Skip to content

fix: list GitHub App installations instead of per-org lookups #446

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
joshfriend merged 1 commit into from
Nov 11, 2026
Merged

fix: list GitHub App installations instead of per-org lookups #446

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 6 additions & 5 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -290,7 +290,8 @@ github-app {
}
```

Installations can also be discovered dynamically via the GitHub API.
Each App lists its installations via `GET /app/installations` at startup and every five minutes. Orgs matching no
installation (case-insensitively) use the `fallback-org` token without further API calls.

## CLI

Expand Down Expand Up @@ -336,7 +337,7 @@ metrics {
Admin endpoints: `/_liveness`, `/_readiness`, `PUT /admin/log/level`, `/admin/pprof/`.

GitHub App installations are checked at startup and every minute via `/rate_limit`.
The poller includes dynamically discovered installations and the configured fallback organization.
The poller includes every listed installation and the configured fallback organization.

| Metric | Type | Labels |
| --- | --- | --- |
Expand All @@ -346,9 +347,9 @@ The poller includes dynamically discovered installations and the configured fall
| `github.api_calls` | Counter | `owner`, `operation` |

`app` is the configured App ID. `owner` identifies the installation actually used,
including fallback credentials. Installation discovery uses `owner=unknown` because
it authenticates as the App before resolving an installation. API calls for release
metadata/assets, installation discovery, token creation and quota polling are counted. Responses with less than
including fallback credentials. Installation listing uses `owner=unknown` because
it authenticates as the App rather than an installation. API calls for release
metadata/assets, installation listing, token creation and quota polling are counted. Responses with less than
10% of their API quota remaining produce a warning. Static release tokens receive
API call counts and warnings, but are not polled for quota gauges.

Expand Down
3 changes: 3 additions & 0 deletions cmd/cachewd/main.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -393,6 +393,9 @@ func startMetrics(ctx context.Context, client *metrics.Client, provider githubap
return errors.Wrap(err, "create GitHub App token manager")
}
if manager != nil {
go manager.RunInstallationRefresh(ctx, func(err error) {
logging.FromContext(ctx).WarnContext(ctx, "Failed to refresh GitHub App installations", "error", err)
})
go githubapp.NewRateLimitPoller(manager, 0).Run(ctx, func(err error) {
logging.FromContext(ctx).WarnContext(ctx, "Failed to poll GitHub rate limit", "error", err)
})
Expand Down
6 changes: 3 additions & 3 deletions internal/githubapp/poller_external_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,8 +18,8 @@ import (
func TestRateLimitPollerBootstrapsFallback(t *testing.T) {
polled := make(chan struct{})
mux := http.NewServeMux()
mux.HandleFunc("GET /orgs/squareup/installation", func(w http.ResponseWriter, _ *http.Request) {
_, _ = fmt.Fprint(w, `{"id":123}`)
mux.HandleFunc("GET /app/installations", func(w http.ResponseWriter, _ *http.Request) {
_, _ = fmt.Fprint(w, `[{"id":123,"account":{"login":"block"}}]`)
})
mux.HandleFunc("POST /app/installations/123/access_tokens", func(w http.ResponseWriter, _ *http.Request) {
w.WriteHeader(http.StatusCreated)
Expand All @@ -33,7 +33,7 @@ func TestRateLimitPollerBootstrapsFallback(t *testing.T) {
server := httptest.NewServer(mux)
defer server.Close()
manager := githubapp.NewTokenManagerForTest(t, []githubapp.Config{{}, {
AppID: "111", PrivateKeyPath: generateTestKey(t), FallbackOrg: "squareup",
AppID: "111", PrivateKeyPath: generateTestKey(t), FallbackOrg: "block",
}}, slog.Default(), server.URL, server.Client())
ctx, cancel := context.WithCancel(logging.ContextWithLogger(t.Context(), slog.Default()))
defer cancel()
Expand Down
59 changes: 17 additions & 42 deletions internal/githubapp/ratelimit_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -60,11 +60,10 @@ func collectMetrics(t *testing.T, reader *sdkmetric.ManualReader) map[string][]m
func cachedTestApp(server *httptest.Server, appID, owner, installationID string) *appState {
app := &appState{
appID: appID, apiBase: server.URL, httpClient: NewAPIClient(server.Client()),
cacheConfig: DefaultTokenCacheConfig(), installationCache: make(map[string]string),
installationOwners: make(map[string]string),
tokens: map[string]*cachedToken{installationID: {token: "token-" + appID, expiresAt: time.Now().Add(time.Hour)}},
cacheConfig: DefaultTokenCacheConfig(),
tokens: map[string]*cachedToken{installationID: {token: "token-" + appID, expiresAt: time.Now().Add(time.Hour)}},
}
app.rememberInstallation(owner, installationID, owner)
app.setInstallations(map[string]string{installationID: owner})
return app
}

Expand All @@ -80,10 +79,9 @@ func TestRateLimitPollerSamplesInstallations(t *testing.T) {
_, _ = fmt.Fprint(w, `{"resources":{"core":{"remaining":4900,"limit":5000,"used":100},"search":{"remaining":30,"limit":30,"used":0}}}`)
}))
defer server.Close()
app := cachedTestApp(server, "111", "squareup", "123")
app.rememberInstallation("alias", "123", "alias")
otherApp := cachedTestApp(server, "222", "block", "456")
manager := &TokenManager{apps: []*appState{app, otherApp}, fallbackApp: app, fallbackOrg: "squareup"}
app := cachedTestApp(server, "111", "block", "123")
otherApp := cachedTestApp(server, "222", "example-org", "456")
manager := &TokenManager{apps: []*appState{app, otherApp}, fallbackApp: app, fallbackOrg: "block"}
assert.NoError(t, NewRateLimitPoller(manager, 0).poll(pollerContext(t)))
assert.Equal(t, int64(2), calls.Load())
data := collectMetrics(t, reader)
Expand All @@ -101,11 +99,11 @@ func TestRateLimitPollerSamplesInstallations(t *testing.T) {
owner, _ := point.Attributes.Value("owner")
if appID.AsString() == "111" {
assert.Equal(t, "123", installation.AsString())
assert.Equal(t, "squareup", owner.AsString())
assert.Equal(t, "block", owner.AsString())
} else {
assert.Equal(t, "222", appID.AsString())
assert.Equal(t, "456", installation.AsString())
assert.Equal(t, "block", owner.AsString())
assert.Equal(t, "example-org", owner.AsString())
}
}
}
Expand All @@ -117,11 +115,11 @@ func TestRateLimitPollerFindsNewInstallations(t *testing.T) {
_, _ = fmt.Fprint(w, `{"resources":{"core":{"remaining":50,"limit":100,"used":50}}}`)
}))
defer server.Close()
app := cachedTestApp(server, "111", "squareup", "123")
app := cachedTestApp(server, "111", "block", "123")
poller := NewRateLimitPoller(&TokenManager{apps: []*appState{app}}, 0)
assert.NoError(t, poller.poll(pollerContext(t)))
assert.Equal(t, 1, len(collectMetrics(t, reader)["github.ratelimit_remaining"]))
app.rememberInstallation("block", "456", "block")
app.setInstallations(map[string]string{"123": "block", "456": "example-org"})
app.mu.Lock()
app.tokens["456"] = &cachedToken{token: "later-token", expiresAt: time.Now().Add(time.Hour)}
app.mu.Unlock()
Expand All @@ -139,41 +137,18 @@ func TestRateLimitPrometheusExport(t *testing.T) {
_, _ = fmt.Fprint(w, `{"resources":{"core":{"remaining":42,"limit":100,"used":58}}}`)
}))
defer server.Close()
app := cachedTestApp(server, "111", "squareup", "123")
app := cachedTestApp(server, "111", "block", "123")
assert.NoError(t, NewRateLimitPoller(&TokenManager{apps: []*appState{app}}, 0).poll(pollerContext(t)))
response := httptest.NewRecorder()
client.Handler().ServeHTTP(response, httptest.NewRequestWithContext(t.Context(), http.MethodGet, "/metrics", nil))
assert.Equal(t, http.StatusOK, response.Code)
assert.Contains(t, response.Body.String(), `github_ratelimit_remaining{`)
assert.Contains(t, response.Body.String(), `app="111"`)
assert.Contains(t, response.Body.String(), `installation_id="123"`)
assert.Contains(t, response.Body.String(), `owner="squareup"`)
assert.Contains(t, response.Body.String(), `owner="block"`)
assert.Contains(t, response.Body.String(), `resource="core"`)
}

func TestRateLimitPollerRetainsOwnerAcrossAliases(t *testing.T) {
reader := newMetricReader(t)
server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) {
_, _ = fmt.Fprint(w, `{"resources":{"core":{"remaining":42,"limit":100,"used":58}}}`)
}))
defer server.Close()
app := cachedTestApp(server, "111", "example-org", "123")
poller := NewRateLimitPoller(&TokenManager{apps: []*appState{app}}, 0)
for _, alias := range []string{"Example-Org", "old-name", "renamed-org"} {
app.rememberInstallation(alias, "123", alias)
token, installation, err := app.getToken(pollerContext(t), alias)
assert.NoError(t, err)
assert.Equal(t, "token-111", token)
assert.Equal(t, Installation{ID: "123", Owner: "example-org"}, installation)
assert.Equal(t, map[string]string{"123": "example-org"}, app.knownInstallations())
assert.NoError(t, poller.poll(pollerContext(t)))
points := collectMetrics(t, reader)["github.ratelimit_remaining"]
assert.Equal(t, 1, len(points))
assert.Equal(t, attribute.NewSet(attribute.String("app", "111"), attribute.String("owner", "example-org"),
attribute.String("installation_id", "123"), attribute.String("resource", "core")), points[0].Attributes)
}
}

func TestRateLimitPollerErrorsDoNotEmitZeroes(t *testing.T) {
for _, tc := range []struct {
name string
Expand All @@ -190,7 +165,7 @@ func TestRateLimitPollerErrorsDoNotEmitZeroes(t *testing.T) {
_, _ = fmt.Fprint(w, tc.body)
}))
defer server.Close()
app := cachedTestApp(server, "111", "squareup", "123")
app := cachedTestApp(server, "111", "block", "123")
poller := NewRateLimitPoller(&TokenManager{apps: []*appState{app}}, 0)
assert.Error(t, poller.poll(pollerContext(t)))
assert.Zero(t, len(collectMetrics(t, reader)["github.ratelimit_remaining"]))
Expand All @@ -209,13 +184,13 @@ func TestRateLimitPollerContinuesAfterFailure(t *testing.T) {
}))
defer server.Close()
manager := &TokenManager{apps: []*appState{
cachedTestApp(server, "111", "squareup", "123"), cachedTestApp(server, "222", "block", "456"),
cachedTestApp(server, "111", "block", "123"), cachedTestApp(server, "222", "example-org", "456"),
}}
assert.Error(t, NewRateLimitPoller(manager, 0).poll(pollerContext(t)))
points := collectMetrics(t, reader)["github.ratelimit_remaining"]
assert.Equal(t, 1, len(points))
assert.Equal(t, int64(42), points[0].Value)
assert.Equal(t, attribute.NewSet(attribute.String("app", "222"), attribute.String("owner", "block"),
assert.Equal(t, attribute.NewSet(attribute.String("app", "222"), attribute.String("owner", "example-org"),
attribute.String("installation_id", "456"), attribute.String("resource", "core")), points[0].Attributes)
}

Expand All @@ -232,7 +207,7 @@ func TestRateLimitPollerRun(t *testing.T) {
defer server.Close()
ctx, cancel := context.WithTimeout(pollerContext(t), time.Second)
defer cancel()
poller := NewRateLimitPoller(&TokenManager{apps: []*appState{cachedTestApp(server, "111", "squareup", "123")}}, time.Millisecond)
poller := NewRateLimitPoller(&TokenManager{apps: []*appState{cachedTestApp(server, "111", "block", "123")}}, time.Millisecond)
done := make(chan struct{})
go func() {
poller.Run(ctx, func(err error) { t.Error(err) })
Expand Down Expand Up @@ -307,7 +282,7 @@ func TestRateLimitPollerCancelsRequest(t *testing.T) {
defer server.Close()
ctx, cancel := context.WithCancel(pollerContext(t))
defer cancel()
poller := NewRateLimitPoller(&TokenManager{apps: []*appState{cachedTestApp(server, "111", "squareup", "123")}}, 0)
poller := NewRateLimitPoller(&TokenManager{apps: []*appState{cachedTestApp(server, "111", "block", "123")}}, 0)
done := make(chan struct{})
go func() {
poller.Run(ctx, func(err error) { t.Error(err) })
Expand Down
Loading
Loading