fix: more proper setup and lots of quality of life changes (#10)

.github/workflows/pr-check.yml

@@ -5,20 +5,14 @@ on:
     branches: [main]
 
 jobs:
-  build:
+  test:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
 
-      - name: Set up QEMU
-        uses: docker/setup-qemu-action@v3
-
-      - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v3
-
-      - name: Build
-        uses: docker/build-push-action@v6
+      - uses: actions/setup-go@v5
         with:
-          context: .
-          push: false
-          platforms: linux/amd64,linux/arm64
+          go-version-file: go.mod
+
+      - name: Test
+        run: go test -race ./...

Dockerfile

@@ -1,4 +1,4 @@
-FROM golang:1.22-alpine AS builder
+FROM golang:1.24-alpine AS builder
 WORKDIR /src
 COPY go.mod go.sum ./
 RUN go mod download

README.md

@@ -63,6 +63,20 @@ docker run -d \
   rwejlgaard/octopus_exporter
 ```
 
+## Docker Compose
+
+```yaml
+services:
+  octopus-exporter:
+    image: rwejlgaard/octopus_exporter:latest
+    restart: unless-stopped
+    environment:
+      OCTOPUS_API_KEY: sk_live_...
+      OCTOPUS_MPAN: "1234567890123"   # optional, auto-discovered if omitted
+    ports:
+      - "9359:9359"
+```
+
 ## Running from source
 
 Requires Go 1.22+.

cmd/octopus_exporter/agile.go

@@ -9,7 +9,7 @@ import (
 
 // getCurrentAgileRate returns the unit rate (inc. VAT, pence/kWh) for the current
 // half-hour slot from the Agile tariff REST endpoint.
-func getCurrentAgileRate(productCode, tariffCode string) (float64, error) {
+func getCurrentAgileRate(productCode, tariffCode, key string) (float64, error) {
 	now := time.Now().UTC()
 	slotStart := now.Truncate(30 * time.Minute)
 	slotEnd := slotStart.Add(30 * time.Minute)
@@ -18,7 +18,7 @@ func getCurrentAgileRate(productCode, tariffCode string) (float64, error) {
 	result, err := doREST(path, url.Values{
 		"period_from": {slotStart.Format(time.RFC3339)},
 		"period_to":   {slotEnd.Format(time.RFC3339)},
-	})
+	}, key)
 	if err != nil {
 		return 0, err
 	}

cmd/octopus_exporter/agile_test.go

@@ -15,9 +15,8 @@ func TestGetCurrentAgileRate_Success(t *testing.T) {
 	}))
 	defer srv.Close()
 	octopusREST = srv.URL
-	apiKey = "test"
 
-	rate, err := getCurrentAgileRate("AGILE-24-10-01", "E-1R-AGILE-24-10-01-C")
+	rate, err := getCurrentAgileRate("AGILE-24-10-01", "E-1R-AGILE-24-10-01-C", "test")
 	if err != nil {
 		t.Fatalf("unexpected error: %v", err)
 	}
@@ -32,9 +31,8 @@ func TestGetCurrentAgileRate_NoSlot(t *testing.T) {
 	}))
 	defer srv.Close()
 	octopusREST = srv.URL
-	apiKey = "test"
 
-	_, err := getCurrentAgileRate("AGILE-24-10-01", "E-1R-AGILE-24-10-01-C")
+	_, err := getCurrentAgileRate("AGILE-24-10-01", "E-1R-AGILE-24-10-01-C", "test")
 	if err == nil {
 		t.Error("expected error for empty slot, got nil")
 	}
@@ -48,9 +46,8 @@ func TestGetCurrentAgileRate_CorrectPath(t *testing.T) {
 	}))
 	defer srv.Close()
 	octopusREST = srv.URL
-	apiKey = "test"
 
-	getCurrentAgileRate("AGILE-24-10-01", "E-1R-AGILE-24-10-01-C")
+	getCurrentAgileRate("AGILE-24-10-01", "E-1R-AGILE-24-10-01-C", "test")
 
 	want := "/v1/products/AGILE-24-10-01/electricity-tariffs/E-1R-AGILE-24-10-01-C/standard-unit-rates/"
 	if capturedPath != want {

cmd/octopus_exporter/client.go

@@ -6,11 +6,20 @@ import (
 	"errors"
 	"fmt"
 	"io"
+	"log"
 	"net/http"
 	"strconv"
+	"strings"
+	"time"
+
+	"github.com/prometheus/client_golang/prometheus"
 )
 
-var octopusGraphQL = "https://api.octopus.energy/v1/graphql/"
+var (
+	octopusGraphQL   = "https://api.octopus.energy/v1/graphql/"
+	httpClient       = &http.Client{Timeout: 15 * time.Second}
+	rateLimitRetries prometheus.Counter
+)
 
 type gqlRequest struct {
 	OperationName string         `json:"operationName,omitempty"`
@@ -39,11 +48,62 @@ func (f *jsonFloat) UnmarshalJSON(data []byte) error {
 	return nil
 }
 
+// executeWithRetry executes an HTTP request, retrying on 429 with exponential
+// backoff (honouring Retry-After when present). Returns the raw response body.
+func executeWithRetry(makeReq func() (*http.Request, error)) ([]byte, error) {
+	const maxRetries = 5
+	backoff := 30 * time.Second
+	for attempt := 0; attempt <= maxRetries; attempt++ {
+		req, err := makeReq()
+		if err != nil {
+			return nil, err
+		}
+		resp, err := httpClient.Do(req)
+		if err != nil {
+			return nil, err
+		}
+		if resp.StatusCode == http.StatusTooManyRequests {
+			resp.Body.Close()
+			if attempt == maxRetries {
+				return nil, errors.New("rate limited: max retries exceeded")
+			}
+			if rateLimitRetries != nil {
+				rateLimitRetries.Inc()
+			}
+			wait := backoff
+			if ra := resp.Header.Get("Retry-After"); ra != "" {
+				if secs, err := strconv.Atoi(ra); err == nil {
+					wait = time.Duration(secs) * time.Second
+				}
+			}
+			log.Printf("rate limited; retrying in %v (attempt %d/%d)", wait, attempt+1, maxRetries)
+			time.Sleep(wait)
+			backoff *= 2
+			continue
+		}
+		raw, err := io.ReadAll(resp.Body)
+		resp.Body.Close()
+		if err != nil {
+			return nil, err
+		}
+		if resp.StatusCode < 200 || resp.StatusCode >= 300 {
+			snippet := strings.TrimSpace(string(raw))
+			if len(snippet) > 200 {
+				snippet = snippet[:200]
+			}
+			return nil, fmt.Errorf("HTTP %d: %s", resp.StatusCode, snippet)
+		}
+		return raw, nil
+	}
+	return nil, errors.New("rate limited: max retries exceeded")
+}
+
 func doGraphQL(req gqlRequest, authToken string) (map[string]any, error) {
 	body, err := json.Marshal(req)
 	if err != nil {
 		return nil, err
 	}
+	raw, err := executeWithRetry(func() (*http.Request, error) {
 		httpReq, err := http.NewRequest(http.MethodPost, octopusGraphQL, bytes.NewReader(body))
 		if err != nil {
 			return nil, err
@@ -52,12 +112,8 @@ func doGraphQL(req gqlRequest, authToken string) (map[string]any, error) {
 		if authToken != "" {
 			httpReq.Header.Set("Authorization", "JWT "+authToken)
 		}
-	resp, err := http.DefaultClient.Do(httpReq)
-	if err != nil {
-		return nil, err
-	}
-	defer resp.Body.Close()
-	raw, err := io.ReadAll(resp.Body)
+		return httpReq, nil
+	})
 	if err != nil {
 		return nil, err
 	}
@@ -67,7 +123,11 @@ func doGraphQL(req gqlRequest, authToken string) (map[string]any, error) {
 	}
 	if errs, ok := result["errors"].([]any); ok && len(errs) > 0 {
 		if e, ok := errs[0].(map[string]any); ok {
-			return nil, fmt.Errorf("GraphQL error: %s", e["message"])
+			msg, _ := e["message"].(string)
+			if strings.Contains(msg, "JWT") && strings.Contains(msg, "expired") {
+				return nil, errTokenExpired
+			}
+			return nil, fmt.Errorf("GraphQL error: %s", msg)
 		}
 		return nil, errors.New("GraphQL error")
 	}

cmd/octopus_exporter/client_test.go

@@ -2,6 +2,11 @@ package main
 
 import (
 	"encoding/json"
+	"fmt"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"sync/atomic"
 	"testing"
 )
 
@@ -51,3 +56,66 @@ func TestToSlice_WrongType(t *testing.T) {
 		t.Errorf("expected nil for wrong type, got %v", got)
 	}
 }
+
+func TestExecuteWithRetry_RateLimitRetry(t *testing.T) {
+	var attempts atomic.Int32
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		n := attempts.Add(1)
+		if n < 3 {
+			w.Header().Set("Retry-After", "0")
+			w.WriteHeader(http.StatusTooManyRequests)
+			return
+		}
+		fmt.Fprint(w, `ok`)
+	}))
+	defer srv.Close()
+
+	raw, err := executeWithRetry(func() (*http.Request, error) {
+		return http.NewRequest(http.MethodGet, srv.URL, nil)
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if string(raw) != "ok" {
+		t.Errorf("got body %q, want %q", string(raw), "ok")
+	}
+	if attempts.Load() != 3 {
+		t.Errorf("got %d attempts, want 3", attempts.Load())
+	}
+}
+
+func TestExecuteWithRetry_Non200Error(t *testing.T) {
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		w.WriteHeader(http.StatusUnauthorized)
+		fmt.Fprint(w, `{"error":"unauthorized"}`)
+	}))
+	defer srv.Close()
+
+	_, err := executeWithRetry(func() (*http.Request, error) {
+		return http.NewRequest(http.MethodGet, srv.URL, nil)
+	})
+	if err == nil {
+		t.Error("expected error for 401, got nil")
+	}
+	if !strings.Contains(err.Error(), "401") {
+		t.Errorf("expected error to contain status code, got: %v", err)
+	}
+}
+
+func TestExecuteWithRetry_ServerError(t *testing.T) {
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		w.WriteHeader(http.StatusInternalServerError)
+		fmt.Fprint(w, `internal server error`)
+	}))
+	defer srv.Close()
+
+	_, err := executeWithRetry(func() (*http.Request, error) {
+		return http.NewRequest(http.MethodGet, srv.URL, nil)
+	})
+	if err == nil {
+		t.Error("expected error for 500, got nil")
+	}
+	if !strings.Contains(err.Error(), "500") {
+		t.Errorf("expected error to contain status code, got: %v", err)
+	}
+}

cmd/octopus_exporter/consumption.go

@@ -12,7 +12,7 @@ type consumptionReading struct {
 	IntervalStart time.Time
 }
 
-func getLatestConsumption(kind meterKind, id, serial string) (*consumptionReading, error) {
+func getLatestConsumption(kind meterKind, id, serial, key string) (*consumptionReading, error) {
 	var path string
 	switch kind {
 	case electricity:
@@ -25,7 +25,7 @@ func getLatestConsumption(kind meterKind, id, serial string) (*consumptionReadin
 	result, err := doREST(path, url.Values{
 		"period_from": {time.Now().UTC().Add(-24 * time.Hour).Format(time.RFC3339)},
 		"order_by":    {"period"},
-	})
+	}, key)
 	if err != nil {
 		return nil, err
 	}
@@ -35,9 +35,16 @@ func getLatestConsumption(kind meterKind, id, serial string) (*consumptionReadin
 		return nil, errors.New("no consumption data in last 24h")
 	}
 
-	latest := results[len(results)-1].(map[string]any)
+	latest, ok := results[len(results)-1].(map[string]any)
+	if !ok {
+		return nil, errors.New("unexpected API response: invalid result entry")
+	}
 	kwh, _ := latest["consumption"].(float64)
-	start, err := time.Parse(time.RFC3339, latest["interval_start"].(string))
+	startStr, ok := latest["interval_start"].(string)
+	if !ok {
+		return nil, errors.New("unexpected API response: missing interval_start")
+	}
+	start, err := time.Parse(time.RFC3339, startStr)
 	if err != nil {
 		return nil, fmt.Errorf("failed to parse interval_start: %w", err)
 	}

cmd/octopus_exporter/consumption_test.go

@@ -16,9 +16,8 @@ func TestGetLatestConsumption_ReturnsLatestInterval(t *testing.T) {
 	}))
 	defer srv.Close()
 	octopusREST = srv.URL
-	apiKey = "test"
 
-	c, err := getLatestConsumption(electricity, "MPAN123", "SERIAL456")
+	c, err := getLatestConsumption(electricity, "MPAN123", "SERIAL456", "test")
 	if err != nil {
 		t.Fatalf("unexpected error: %v", err)
 	}
@@ -33,9 +32,8 @@ func TestGetLatestConsumption_Empty(t *testing.T) {
 	}))
 	defer srv.Close()
 	octopusREST = srv.URL
-	apiKey = "test"
 
-	_, err := getLatestConsumption(electricity, "MPAN123", "SERIAL456")
+	_, err := getLatestConsumption(electricity, "MPAN123", "SERIAL456", "test")
 	if err == nil {
 		t.Error("expected error for empty results, got nil")
 	}
@@ -51,12 +49,25 @@ func TestGetLatestConsumption_GasPath(t *testing.T) {
 	}))
 	defer srv.Close()
 	octopusREST = srv.URL
-	apiKey = "test"
 
-	getLatestConsumption(gas, "MPRN789", "SERIAL456")
+	getLatestConsumption(gas, "MPRN789", "SERIAL456", "test")
 
 	want := "/v1/gas-meter-points/MPRN789/meters/SERIAL456/consumption/"
 	if capturedPath != want {
 		t.Errorf("got path %q, want %q", capturedPath, want)
 	}
 }
+
+func TestGetLatestConsumption_Non200Error(t *testing.T) {
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		w.WriteHeader(http.StatusUnauthorized)
+		fmt.Fprint(w, `{"error":"unauthorized"}`)
+	}))
+	defer srv.Close()
+	octopusREST = srv.URL
+
+	_, err := getLatestConsumption(electricity, "MPAN123", "SERIAL456", "bad-key")
+	if err == nil {
+		t.Error("expected error for 401, got nil")
+	}
+}

cmd/octopus_exporter/main.go

@@ -1,9 +1,12 @@
 package main
 
 import (
+	"errors"
 	"log"
 	"net/http"
 	"os"
+	"sync"
+	"sync/atomic"
 	"time"
 
 	"github.com/prometheus/client_golang/prometheus"
@@ -34,6 +37,10 @@ func gauge(name, help string) prometheus.Gauge {
 	return prometheus.NewGauge(prometheus.GaugeOpts{Name: name, Help: help})
 }
 
+func counter(name, help string) prometheus.Counter {
+	return prometheus.NewCounter(prometheus.CounterOpts{Name: name, Help: help})
+}
+
 func main() {
 	apiKey = mustEnv("OCTOPUS_API_KEY")
 	port = envOrDefault("PORT", "9359")
@@ -43,7 +50,13 @@ func main() {
 		log.Fatalf("failed to get initial token: %v", err)
 	}
 
-	elecMeter, err := resolveMeter(token, electricity)
+	log.Println("discovering meters from account...")
+	candidates, err := getMeters(token)
+	if err != nil {
+		log.Fatalf("failed to discover meters: %v", err)
+	}
+
+	elecMeter, err := resolveMeter(candidates, electricity)
 	if err != nil {
 		log.Fatalf("failed to resolve electricity meter: %v", err)
 	}
@@ -51,7 +64,7 @@ func main() {
 		log.Fatal("no electricity smart meter found on account")
 	}
 
-	gasMeter, err := resolveMeter(token, gas)
+	gasMeter, err := resolveMeter(candidates, gas)
 	if err != nil {
 		log.Fatalf("failed to resolve gas meter: %v", err)
 	}
@@ -76,11 +89,18 @@ func main() {
 	// Account
 	accountBalance := gauge("octopus_account_balance_pence", "Account balance in pence (positive = credit, negative = debit)")
 
+	// Exporter health
+	exporterUp := gauge("octopus_up", "1 if the last poll cycle completed without errors, 0 otherwise")
+	pollErrors := counter("octopus_poll_errors_total", "Total number of collector errors per poll cycle")
+	tokenRefreshCount := counter("octopus_token_refreshes_total", "Total number of successful JWT token refreshes")
+	rateLimitRetries = counter("octopus_rate_limit_retries_total", "Total number of 429 rate-limit retries across all requests")
+
 	toRegister := []prometheus.Collector{
 		elecDemand, elecLastRead,
 		elecConsumption, elecConsumptionInterval,
 		elecUnitRate, elecStandingCharge,
 		accountBalance,
+		exporterUp, pollErrors, tokenRefreshCount, rateLimitRetries,
 	}
 
 	var (
@@ -111,87 +131,170 @@ func main() {
 		}
 	}()
 
+	// tokenMu guards token across concurrent poll goroutines.
+	var tokenMu sync.RWMutex
+
+	// withToken calls fn with the current token, refreshing once on JWT expiry.
+	withToken := func(fn func(string) error) error {
+		tokenMu.RLock()
+		t := token
+		tokenMu.RUnlock()
+
+		err := fn(t)
+		if !errors.Is(err, errTokenExpired) {
+			return err
+		}
+
+		// Only one goroutine refreshes; others will pick up the new token.
+		tokenMu.Lock()
+		if token == t {
+			newT, e := getKrakenToken(apiKey)
+			if e != nil {
+				tokenMu.Unlock()
+				log.Printf("token refresh failed: %v", e)
+				return err
+			}
+			token = newT
+			tokenRefreshCount.Inc()
+		}
+		newT := token
+		tokenMu.Unlock()
+
+		return fn(newT)
+	}
+
 	for {
+		var (
+			wg        sync.WaitGroup
+			failedAny atomic.Bool
+		)
+
+		fail := func(format string, args ...any) {
+			log.Printf(format, args...)
+			pollErrors.Inc()
+			failedAny.Store(true)
+		}
+
+		collect := func(name string, fn func() error) {
+			wg.Add(1)
+			go func() {
+				defer wg.Done()
+				if err := fn(); err != nil {
+					fail("%s error: %v", name, err)
+				}
+			}()
+		}
+
 		// Electricity telemetry (live demand)
 		if elecMeter.deviceID != "" {
-			reading, err := getLiveConsumption(token, elecMeter.deviceID)
+			collect("electricity telemetry", func() error {
+				return withToken(func(t string) error {
+					reading, err := getLiveConsumption(t, elecMeter.deviceID)
 					if err != nil {
-				log.Printf("electricity telemetry error: %v", err)
-				if token, err = getKrakenToken(apiKey); err != nil {
-					log.Printf("token refresh failed: %v", err)
+						return err
 					}
-			} else {
 					elecDemand.Set(float64(reading.Demand))
-				if t, err := time.Parse("2006-01-02T15:04:05+00:00", reading.ReadAt); err == nil {
-					elecLastRead.Set(float64(t.Unix()))
-				}
+					if ts, err := time.Parse(time.RFC3339, reading.ReadAt); err == nil {
+						elecLastRead.Set(float64(ts.Unix()))
 					}
+					return nil
+				})
+			})
 		}
 
 		// Gas telemetry (live demand)
 		if gasMeter != nil && gasMeter.deviceID != "" {
-			reading, err := getLiveConsumption(token, gasMeter.deviceID)
+			collect("gas telemetry", func() error {
+				return withToken(func(t string) error {
+					reading, err := getLiveConsumption(t, gasMeter.deviceID)
 					if err != nil {
-				log.Printf("gas telemetry error: %v", err)
-			} else {
+						return err
+					}
 					gasDemand.Set(float64(reading.Demand))
-				if t, err := time.Parse("2006-01-02T15:04:05+00:00", reading.ReadAt); err == nil {
-					gasLastRead.Set(float64(t.Unix()))
-				}
+					if ts, err := time.Parse(time.RFC3339, reading.ReadAt); err == nil {
+						gasLastRead.Set(float64(ts.Unix()))
 					}
+					return nil
+				})
+			})
 		}
 
 		// Electricity half-hourly consumption (REST)
 		if elecMeter.mpan != "" && elecMeter.serial != "" {
-			c, err := getLatestConsumption(electricity, elecMeter.mpan, elecMeter.serial)
+			collect("electricity consumption", func() error {
+				c, err := getLatestConsumption(electricity, elecMeter.mpan, elecMeter.serial, apiKey)
 				if err != nil {
-				log.Printf("electricity consumption error: %v", err)
-			} else {
+					return err
+				}
 				elecConsumption.Set(c.KWh)
 				elecConsumptionInterval.Set(float64(c.IntervalStart.Unix()))
-			}
+				return nil
+			})
 		}
 
 		// Gas half-hourly consumption (REST)
 		if gasMeter != nil && gasMeter.mprn != "" && gasMeter.serial != "" {
-			c, err := getLatestConsumption(gas, gasMeter.mprn, gasMeter.serial)
+			collect("gas consumption", func() error {
+				c, err := getLatestConsumption(gas, gasMeter.mprn, gasMeter.serial, apiKey)
 				if err != nil {
-				log.Printf("gas consumption error: %v", err)
-			} else {
+					return err
+				}
 				gasConsumption.Set(c.KWh)
 				gasConsumptionInterval.Set(float64(c.IntervalStart.Unix()))
-			}
+				return nil
+			})
 		}
 
-		// Tariff rates
-		rates, err := getRates(token)
+		// Tariff rates (result needed for optional agile lookup after wg.Wait)
+		var collectedRates *tariffRates
+		collect("rates", func() error {
+			return withToken(func(t string) error {
+				r, err := getRates(t)
 				if err != nil {
-			log.Printf("rates error: %v", err)
-		} else {
-			unitRate := rates.ElectricityUnitRate
-			if rates.ElectricityIsAgile && rates.ElectricityProductCode != "" && rates.ElectricityTariffCode != "" {
-				agileRate, err := getCurrentAgileRate(rates.ElectricityProductCode, rates.ElectricityTariffCode)
+					return err
+				}
+				collectedRates = r
+				return nil
+			})
+		})
+
+		// Account balance
+		collect("account balance", func() error {
+			return withToken(func(t string) error {
+				balance, err := getAccountBalance(t)
 				if err != nil {
-					log.Printf("agile rate error: %v", err)
+					return err
+				}
+				accountBalance.Set(balance)
+				return nil
+			})
+		})
+
+		wg.Wait()
+
+		// Agile rate depends on the rates result, so it runs after the parallel phase.
+		if collectedRates != nil {
+			unitRate := collectedRates.ElectricityUnitRate
+			if collectedRates.ElectricityIsAgile && collectedRates.ElectricityProductCode != "" && collectedRates.ElectricityTariffCode != "" {
+				agileRate, err := getCurrentAgileRate(collectedRates.ElectricityProductCode, collectedRates.ElectricityTariffCode, apiKey)
+				if err != nil {
+					fail("agile rate error: %v", err)
 				} else {
 					unitRate = agileRate
 				}
 			}
 			elecUnitRate.Set(unitRate)
-			elecStandingCharge.Set(rates.ElectricityStandingCharge)
-
+			elecStandingCharge.Set(collectedRates.ElectricityStandingCharge)
 			if gasMeter != nil {
-				gasUnitRate.Set(rates.GasUnitRate)
-				gasStandCharge.Set(rates.GasStandingCharge)
+				gasUnitRate.Set(collectedRates.GasUnitRate)
+				gasStandCharge.Set(collectedRates.GasStandingCharge)
 			}
 		}
 
-		// Account balance
-		balance, err := getAccountBalance(token)
-		if err != nil {
-			log.Printf("account balance error: %v", err)
+		if failedAny.Load() {
+			exporterUp.Set(0)
 		} else {
-			accountBalance.Set(balance)
+			exporterUp.Set(1)
 		}
 
 		time.Sleep(60 * time.Second)

cmd/octopus_exporter/meters.go

@@ -53,40 +53,71 @@ func getMeters(token string) ([]meterCandidate, error) {
 
 	var candidates []meterCandidate
 
-	accounts, _ := result["data"].(map[string]any)["viewer"].(map[string]any)["accounts"].([]any)
+	data, _ := result["data"].(map[string]any)
+	viewer, _ := data["viewer"].(map[string]any)
+	accounts, _ := viewer["accounts"].([]any)
 	for _, a := range accounts {
 		props, _ := a.(map[string]any)["properties"].([]any)
 		for _, p := range props {
-			pm := p.(map[string]any)
+			pm, ok := p.(map[string]any)
+			if !ok {
+				continue
+			}
 
 			for _, mp := range toSlice(pm["electricityMeterPoints"]) {
-				mpan, _ := mp.(map[string]any)["mpan"].(string)
-				for _, m := range toSlice(mp.(map[string]any)["meters"]) {
-					serial, _ := m.(map[string]any)["serialNumber"].(string)
-					for _, d := range toSlice(m.(map[string]any)["smartDevices"]) {
-						deviceID, _ := d.(map[string]any)["deviceId"].(string)
+				mpPoint, ok := mp.(map[string]any)
+				if !ok {
+					continue
+				}
+				mpan, _ := mpPoint["mpan"].(string)
+				for _, m := range toSlice(mpPoint["meters"]) {
+					meterMap, ok := m.(map[string]any)
+					if !ok {
+						continue
+					}
+					serial, _ := meterMap["serialNumber"].(string)
+					devices := toSlice(meterMap["smartDevices"])
+					for _, d := range devices {
+						dMap, ok := d.(map[string]any)
+						if !ok {
+							continue
+						}
+						deviceID, _ := dMap["deviceId"].(string)
 						if deviceID != "" {
 							candidates = append(candidates, meterCandidate{kind: electricity, mpan: mpan, serial: serial, deviceID: deviceID})
 						}
 					}
 					// Include meters without smart devices so we can still use the REST consumption endpoint.
-					if len(toSlice(m.(map[string]any)["smartDevices"])) == 0 && serial != "" {
+					if len(devices) == 0 && serial != "" {
 						candidates = append(candidates, meterCandidate{kind: electricity, mpan: mpan, serial: serial})
 					}
 				}
 			}
 
 			for _, mp := range toSlice(pm["gasMeterPoints"]) {
-				mprn, _ := mp.(map[string]any)["mprn"].(string)
-				for _, m := range toSlice(mp.(map[string]any)["meters"]) {
-					serial, _ := m.(map[string]any)["serialNumber"].(string)
-					for _, d := range toSlice(m.(map[string]any)["smartDevices"]) {
-						deviceID, _ := d.(map[string]any)["deviceId"].(string)
+				mpPoint, ok := mp.(map[string]any)
+				if !ok {
+					continue
+				}
+				mprn, _ := mpPoint["mprn"].(string)
+				for _, m := range toSlice(mpPoint["meters"]) {
+					meterMap, ok := m.(map[string]any)
+					if !ok {
+						continue
+					}
+					serial, _ := meterMap["serialNumber"].(string)
+					devices := toSlice(meterMap["smartDevices"])
+					for _, d := range devices {
+						dMap, ok := d.(map[string]any)
+						if !ok {
+							continue
+						}
+						deviceID, _ := dMap["deviceId"].(string)
 						if deviceID != "" {
 							candidates = append(candidates, meterCandidate{kind: gas, mprn: mprn, serial: serial, deviceID: deviceID})
 						}
 					}
-					if len(toSlice(m.(map[string]any)["smartDevices"])) == 0 && serial != "" {
+					if len(devices) == 0 && serial != "" {
 						candidates = append(candidates, meterCandidate{kind: gas, mprn: mprn, serial: serial})
 					}
 				}
@@ -99,7 +130,7 @@ func getMeters(token string) ([]meterCandidate, error) {
 
 // resolveMeter finds the meter matching the env var filters for the given kind.
 // Returns (nil, nil) if no meter of that kind exists on the account.
-func resolveMeter(token string, kind meterKind) (*resolvedMeter, error) {
+func resolveMeter(candidates []meterCandidate, kind meterKind) (*resolvedMeter, error) {
 	var wantDeviceID, wantID, wantSerial string
 	switch kind {
 	case electricity:
@@ -112,12 +143,6 @@ func resolveMeter(token string, kind meterKind) (*resolvedMeter, error) {
 		wantSerial = os.Getenv("OCTOPUS_GAS_SERIAL")
 	}
 
-	log.Printf("discovering %s meters from account...", kind)
-	candidates, err := getMeters(token)
-	if err != nil {
-		return nil, err
-	}
-
 	for _, c := range candidates {
 		if c.kind != kind {
 			continue

cmd/octopus_exporter/meters_test.go

@@ -0,0 +1,209 @@
+package main
+
+import (
+	"fmt"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+)
+
+func TestResolveMeter_NoFilters_ReturnsFirst(t *testing.T) {
+	t.Setenv("OCTOPUS_DEVICE_ID", "")
+	t.Setenv("OCTOPUS_MPAN", "")
+	t.Setenv("OCTOPUS_SERIAL", "")
+
+	candidates := []meterCandidate{
+		{kind: electricity, mpan: "1000000000001", serial: "A001", deviceID: "dev1"},
+		{kind: electricity, mpan: "1000000000002", serial: "A002", deviceID: "dev2"},
+	}
+	m, err := resolveMeter(candidates, electricity)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if m == nil {
+		t.Fatal("expected meter, got nil")
+	}
+	if m.mpan != "1000000000001" {
+		t.Errorf("got mpan %q, want 1000000000001", m.mpan)
+	}
+}
+
+func TestResolveMeter_FilterByMPAN(t *testing.T) {
+	t.Setenv("OCTOPUS_DEVICE_ID", "")
+	t.Setenv("OCTOPUS_MPAN", "1000000000002")
+	t.Setenv("OCTOPUS_SERIAL", "")
+
+	candidates := []meterCandidate{
+		{kind: electricity, mpan: "1000000000001", serial: "A001", deviceID: "dev1"},
+		{kind: electricity, mpan: "1000000000002", serial: "A002", deviceID: "dev2"},
+	}
+	m, err := resolveMeter(candidates, electricity)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if m == nil {
+		t.Fatal("expected meter, got nil")
+	}
+	if m.mpan != "1000000000002" {
+		t.Errorf("got mpan %q, want 1000000000002", m.mpan)
+	}
+}
+
+func TestResolveMeter_FilterByDeviceID(t *testing.T) {
+	t.Setenv("OCTOPUS_DEVICE_ID", "dev2")
+	t.Setenv("OCTOPUS_MPAN", "")
+	t.Setenv("OCTOPUS_SERIAL", "")
+
+	candidates := []meterCandidate{
+		{kind: electricity, mpan: "1000000000001", serial: "A001", deviceID: "dev1"},
+		{kind: electricity, mpan: "1000000000002", serial: "A002", deviceID: "dev2"},
+	}
+	m, err := resolveMeter(candidates, electricity)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if m.deviceID != "dev2" {
+		t.Errorf("got deviceID %q, want dev2", m.deviceID)
+	}
+}
+
+func TestResolveMeter_FilterMismatch_ReturnsError(t *testing.T) {
+	t.Setenv("OCTOPUS_MPAN", "9999999999999")
+	t.Setenv("OCTOPUS_DEVICE_ID", "")
+	t.Setenv("OCTOPUS_SERIAL", "")
+
+	candidates := []meterCandidate{
+		{kind: electricity, mpan: "1000000000001", serial: "A001"},
+	}
+	_, err := resolveMeter(candidates, electricity)
+	if err == nil {
+		t.Error("expected error for unmatched filter, got nil")
+	}
+}
+
+func TestResolveMeter_NoMetersOfKind_ReturnsNil(t *testing.T) {
+	t.Setenv("OCTOPUS_DEVICE_ID", "")
+	t.Setenv("OCTOPUS_MPAN", "")
+	t.Setenv("OCTOPUS_SERIAL", "")
+
+	candidates := []meterCandidate{
+		{kind: gas, mprn: "1111111111", serial: "G001"},
+	}
+	m, err := resolveMeter(candidates, electricity)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if m != nil {
+		t.Errorf("expected nil for no electricity meters, got %+v", m)
+	}
+}
+
+func TestResolveMeter_GasFilterByMPRN(t *testing.T) {
+	t.Setenv("OCTOPUS_GAS_DEVICE_ID", "")
+	t.Setenv("OCTOPUS_GAS_MPRN", "2222222222")
+	t.Setenv("OCTOPUS_GAS_SERIAL", "")
+
+	candidates := []meterCandidate{
+		{kind: gas, mprn: "1111111111", serial: "G001"},
+		{kind: gas, mprn: "2222222222", serial: "G002"},
+	}
+	m, err := resolveMeter(candidates, gas)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if m.mprn != "2222222222" {
+		t.Errorf("got mprn %q, want 2222222222", m.mprn)
+	}
+}
+
+func TestGetMeters_ElectricityWithSmartDevice(t *testing.T) {
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		fmt.Fprint(w, `{"data":{"viewer":{"accounts":[{"properties":[{
+			"electricityMeterPoints":[{
+				"mpan":"1000000000001",
+				"meters":[{"serialNumber":"A001","smartDevices":[{"deviceId":"dev1"}]}]
+			}],
+			"gasMeterPoints":[]
+		}]}]}}}`)
+	}))
+	defer srv.Close()
+	octopusGraphQL = srv.URL + "/"
+
+	candidates, err := getMeters("token")
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if len(candidates) != 1 {
+		t.Fatalf("got %d candidates, want 1", len(candidates))
+	}
+	c := candidates[0]
+	if c.kind != electricity {
+		t.Errorf("kind: got %q, want electricity", c.kind)
+	}
+	if c.mpan != "1000000000001" {
+		t.Errorf("mpan: got %q, want 1000000000001", c.mpan)
+	}
+	if c.deviceID != "dev1" {
+		t.Errorf("deviceID: got %q, want dev1", c.deviceID)
+	}
+}
+
+func TestGetMeters_MeterWithoutSmartDevice(t *testing.T) {
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		fmt.Fprint(w, `{"data":{"viewer":{"accounts":[{"properties":[{
+			"electricityMeterPoints":[{
+				"mpan":"1000000000001",
+				"meters":[{"serialNumber":"A001","smartDevices":[]}]
+			}],
+			"gasMeterPoints":[]
+		}]}]}}}`)
+	}))
+	defer srv.Close()
+	octopusGraphQL = srv.URL + "/"
+
+	candidates, err := getMeters("token")
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if len(candidates) != 1 {
+		t.Fatalf("got %d candidates, want 1", len(candidates))
+	}
+	if candidates[0].deviceID != "" {
+		t.Errorf("expected empty deviceID for meter without smart device, got %q", candidates[0].deviceID)
+	}
+}
+
+func TestGetMeters_BothElectricityAndGas(t *testing.T) {
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		fmt.Fprint(w, `{"data":{"viewer":{"accounts":[{"properties":[{
+			"electricityMeterPoints":[{
+				"mpan":"1000000000001",
+				"meters":[{"serialNumber":"A001","smartDevices":[{"deviceId":"dev1"}]}]
+			}],
+			"gasMeterPoints":[{
+				"mprn":"1111111111",
+				"meters":[{"serialNumber":"G001","smartDevices":[{"deviceId":"gdev1"}]}]
+			}]
+		}]}]}}}`)
+	}))
+	defer srv.Close()
+	octopusGraphQL = srv.URL + "/"
+
+	candidates, err := getMeters("token")
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if len(candidates) != 2 {
+		t.Fatalf("got %d candidates, want 2", len(candidates))
+	}
+	kinds := map[meterKind]bool{}
+	for _, c := range candidates {
+		kinds[c.kind] = true
+	}
+	if !kinds[electricity] {
+		t.Error("expected electricity candidate")
+	}
+	if !kinds[gas] {
+		t.Error("expected gas candidate")
+	}
+}

cmd/octopus_exporter/rates.go

@@ -31,11 +31,16 @@ func getRates(token string) (*tariffRates, error) {
 
 	rates := &tariffRates{}
 
-	accounts, _ := result["data"].(map[string]any)["viewer"].(map[string]any)["accounts"].([]any)
+	data, _ := result["data"].(map[string]any)
+	viewer, _ := data["viewer"].(map[string]any)
+	accounts, _ := viewer["accounts"].([]any)
 	for _, a := range accounts {
 		props, _ := a.(map[string]any)["properties"].([]any)
 		for _, p := range props {
-			pm := p.(map[string]any)
+			pm, ok := p.(map[string]any)
+			if !ok {
+				continue
+			}
 
 			for _, mp := range toSlice(pm["electricityMeterPoints"]) {
 				if tariff := activeAgreementTariff(mp); tariff != nil {
@@ -44,7 +49,6 @@ func getRates(token string) (*tariffRates, error) {
 					rates.ElectricityProductCode, _ = tariff["productCode"].(string)
 					rates.ElectricityTariffCode, _ = tariff["tariffCode"].(string)
 					// HalfHourlyTariff has no unitRate field — detect Agile by absence.
-					_, rates.ElectricityIsAgile = tariff["unitRates"]
 					if _, hasUnit := tariff["unitRate"]; !hasUnit {
 						rates.ElectricityIsAgile = true
 					}
@@ -65,8 +69,15 @@ func getRates(token string) (*tariffRates, error) {
 
 // activeAgreementTariff returns the tariff map for the agreement with validTo == null.
 func activeAgreementTariff(meterPoint any) map[string]any {
-	for _, ag := range toSlice(meterPoint.(map[string]any)["agreements"]) {
-		agm := ag.(map[string]any)
+	mp, ok := meterPoint.(map[string]any)
+	if !ok {
+		return nil
+	}
+	for _, ag := range toSlice(mp["agreements"]) {
+		agm, ok := ag.(map[string]any)
+		if !ok {
+			continue
+		}
 		if agm["validTo"] == nil {
 			tariff, _ := agm["tariff"].(map[string]any)
 			return tariff

cmd/octopus_exporter/rest.go

@@ -2,29 +2,25 @@ package main
 
 import (
 	"encoding/json"
-	"io"
 	"net/http"
 	"net/url"
 )
 
 var octopusREST = "https://api.octopus.energy"
 
-func doREST(path string, params url.Values) (map[string]any, error) {
+func doREST(path string, params url.Values, key string) (map[string]any, error) {
 	u := octopusREST + path
 	if len(params) > 0 {
 		u += "?" + params.Encode()
 	}
+	raw, err := executeWithRetry(func() (*http.Request, error) {
 		req, err := http.NewRequest(http.MethodGet, u, nil)
 		if err != nil {
 			return nil, err
 		}
-	req.SetBasicAuth(apiKey, "")
-	resp, err := http.DefaultClient.Do(req)
-	if err != nil {
-		return nil, err
-	}
-	defer resp.Body.Close()
-	raw, err := io.ReadAll(resp.Body)
+		req.SetBasicAuth(key, "")
+		return req, nil
+	})
 	if err != nil {
 		return nil, err
 	}

cmd/octopus_exporter/telemetry.go

@@ -3,6 +3,7 @@ package main
 import (
 	"encoding/json"
 	"errors"
+	"fmt"
 )
 
 type telemetryReading struct {
@@ -18,13 +19,14 @@ func getLiveConsumption(token, deviceID string) (*telemetryReading, error) {
 		Query:         "query getSmartMeterTelemetry($meterDeviceId: String!, $start: DateTime, $end: DateTime, $grouping: TelemetryGrouping) {\n  smartMeterTelemetry(deviceId: $meterDeviceId, start: $start, end: $end, grouping: $grouping) {\n    readAt\n    consumption\n    demand\n    __typename\n  }\n}\n",
 	}, token)
 	if err != nil {
-		if err.Error() == "GraphQL error: Signature of the JWT has expired." {
-			return nil, errTokenExpired
-		}
 		return nil, err
 	}
 
-	telemetry := toSlice(result["data"].(map[string]any)["smartMeterTelemetry"])
+	data, ok := result["data"].(map[string]any)
+	if !ok {
+		return nil, fmt.Errorf("unexpected API response: missing data field")
+	}
+	telemetry := toSlice(data["smartMeterTelemetry"])
 	if len(telemetry) == 0 {
 		return nil, errors.New("no telemetry data returned")
 	}

cmd/octopus_exporter/telemetry_test.go

@@ -0,0 +1,71 @@
+package main
+
+import (
+	"fmt"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+)
+
+func TestGetLiveConsumption_Success(t *testing.T) {
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		fmt.Fprint(w, `{"data":{"smartMeterTelemetry":[
+			{"readAt":"2026-04-24T10:00:00Z","consumption":"1.23","demand":"456.7"}
+		]}}`)
+	}))
+	defer srv.Close()
+	octopusGraphQL = srv.URL + "/"
+
+	reading, err := getLiveConsumption("test-token", "device-123")
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if float64(reading.Demand) != 456.7 {
+		t.Errorf("demand: got %v, want 456.7", reading.Demand)
+	}
+	if float64(reading.Consumption) != 1.23 {
+		t.Errorf("consumption: got %v, want 1.23", reading.Consumption)
+	}
+	if reading.ReadAt != "2026-04-24T10:00:00Z" {
+		t.Errorf("readAt: got %q, want %q", reading.ReadAt, "2026-04-24T10:00:00Z")
+	}
+}
+
+func TestGetLiveConsumption_EmptyTelemetry(t *testing.T) {
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		fmt.Fprint(w, `{"data":{"smartMeterTelemetry":[]}}`)
+	}))
+	defer srv.Close()
+	octopusGraphQL = srv.URL + "/"
+
+	_, err := getLiveConsumption("test-token", "device-123")
+	if err == nil {
+		t.Error("expected error for empty telemetry, got nil")
+	}
+}
+
+func TestGetLiveConsumption_MissingDataField(t *testing.T) {
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		fmt.Fprint(w, `{"data":null}`)
+	}))
+	defer srv.Close()
+	octopusGraphQL = srv.URL + "/"
+
+	_, err := getLiveConsumption("test-token", "device-123")
+	if err == nil {
+		t.Error("expected error for null data, got nil")
+	}
+}
+
+func TestGetLiveConsumption_GraphQLError(t *testing.T) {
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		fmt.Fprint(w, `{"errors":[{"message":"device not found"}]}`)
+	}))
+	defer srv.Close()
+	octopusGraphQL = srv.URL + "/"
+
+	_, err := getLiveConsumption("test-token", "device-123")
+	if err == nil {
+		t.Error("expected error for GraphQL error, got nil")
+	}
+}