fix(enginenetx): stabilize happy eyeballs algorithm #1301
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
The current algorithm is broken in that, after doubling the delay for a while (a) we could overflow if index is too big and (b) we end up scheduling all attempts 30 seconds after the beginning. Rather, we should switch to increasingly the delay linearly after we have reached a given cutoff and we should not use an algorithm that depends on the number be small to avoid overflow. Part of ooni/probe#2531
Murphy-OrangeMud
pushed a commit
to Murphy-OrangeMud/probe-cli
that referenced
this pull request
Feb 13, 2024
* Use 1s as the base delay (which leads to simpler computations) * Acknowledge that with very large indexes we still need to produce reasonable positive values, while the current algorithm breaks for indexes large than `63` * Acknowledge that, after incrementing exponentially for a while, it makes sense to continue in a flat fashion, where we increment linearly, while the current algorithm was aiming to always return 30s, which means several attempts would actually run in parallel * Acknowledge that we should use the same zero for all timings rather than having a goroutine dependent zero * Acknowledge that for now we don't need to mock `time.Now` Part of ooni/probe#2531
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Outdated suggestions cannot be applied.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.
Use 1s as the base delay (which leads to simpler computations)
Acknowledge that with very large indexes we still need to produce reasonable positive values, while the current algorithm breaks for indexes large than
63Acknowledge that, after incrementing exponentially for a while, it makes sense to continue in a flat fashion, where we increment linearly, while the current algorithm was aiming to always return 30s, which means several attempts would actually run in parallel
Acknowledge that we should use the same zero for all timings rather than having a goroutine dependent zero
Acknowledge that for now we don't need to mock
time.NowPart of ooni/probe#2531