perf(github): cache check runs with sync.Once

Extract fetchAllCheckRunPages to collect all paginated check
runs in one pass, then cache the result on Provider via
sync.Once. Concurrent getExistingCheckRunID calls for
multiple matched PipelineRuns share a single
ListCheckRunsForRef API round-trip instead of each fetching
independently.

Also extract searchCheckRuns helper and adds a nil guard
on ExternalID.

Signed-off-by: Akshay Pant <akpant@redhat.com>
Assisted-by: Claude <noreply@anthropic.com>

Author: theakshaypant

pkg/provider/github/github.go

@@ -67,13 +67,20 @@ type Provider struct {
 	pacUserLogin       string // user/bot login used by PAC
 	clock              clockwork.Clock
 	graphQLClient      *graphQLClient
+	checkRunsCache     checkRunsCache
 }
 
 type skippedRun struct {
 	mutex      *sync.Mutex
 	checkRunID int64
 }
 
+type checkRunsCache struct {
+	once sync.Once
+	runs []*github.CheckRun
+	err  error
+}
+
 func New() *Provider {
 	return &Provider{
 		APIURL:        github.Ptr(keys.PublicGithubAPIURL),

pkg/provider/github/status.go

@@ -42,7 +42,9 @@ const taskStatusTemplate = `
 {{- end }}
 </table>`
 
-func (v *Provider) getExistingCheckRunID(ctx context.Context, runevent *info.Event, status providerstatus.StatusOpts) (*int64, error) {
+// fetchAllCheckRunPages retrieves every page of check runs for the event SHA.
+func (v *Provider) fetchAllCheckRunPages(ctx context.Context, runevent *info.Event) ([]*github.CheckRun, error) {
+	var all []*github.CheckRun
 	opt := github.ListOptions{PerPage: v.PaginedNumber}
 	for {
 		res, resp, err := wrapAPI(v, "list_check_runs_for_ref", func() (*github.ListCheckRunsResults, *github.Response, error) {
@@ -55,25 +57,37 @@ func (v *Provider) getExistingCheckRunID(ctx context.Context, runevent *info.Eve
 		if err != nil {
 			return nil, err
 		}
-
-		for _, checkrun := range res.CheckRuns {
-			// if it is a Pending approval CheckRun then overwrite it
-			if isPendingApprovalCheckrun(checkrun) || isFailedCheckrun(checkrun) {
-				if v.canIUseCheckrunID(checkrun.ID) {
-					return checkrun.ID, nil
-				}
-			}
-			if *checkrun.ExternalID == status.PipelineRunName {
-				return checkrun.ID, nil
-			}
-		}
+		all = append(all, res.CheckRuns...)
 		if resp.NextPage == 0 {
 			break
 		}
 		opt.Page = resp.NextPage
 	}
+	return all, nil
+}
 
-	return nil, nil
+func (v *Provider) getExistingCheckRunID(ctx context.Context, runevent *info.Event, status providerstatus.StatusOpts) (*int64, error) {
+	v.checkRunsCache.once.Do(func() {
+		v.checkRunsCache.runs, v.checkRunsCache.err = v.fetchAllCheckRunPages(ctx, runevent)
+	})
+	if v.checkRunsCache.err != nil {
+		return nil, v.checkRunsCache.err
+	}
+	return searchCheckRuns(v.checkRunsCache.runs, status, v), nil
+}
+
+func searchCheckRuns(runs []*github.CheckRun, status providerstatus.StatusOpts, v *Provider) *int64 {
+	for _, checkrun := range runs {
+		if isPendingApprovalCheckrun(checkrun) || isFailedCheckrun(checkrun) {
+			if v.canIUseCheckrunID(checkrun.ID) {
+				return checkrun.ID
+			}
+		}
+		if checkrun.ExternalID != nil && *checkrun.ExternalID == status.PipelineRunName {
+			return checkrun.ID
+		}
+	}
+	return nil
 }
 
 func isPendingApprovalCheckrun(run *github.CheckRun) bool {

pkg/provider/github/status_test.go

@@ -8,6 +8,8 @@ import (
 	"reflect"
 	"strconv"
 	"strings"
+	"sync"
+	"sync/atomic"
 	"testing"
 
 	"github.com/google/go-github/v84/github"
@@ -686,3 +688,71 @@ func TestProviderGetExistingCheckRunID(t *testing.T) {
 		})
 	}
 }
+
+func TestGetExistingCheckRunIDCacheHit(t *testing.T) {
+	ctx, _ := rtesting.SetupFakeContext(t)
+	client, mux, _, teardown := ghtesthelper.SetupGH()
+	defer teardown()
+
+	event := &info.Event{
+		Organization: "owner",
+		Repository:   "repository",
+		SHA:          "sha",
+	}
+
+	apiHits := 0
+	mux.HandleFunc(fmt.Sprintf("/repos/%v/%v/commits/%v/check-runs", event.Organization, event.Repository, event.SHA), func(w http.ResponseWriter, _ *http.Request) {
+		apiHits++
+		fmt.Fprint(w, `{"total_count": 1, "check_runs": [{"id": 55555, "external_id": "mypr"}]}`)
+	})
+
+	cnx := New()
+	cnx.SetGithubClient(client)
+
+	// First call — should hit the API and populate the cache.
+	id, err := cnx.getExistingCheckRunID(ctx, event, providerstatus.StatusOpts{PipelineRunName: "mypr"})
+	assert.NilError(t, err)
+	assert.Assert(t, id != nil)
+
+	// Second call — should serve from cache, no additional API call.
+	id2, err := cnx.getExistingCheckRunID(ctx, event, providerstatus.StatusOpts{PipelineRunName: "mypr"})
+	assert.NilError(t, err)
+	assert.Assert(t, id2 != nil)
+	assert.Equal(t, *id, *id2)
+	assert.Equal(t, apiHits, 1)
+}
+
+func TestGetExistingCheckRunIDConcurrent(t *testing.T) {
+	ctx, _ := rtesting.SetupFakeContext(t)
+	client, mux, _, teardown := ghtesthelper.SetupGH()
+	defer teardown()
+
+	event := &info.Event{
+		Organization: "owner",
+		Repository:   "repository",
+		SHA:          "sha",
+	}
+
+	var apiHits atomic.Int64
+	mux.HandleFunc(fmt.Sprintf("/repos/%v/%v/commits/%v/check-runs", event.Organization, event.Repository, event.SHA), func(w http.ResponseWriter, _ *http.Request) {
+		apiHits.Add(1)
+		fmt.Fprint(w, `{"total_count": 1, "check_runs": [{"id": 55555, "external_id": "mypr"}]}`)
+	})
+
+	cnx := New()
+	cnx.SetGithubClient(client)
+
+	const goroutines = 10
+	var wg sync.WaitGroup
+	wg.Add(goroutines)
+	for range goroutines {
+		go func() {
+			defer wg.Done()
+			_, _ = cnx.getExistingCheckRunID(ctx, event, providerstatus.StatusOpts{PipelineRunName: "mypr"})
+		}()
+	}
+	wg.Wait()
+
+	// Only one goroutine should have fetched from the API; the rest hit the cache.
+	assert.Equal(t, apiHits.Load(), int64(1))
+}