[viceice]

Changes

Responses served from an HTTP cache provider (bypassServer) no longer wait for a throttle (maxRequestsPerSecond) or concurrency queue (concurrentRequestLimit) slot.

Since #38986 the cache check runs inside the task that throttle.add() and queue.add() wrap, so every cache hit used up a rate-limit slot although it never reaches the server. With a low maxRequestsPerSecond, lookups that are mostly served from the cache slow down a lot.

Now only the actual fetch is throttled and queued. The cache check, including the per-URL lock from #38986, runs before it, so the lock behaves as before.

Context

Please select one of the following:

• This closes an existing Issue, Closes: HTTP cache hits are throttled by maxRequestsPerSecond and concurrentRequestLimit #46624
• This doesn't close an Issue, but I accept the risk that this PR may be closed if maintainers disagree with its opening or implementation. Because this does not close an Issue, I will add a minimal reproduction repository as part of this PR.

AI assistance disclosure

Did you use AI tools to create any part of this pull request?

Please select one option and, if yes, briefly describe how AI was used (e.g., code, tests, docs) and which tool(s) you used.

• No — I did not use AI for this contribution.
• Yes — minimal assistance (e.g., IDE autocomplete, small code completions, grammar fixes).
• Yes — substantive assistance (AI-generated non‑trivial portions of code, tests, or documentation).
• Yes — other (please describe):

Code, test and PR description written with Claude Code (Claude Opus 5.5), reviewed by @viceice.

Use of AI in replying to PR comments

Who answers review comments:

• @viceice will read and reply directly. Name the account.
• An agent will draft replies and @username will read them before they are posted. Name the account.
• An agent will draft replies and reply autonomously. This is heavily discouraged, and we prefer that there are humans in the loop
• Nobody has explicitly committed to replying.

Documentation (please check one with an [x])

• I have updated the documentation, or
• No documentation update is required

How I've tested my work (please select one)

I have verified these changes via:

• Code inspection only, or
• Newly added/modified unit tests, or
• No unit tests, but ran on a real repository, or
• Both unit tests + ran on a real repository

The public repository:

A reproduction is compulsory for PRs that do not close an Issue: link a public repository above that reproduces the bug or shows the new behavior.
If this PR closes an Issue, confirm the fix against the Issue's reproduction.
If you are not adding a reproduction, state your reason here:

Confirmed against the issue with the new unit test: with maxRequestsPerSecond: 0.25, a second request served from the cache provider waited the full 4 s throttle interval before this change and returns immediately now.

PRs without a reproduction or a stated reason will not be reviewed and may be closed.

🤖 Generated with Claude Code

[github-actions]

🎉 This PR is included in version 44.127.1 🎉

The release is available on:

• GitHub release
• 44.127.1

Your semantic-release bot 📦🚀