflicting pull requests. Ideally, start with the one that should be merged first.
tACK d6b24e14d2a828ba7328f0cae1d0d7c5f4f6b3bd. Tested on NixOS 22.05 64 bits.
I was able to reproduce the bug (https://github.com/bitcoin/bitcoin/pull/24421#issuecomment-1067451667) prior to this fix using: seq 100 | xargs -i -P 20 python test/functional/feature_segwit.py (100 tests in batches of 20).
Could not reproduce the bug after this PR using the same command with 1000 tests.
Code review ACK d6b24e14d2a828ba7328f0cae1d0d7c5f4f6b3bd
Is there a way to introduce random sleeps into the test logic to flush out other racy test bugs?
Sure, it should be possible to inject a random sleep after sendrawtransaction. Though, I am not sure if this is going to give consistent results either. I think the only way would be to do the same for txs that was done for blocks. (Always sync unless opted out).
post merge ACK