ananke/internal/service/daemon_poststart_autorepair_test.go

package service

import (
	"bytes"
	"context"
	"errors"
	"log"
	"strings"

	"scm.bstein.dev/bstein/ananke/internal/cluster"
	"scm.bstein.dev/bstein/ananke/internal/execx"
	"testing"
	"time"

	"scm.bstein.dev/bstein/ananke/internal/config"
)

// TestDaemonMaybeRunPostStartAutoHeal runs one orchestration or CLI step.
// Signature: TestDaemonMaybeRunPostStartAutoHeal(t *testing.T).
// Why: covers the daemon-side interval and on-battery guards for the new
// post-start repair loop.
func TestDaemonMaybeRunPostStartAutoHeal(t *testing.T) {
	calls := 0
	d := &Daemon{
		cfg: config.Config{
			Startup: config.Startup{
				PostStartAutoHealSeconds: 10,
			},
		},
		postStartAutoHealOverride: func(context.Context) error {
			calls++
			return nil
		},
	}

	d.cfg.Coordination.Role = "peer"
	var last time.Time
	d.maybeRunPostStartAutoHeal(context.Background(), &last, false)
	if calls != 0 || !last.IsZero() {
		t.Fatal("peer must not mutate shared cluster recovery state")
	}
	d.cfg.Coordination.Role = "coordinator"
	d.maybeRunPostStartAutoHeal(context.Background(), &last, false)
	if calls != 1 {
		t.Fatalf("expected first auto-heal invocation, got %d", calls)
	}

	d.maybeRunPostStartAutoHeal(context.Background(), &last, false)
	if calls != 1 {
		t.Fatalf("expected interval guard to suppress second call, got %d", calls)
	}

	last = time.Now().Add(-11 * time.Second)
	d.maybeRunPostStartAutoHeal(context.Background(), &last, true)
	if calls != 1 {
		t.Fatalf("expected on-battery guard to suppress call, got %d", calls)
	}

	last = time.Now().Add(-11 * time.Second)
	d.maybeRunPostStartAutoHeal(context.Background(), &last, false)
	if calls != 2 {
		t.Fatalf("expected second allowed auto-heal call, got %d", calls)
	}
}

// TestDaemonRecoveryFailureDoesNotStopMonitoring exercises optional repair isolation.
// Signature: TestDaemonRecoveryFailureDoesNotStopMonitoring(t *testing.T).
// Why: a missing or failed repair helper must not terminate UPS monitoring.
func TestDaemonRecoveryFailureDoesNotStopMonitoring(t *testing.T) {
	var messages bytes.Buffer
	d := &Daemon{cfg: config.Config{Startup: config.Startup{PostStartAutoHealSeconds: 60}}, log: log.New(&messages, "", 0)}
	var last time.Time
	d.maybeRunPostStartAutoHeal(context.Background(), &last, false)
	if !last.IsZero() {
		t.Fatal("missing helper must not consume an interval")
	}
	if err := d.runPostStartAutoHeal(context.Background()); err != nil {
		t.Fatal(err)
	}
	d.orch = cluster.New(config.Config{}, &execx.Runner{DryRun: true}, nil, d.log)
	if err := d.runPostStartAutoHeal(context.Background()); err != nil {
		t.Fatal(err)
	}
	d.postStartAutoHealOverride = func(context.Context) error { return errors.New("synthetic repair failure") }
	d.maybeRunPostStartAutoHeal(context.Background(), &last, false)
	if last.IsZero() || !strings.Contains(messages.String(), "synthetic repair failure") {
		t.Fatal("failed repair must be logged and rate limited")
	}
}