Skip to content

Conversation

@Matilda-Clerke
Copy link
Contributor

PR description

Improve performance during backward sync by using bestPeerComparator. Also implement recording useless responses from peers in PeerTaskExecutor to eventually disconnect bad peers.

…ger#9411)"

This reverts commit 8784c4e.

Signed-off-by: Matilda Clerke <matilda.clerke@consensys.net>
…useless responses for some invalid responses

Signed-off-by: Matilda Clerke <matilda.clerke@consensys.net>
Signed-off-by: Matilda-Clerke <matilda.clerke@consensys.net>
Signed-off-by: Matilda Clerke <matilda.clerke@consensys.net>
Signed-off-by: Matilda Clerke <matilda.clerke@consensys.net>
…r task feature toggle in BackwardSyncStep

Signed-off-by: Matilda Clerke <matilda.clerke@consensys.net>
@Matilda-Clerke
Copy link
Contributor Author

Performance between test and control nodes looks to be the same now
image

@Matilda-Clerke Matilda-Clerke marked this pull request as ready for review December 19, 2025 00:35
@jframe
Copy link
Contributor

jframe commented Dec 19, 2025

Do you have times for BWS header download with and without your PR for comparison?

@Matilda-Clerke
Copy link
Contributor Author

Control node BWS header download was pretty consistant around 20 minutes.
Test node BWS header download was more variable but generally 3x as long at 60 minutes.

With these changes, they're both around the 20 minute mark.

…uses memory leaks

Signed-off-by: Matilda Clerke <matilda.clerke@consensys.net>
…tor-peers' into improve-performance-of-peerselector-peers
…if it causes memory leaks"

This reverts commit b9188d7.

Signed-off-by: Matilda Clerke <matilda.clerke@consensys.net>
@Matilda-Clerke Matilda-Clerke force-pushed the improve-performance-of-peerselector-peers branch from 0d1d1b9 to 8ef2bc5 Compare January 5, 2026 22:17
Copy link
Contributor

@pinges pinges left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Good to merge if you have a good answer for my question :-)

TOO_MANY_RESULTS_RETURNED(null),
RESULTS_DO_NOT_MATCH_QUERY(null),
NO_RESULTS_RETURNED(null, true),
TOO_MANY_RESULTS_RETURNED(null, false),
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why do we not record a useless response when there are too many results returned?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Tbh I was mostly just guessing. Assuming the results returned contains the results we expected, plus others, it's still useful.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looking at the code it looks like we are not making use of the response when the validation response is TOO_MANY_RESULTS_RETURNED. So I guess it should at least be a useless response.

@Matilda-Clerke Matilda-Clerke enabled auto-merge (squash) January 7, 2026 03:14
@Matilda-Clerke Matilda-Clerke requested a review from pinges January 7, 2026 03:49
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants