Broadcast produced transition immediately #13551
Merged
Conversation
This file contains 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
nholland94
requested changes
Jul 10, 2023
nholland94
approved these changes
Jul 12, 2023
georgeee
force-pushed
the
georgeee/broadcast-transition-immediately-upon-finalization
branch
2 times, most recently
from
4cb69af to
615d11b
Compare
|
!ci-build-me |
| @@ -2056,9 +2056,6 @@ let create ?wallets (config : Config.t) = | |||
| valid_cb | |||
| | `Internal -> | |||
| (*Send callback to publish the new block. Don't log rebroadcast message if it is internally generated; There is a broadcast log*) | |||
| don't_wait_for | |||
| (Mina_networking.broadcast_state net | |||
| (Mina_block.Validated.forget transition) ) ; | |||
| Option.iter | |||
| ~f: | |||
| (Fn.flip | |||
There was a problem hiding this comment.
isn't the broadcast blocked by this validation callback? Like whatever was causing the delay to get to here, wouldn't that still affect block broadcasts?
There was a problem hiding this comment.
Nope, because there is no validation callback for internal blocks. This code does nothing. I rewrote it subtantially in Bitswap branch, but here attempted to keep the code change minimal
Problem: after block is produced, it is scheduled for broadcast only in next cycle (or even in the cycle after next). If some long async job appears in between (and this is what happens in Berkeley now), broadcast will be delayed, potentially causing block to appear later-than-needed on other nodes (especially on clusters using block window below 1 minute). Solution: broadcast produced transition immediately after it's added to frontier.
georgeee
force-pushed
the
georgeee/broadcast-transition-immediately-upon-finalization
branch
from
615d11b to
2b89556
Compare
|
!ci-build-me |
nholland94
deleted the
georgeee/broadcast-transition-immediately-upon-finalization
branch
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
This PR addresses two conditions related to block processing/production.
First problem is that after block is produced, it is scheduled for broadcast only in next cycle (or even in the cycle after next). If some long async job appears in between (and this is what happens in Berkeley now), broadcast will be delayed, potentially causing block to appear later-than-needed on other nodes (especially on clusters using block window below 1 minute).
Second problem appears when node goes to bootstrap while a block is being processed in
add_and_finalize. It happens very infrequently, but shuts the node down when happens.Solution: broadcast produced transition immediately after it's added to frontier and check for pipe not being closed before writing to it.
Explain how you tested your changes:
k=12), monitoring long async cycles and ensuring cluster's healthChecklist:
Closes #13508 with the help of #13550