process_registry_updates without a full copy of the validator set #14130
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
prestonvanloon
force-pushed
the
process_registry_updates-readFromEveryValidator
branch
from
11ee9c3
to
3f55435
Compare
prestonvanloon
force-pushed
the
process_registry_updates-readFromEveryValidator
branch
from
3f55435
to
96f8db7
Compare
potuz
reviewed
Jun 21, 2024
Comment on lines
+78
to
+88
| // To avoid copying the state validator set via st.Validators(), we will perform a read only pass | ||
| // over the validator set while collecting validator indices where the validator copy is actually | ||
| // necessary, then we will process these operations. | ||
| eligibleForActivationQ := make([]primitives.ValidatorIndex, 0) | ||
| eligibleForActivation := make([]primitives.ValidatorIndex, 0) | ||
| eligibleForEjection := make([]primitives.ValidatorIndex, 0) | ||
|
|
||
| if err := st.ReadFromEveryValidator(func(idx int, val state.ReadOnlyValidator) error { | ||
| // Collect validators eligible to enter the activation queue. | ||
| if helpers.IsEligibleForActivationQueue(val, currentEpoch) { | ||
| eligibleForActivationQ = append(eligibleForActivationQ, primitives.ValidatorIndex(idx)) |
There was a problem hiding this comment.
This pattern seems useful generically: we can probably split out a helper that takes a predicate and returns the list of validator indices that satisfies the predicate. We could have it take a slice of predicates and then return a slice of slices with the val indices (in this case it would be three predicates). PErhaps a benchmark shows that three loops vs 1 loop with three predicates is more or less the same.
There was a problem hiding this comment.
Yes, I think that would be a good addition as well. I am thinking something like
func (s BeaconState) FilterValidators(fns []func(ReadOnlyValidator) bool) [][]primitives.ValidatorIndex {
res := make([]primitives.ValidatorIndex, len(fns))
for i, v := range s.ReadOnlyValidators {
for j, fn := range fns {
if fn(v) {
res[j] = append(res[j], primitives.ValidatorIndex(i))
}
}
}
return res
}Can I add it in a follow up PR?
terencechain
approved these changes
Jun 26, 2024
github-merge-queue
bot
removed this pull request from the merge queue due to no response for status checks
github-merge-queue
bot
removed this pull request from the merge queue due to no response for status checks
prestonvanloon
deleted the
process_registry_updates-readFromEveryValidator
branch
This pull request was closed.
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.
What type of PR is this?
Bug fix
What does this PR do? Why is it needed?
state.Validators() is a full copy of 1M+ validators. By using a series of validator index slices, we can reduce the worst case copy to ~8M bytes ~= 8Mb of indices to process.
The average case is near zero allocation.
Special shoutout to @terencechain for the idea and the debugging of my implementation.
Soft requirement for #14005.
Which issues(s) does this PR fix?
Other notes for review
I moved sortableIndices to its own file and added unit tests for this specifically. The lack of unit tests on this struct cost us nearly a full day of debugging a silly mistake.