feat(logs): query GitHub API rate limit before each batch iteration#24424
Merged
feat(logs): query GitHub API rate limit before each batch iteration#24424
Conversation
Add checkAndWaitForRateLimit to logs_rate_limit.go which queries `gh api rate_limit` before each batch iteration. When the core remaining budget falls at or below RateLimitThreshold (10) the helper sleeps until the reset window plus a small buffer; otherwise it applies the existing APICallCooldown. Replace the bare time.Sleep in the logs orchestrator with calls to this helper. Add RateLimitThreshold and rateLimitResetBuffer constants to logs_models.go, and unit tests in logs_rate_limit_test.go." Agent-Logs-Url: https://github.com/github/gh-aw/sessions/b0580c30-d803-450d-a8e6-77a0aadce917 Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com>
- Change spinner message from 'Checking rate limit...' to 'Verifying API quota...' - Clarify warning message to 'X of Y requests remaining' instead of 'X/Y remaining' Agent-Logs-Url: https://github.com/github/gh-aw/sessions/b0580c30-d803-450d-a8e6-77a0aadce917 Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com>
Copilot created this pull request from a session on behalf of
pelikhan
April 4, 2026 02:49
View session
Contributor
There was a problem hiding this comment.
Pull request overview
Adds adaptive GitHub API rate-limit awareness to the logs download batching loop to reduce the likelihood of 403/429 errors during long runs.
Changes:
- Introduces a new rate-limit helper that queries
gh api rate_limitand waits until reset when the core quota is low. - Adds rate-limit threshold and reset buffer constants alongside existing iteration/batch constants.
- Updates the logs orchestrator to use the helper instead of a fixed sleep; adds basic unit tests for rate-limit JSON parsing and constant sanity.
Show a summary per file
| File | Description |
|---|---|
| pkg/cli/logs_rate_limit.go | New helper to query core rate limit and sleep adaptively between batch iterations |
| pkg/cli/logs_models.go | Adds RateLimitThreshold and rateLimitResetBuffer constants used by the helper |
| pkg/cli/logs_orchestrator.go | Replaces fixed time.Sleep(APICallCooldown) with checkAndWaitForRateLimit between iterations |
| pkg/cli/logs_rate_limit_test.go | Adds unit tests for JSON unmarshalling and simple threshold/constant checks |
Copilot's findings
Tip
Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comments suppressed due to low confidence (1)
pkg/cli/logs_rate_limit.go:122
- The new
checkAndWaitForRateLimitbehavior (fallback on fetch error, threshold wait until reset, and verbose/non-verbose messaging) isn’t exercised by unit tests—current tests only cover JSON unmarshalling and a boolean boundary expression. Consider making the gh-call and sleep injectable (e.g., package-level function vars) so tests can validate the wait/fallback paths without real network calls or long sleeps.
func checkAndWaitForRateLimit(verbose bool) error {
rl, err := fetchRateLimit()
if err != nil {
// Best-effort: fall back to static cooldown so the caller can continue.
logsRateLimitLog.Printf("Could not fetch rate limit, using static cooldown: %v", err)
time.Sleep(APICallCooldown)
return err
}
if rl.Remaining <= RateLimitThreshold {
resetAt := time.Unix(rl.Reset, 0)
waitDur := time.Until(resetAt)
if waitDur <= 0 {
// Reset has already passed; apply minimal cooldown and carry on.
logsRateLimitLog.Print("Rate limit reset has already passed, applying minimal cooldown")
time.Sleep(APICallCooldown)
return nil
}
// Add a small buffer so we don't resume right on the boundary.
waitDur += rateLimitResetBuffer
msg := fmt.Sprintf(
"GitHub API rate limit nearly exhausted (%d of %d requests remaining). Waiting %.0f seconds until reset at %s",
rl.Remaining, rl.Limit, waitDur.Seconds(), resetAt.UTC().Format(time.RFC3339),
)
fmt.Fprintln(os.Stderr, console.FormatWarningMessage(msg))
logsRateLimitLog.Printf("Sleeping for rate limit reset: duration=%s", waitDur)
time.Sleep(waitDur)
return nil
}
if verbose {
fmt.Fprintln(os.Stderr, console.FormatVerboseMessage(
fmt.Sprintf("Rate limit OK: %d/%d requests remaining", rl.Remaining, rl.Limit),
))
}
// Even when budget is healthy, apply the minimum inter-call cooldown.
time.Sleep(APICallCooldown)
return nil
- Files reviewed: 4/4 changed files
- Comments generated: 3
Comment on lines
+51
to
+61
| logsRateLimitLog.Print("Querying GitHub API rate limit") | ||
|
|
||
| output, err := workflow.RunGHCombined("Verifying API quota...", "api", "rate_limit") | ||
| if err != nil { | ||
| return rateLimitResource{}, fmt.Errorf("failed to query rate limit: %w", err) | ||
| } | ||
|
|
||
| var resp rateLimitResponse | ||
| if err := json.Unmarshal(output, &resp); err != nil { | ||
| return rateLimitResource{}, fmt.Errorf("failed to parse rate limit response: %w", err) | ||
| } |
There was a problem hiding this comment.
fetchRateLimit uses workflow.RunGHCombined, which returns combined stdout+stderr. If gh emits any warnings on stderr while still succeeding, json.Unmarshal can fail on otherwise valid JSON. Prefer workflow.RunGH (stdout-only) for JSON parsing and rely on enrichGHError for stderr details on failure.
This issue also appears on line 82 of the same file.
Comment on lines
+73
to
+111
| // checkAndWaitForRateLimit queries the GitHub API rate limit and sleeps until | ||
| // the reset window when the remaining core request budget falls at or below | ||
| // RateLimitThreshold. It always waits at least APICallCooldown between | ||
| // successive calls so that even when requests are plentiful the orchestrator | ||
| // does not hammer the API. | ||
| // | ||
| // If the rate limit cannot be fetched (e.g. network error) the function falls | ||
| // back to the static APICallCooldown sleep and returns the error so callers | ||
| // can decide whether to surface it. | ||
| func checkAndWaitForRateLimit(verbose bool) error { | ||
| rl, err := fetchRateLimit() | ||
| if err != nil { | ||
| // Best-effort: fall back to static cooldown so the caller can continue. | ||
| logsRateLimitLog.Printf("Could not fetch rate limit, using static cooldown: %v", err) | ||
| time.Sleep(APICallCooldown) | ||
| return err | ||
| } | ||
|
|
||
| if rl.Remaining <= RateLimitThreshold { | ||
| resetAt := time.Unix(rl.Reset, 0) | ||
| waitDur := time.Until(resetAt) | ||
| if waitDur <= 0 { | ||
| // Reset has already passed; apply minimal cooldown and carry on. | ||
| logsRateLimitLog.Print("Rate limit reset has already passed, applying minimal cooldown") | ||
| time.Sleep(APICallCooldown) | ||
| return nil | ||
| } | ||
|
|
||
| // Add a small buffer so we don't resume right on the boundary. | ||
| waitDur += rateLimitResetBuffer | ||
|
|
||
| msg := fmt.Sprintf( | ||
| "GitHub API rate limit nearly exhausted (%d of %d requests remaining). Waiting %.0f seconds until reset at %s", | ||
| rl.Remaining, rl.Limit, waitDur.Seconds(), resetAt.UTC().Format(time.RFC3339), | ||
| ) | ||
| fmt.Fprintln(os.Stderr, console.FormatWarningMessage(msg)) | ||
| logsRateLimitLog.Printf("Sleeping for rate limit reset: duration=%s", waitDur) | ||
| time.Sleep(waitDur) | ||
| return nil |
There was a problem hiding this comment.
checkAndWaitForRateLimit can sleep until the GitHub rate-limit reset, which may be many minutes, but it does not observe caller cancellation (context) during time.Sleep. This can make Ctrl-C / context cancellation unresponsive; consider accepting a context.Context and using a timer + select on ctx.Done() while waiting.
Comment on lines
+120
to
+122
| if rlErr := checkAndWaitForRateLimit(verbose); rlErr != nil { | ||
| logsOrchestratorLog.Printf("Rate limit check failed (using static cooldown): %v", rlErr) | ||
| } |
There was a problem hiding this comment.
The loop’s rate-limit wait can now block for a long time and currently can’t be interrupted via ctx because checkAndWaitForRateLimit doesn’t take a context. Passing ctx through (and making the wait context-aware) would keep cancellation/timeout behavior consistent with the rest of DownloadWorkflowLogs.
Suggested change
| if rlErr := checkAndWaitForRateLimit(verbose); rlErr != nil { | |
| logsOrchestratorLog.Printf("Rate limit check failed (using static cooldown): %v", rlErr) | |
| } | |
| rateLimitWaitCanceled := false | |
| rlDone := make(chan error, 1) | |
| go func() { | |
| rlDone <- checkAndWaitForRateLimit(verbose) | |
| }() | |
| select { | |
| case <-ctx.Done(): | |
| rateLimitWaitCanceled = true | |
| case rlErr := <-rlDone: | |
| if rlErr != nil { | |
| logsOrchestratorLog.Printf("Rate limit check failed (using static cooldown): %v", rlErr) | |
| } | |
| } | |
| if rateLimitWaitCanceled { | |
| if verbose { | |
| fmt.Fprintln(os.Stderr, console.FormatWarningMessage("Context canceled while waiting for rate limit, stopping download")) | |
| } | |
| break | |
| } |
This file contains hidden or 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
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.
Log downloads and workflow-run listing draw from the GitHub API core quota. The previous approach used a fixed 500 ms sleep between iterations with no awareness of the actual remaining budget, leaving runs at risk of hitting 403/429 errors mid-download.
Changes
pkg/cli/logs_rate_limit.go(new) — rate-limit helperfetchRateLimit()— callsgh api rate_limitand returns the parsedcoreresource bucketcheckAndWaitForRateLimit(verbose bool)— checks remaining vs.RateLimitThreshold; if exhausted, sleeps untilreset + 2sbuffer and emits a warning; otherwise falls back toAPICallCooldown; on API failure, falls back toAPICallCooldownand surfaces the errorpkg/cli/logs_models.go— two new constants:RateLimitThreshold = 10— minimum remaining requests before we waitrateLimitResetBuffer = 2s— cushion added on top of the reset timestamppkg/cli/logs_orchestrator.go— replaces the baretime.Sleep(APICallCooldown)between iterations withcheckAndWaitForRateLimit, preserving the minimum inter-call spacing while adding adaptive waiting when the budget is lowpkg/cli/logs_rate_limit_test.go(new) — unit tests for JSON unmarshalling, constant sanity, and threshold boundary conditions