HH-441: harden durable storage and recovery (#92)

* HH-441: harden durable storage and recovery

* HH-441: clear recovery review blockers

* HH-441: enforce offsite backup failure domain

---------

Co-authored-by: Rogee <rogee@ipao.vip>
This commit is contained in:
Rogee
2026-08-22 02:16:02 +08:00
committed by GitHub
co-authored by rogee
parent 77c91662d2
commit 6d6d80dd86
20 changed files with 966 additions and 202 deletions
+5 -9
View File
@@ -178,6 +178,9 @@ jobs:
MEILI_MASTER_KEY: ci-meili-secret-16
GOCHAT_JWT_SECRET: ci-smoke-jwt-secret-at-least-32-characters
GOCHAT_JWT_PREVIOUS_SECRETS: ci-previous-jwt-secret-at-least-32-characters
GOCHAT_BACKUP_OFFSITE_DIR: /mnt/gochat-offsite
GOCHAT_BACKUP_OFFSITE_SOURCE: backup.example.test:/gochat
GOCHAT_BACKUP_OFFSITE_FSTYPE: nfs4
steps:
- uses: actions/checkout@v4
- name: Build production images from repository root
@@ -200,16 +203,9 @@ jobs:
test -s /app/frontend/dist/index.html'
- name: Start production Compose and smoke core pages
run: |
if MEILI_MASTER_KEY=too-short deploy/docker/preflight.sh; then
echo "preflight accepted a short Meilisearch key" >&2
exit 1
fi
if POSTGRES_IMAGE_REF=pgvector/pgvector:pg16 deploy/docker/preflight.sh; then
echo "preflight accepted a mutable PostgreSQL image" >&2
exit 1
fi
deploy/docker/preflight.sh
deploy/docker/preflight_test.sh
docker compose -f deploy/docker/docker-compose.prod.yml config --format json | python3 -c 'import json, os, sys; config = json.load(sys.stdin); assert all(config["services"][service]["environment"]["GOCHAT_JWT_PREVIOUS_SECRETS"] == os.environ["GOCHAT_JWT_PREVIOUS_SECRETS"] for service in ("gochat", "worker"))'
docker compose -f deploy/docker/docker-compose.prod.yml --profile ops run --rm migrate
docker compose -f deploy/docker/docker-compose.prod.yml up -d --wait gochat
curl -fsS "http://127.0.0.1:$GOCHAT_PORT/health" | grep -q '"status":"ok"'
curl -fsS "http://127.0.0.1:$GOCHAT_PORT/app" | grep -q '/assets/'
+15
View File
@@ -43,6 +43,10 @@ func main() {
fmt.Println("Migrations applied successfully (up)")
case "down":
if err := rejectDestructiveProductionMigration(env, -1); err != nil {
fmt.Fprintf(os.Stderr, "Error: %v\n", err)
os.Exit(1)
}
if err := database.RollbackMigrations(dbURL, migrationsPath); err != nil {
fmt.Fprintf(os.Stderr, "Error running migrations down: %v\n", err)
os.Exit(1)
@@ -89,6 +93,10 @@ func main() {
fmt.Fprintf(os.Stderr, "Error: invalid step count '%s': %v\n", os.Args[2], err)
os.Exit(1)
}
if err := rejectDestructiveProductionMigration(env, steps); err != nil {
fmt.Fprintf(os.Stderr, "Error: %v\n", err)
os.Exit(1)
}
if err := database.MigrateSteps(dbURL, migrationsPath, steps); err != nil {
fmt.Fprintf(os.Stderr, "Error running %d migration steps: %v\n", steps, err)
os.Exit(1)
@@ -102,6 +110,13 @@ func main() {
}
}
func rejectDestructiveProductionMigration(env string, steps int) error {
if env == "production" && steps < 0 {
return fmt.Errorf("destructive schema down is disabled in production; roll back the application image or restore a backup")
}
return nil
}
func printUsage() {
fmt.Println("Usage: migrate <command> [args]")
fmt.Println("")
+5
View File
@@ -218,3 +218,8 @@ func TestCurrentVersion(t *testing.T) {
assert.Equal(t, uint(3), version)
assert.False(t, dirty)
}
func TestRejectDestructiveProductionMigration(t *testing.T) {
require.Error(t, rejectDestructiveProductionMigration("production", -1))
require.NoError(t, rejectDestructiveProductionMigration("production", 1))
require.NoError(t, rejectDestructiveProductionMigration("development", -1))
}
+1 -1
View File
@@ -7,7 +7,7 @@ server:
database:
dsn: "postgres://gochat:CHANGE_ME@postgres:5432/gochat_production?sslmode=disable"
run_migrations: true
run_migrations: false
migrations_path: "/app/migrations"
redis:
+14
View File
@@ -73,6 +73,9 @@ func Bootstrap(env string) (*App, error) {
if err != nil {
return nil, fmt.Errorf("config load failed: %w", err)
}
if err := validateStartupMigrations(env, cfg); err != nil {
return nil, err
}
// Step 2: Validate configuration
if err := config.Validate(cfg); err != nil {
@@ -788,6 +791,10 @@ func Bootstrap(env string) (*App, error) {
WithWidgetAuth(inboxRepo, contactInboxRepo).
WithConversationRepo(conversationRepo).
WithAccessDB(db)
service.RegisterUploadCleanupJobs(workerPool, uploadService)
if _, err := service.EnqueueUploadCleanup(context.Background(), workerPool, time.Now()); err != nil {
applogger.L().Warnf("failed to enqueue upload cleanup: %v", err)
}
uploadHandler := v1.NewUploadHandler(uploadService).WithAccessAuth(jwtService)
// Step 9: Wire handlers (HTTP presentation layer)
@@ -1000,6 +1007,13 @@ func Bootstrap(env string) (*App, error) {
}, nil
}
func validateStartupMigrations(env string, cfg *config.Config) error {
if env == "production" && cfg.Database.RunMigrations {
return fmt.Errorf("database.run_migrations must be false in production; run the one-shot migration job before web/worker startup")
}
return nil
}
// hubTypingAdapter implements service.TypingIndicator by broadcasting
// typing_on/typing_off events directly through the WS hub's SendToAccount method.
// This avoids the Redis dependency required by the full ws.TypingTracker.
@@ -0,0 +1,14 @@
package app
import (
"testing"
"github.com/gochat/gochat/internal/config"
"github.com/stretchr/testify/require"
)
func TestValidateStartupMigrations(t *testing.T) {
cfg := &config.Config{Database: config.DatabaseConfig{RunMigrations: true}}
require.ErrorContains(t, validateStartupMigrations("production", cfg), "one-shot migration job")
require.NoError(t, validateStartupMigrations("development", cfg))
}
@@ -0,0 +1,134 @@
package database
import (
"os"
"path/filepath"
"testing"
"time"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"gorm.io/driver/postgres"
"gorm.io/gorm"
"gorm.io/gorm/logger"
)
func TestHighRiskMigrationsPreserveRollbackData(t *testing.T) {
if os.Getenv("GOCHAT_TEST_DB") == "sqlite" {
t.Skip("requires PostgreSQL migration semantics")
}
dsn := os.Getenv("GOCHAT_TEST_DB_URL")
if dsn == "" {
dsn = "host=localhost port=5432 user=postgres password=postgres dbname=gochat_test sslmode=disable"
}
db, err := gorm.Open(postgres.Open(dsn), &gorm.Config{Logger: logger.Default.LogMode(logger.Silent)})
require.NoError(t, err)
sqlDB, err := db.DB()
require.NoError(t, err)
sqlDB.SetMaxOpenConns(1)
t.Cleanup(func() { _ = sqlDB.Close() })
schema := "high_risk_migrations_" + time.Now().Format("20060102150405000000000")
require.NoError(t, db.Exec("CREATE SCHEMA "+schema).Error)
t.Cleanup(func() { _ = db.Exec("DROP SCHEMA " + schema + " CASCADE").Error })
require.NoError(t, db.Exec("SET search_path TO "+schema).Error)
require.NoError(t, db.Exec(`
CREATE TABLE reporting_events_rollups (
id SERIAL PRIMARY KEY, account_id INTEGER NOT NULL, dimension VARCHAR(50) NOT NULL,
dimension_value VARCHAR(255) NOT NULL, metric_name VARCHAR(50) NOT NULL,
value DOUBLE PRECISION NOT NULL, value_in_business_hours DOUBLE PRECISION,
period VARCHAR(50) NOT NULL, created_at TIMESTAMPTZ DEFAULT NOW(),
updated_at TIMESTAMPTZ DEFAULT NOW(), deleted_at TIMESTAMPTZ,
UNIQUE(account_id, dimension, dimension_value, metric_name, period)
);
INSERT INTO reporting_events_rollups
(account_id, dimension, dimension_value, metric_name, value, value_in_business_hours, period)
VALUES (1, 'inbox', '42', 'conversations_count', 7, 3, '2026-08-21');
CREATE TABLE custom_attribute_definitions (attribute_model TEXT, attribute_name TEXT);
CREATE TABLE conversations (id BIGINT PRIMARY KEY, custom_attributes JSONB, updated_at TIMESTAMPTZ);
INSERT INTO custom_attribute_definitions VALUES ('conversation_attribute', 'swt_source_url');
INSERT INTO conversations VALUES (1, '{"swt_source_url":"https://example.test"}', NOW());
`).Error)
up48, err := os.ReadFile(filepath.Join("..", "..", "migrations", "000048_fix_reporting_events_rollups_schema.up.sql"))
require.NoError(t, err)
require.NoError(t, db.Exec(string(up48)).Error)
var rollup struct {
DimensionID int64
Count int64
SumValue float64
}
require.NoError(t, db.Table("reporting_events_rollups").Select("dimension_id, count, sum_value").Scan(&rollup).Error)
assert.Equal(t, int64(42), rollup.DimensionID)
assert.Equal(t, int64(7), rollup.Count)
assert.Equal(t, float64(7), rollup.SumValue)
up76, err := os.ReadFile(filepath.Join("..", "..", "migrations", "000076_replace_shangwutong_source_attributes_with_messages.up.sql"))
require.NoError(t, err)
require.NoError(t, db.Exec(string(up76)).Error)
var definitions, conversations int64
require.NoError(t, db.Table("custom_attribute_definitions").Count(&definitions).Error)
require.NoError(t, db.Table("conversations").Where("custom_attributes->>'swt_source_url' IS NOT NULL").Count(&conversations).Error)
assert.Equal(t, int64(1), definitions)
assert.Equal(t, int64(1), conversations)
down48, err := os.ReadFile(filepath.Join("..", "..", "migrations", "000048_fix_reporting_events_rollups_schema.down.sql"))
require.NoError(t, err)
require.ErrorContains(t, db.Exec(string(down48)).Error, "irreversible")
}
func TestMigration48RejectsUnsafeLegacyData(t *testing.T) {
if os.Getenv("GOCHAT_TEST_DB") == "sqlite" {
t.Skip("requires PostgreSQL migration semantics")
}
dsn := os.Getenv("GOCHAT_TEST_DB_URL")
if dsn == "" {
dsn = "host=localhost port=5432 user=postgres password=postgres dbname=gochat_test sslmode=disable"
}
db, err := gorm.Open(postgres.Open(dsn), &gorm.Config{Logger: logger.Default.LogMode(logger.Silent)})
require.NoError(t, err)
sqlDB, err := db.DB()
require.NoError(t, err)
sqlDB.SetMaxOpenConns(1)
t.Cleanup(func() { _ = sqlDB.Close() })
up48, err := os.ReadFile(filepath.Join("..", "..", "migrations", "000048_fix_reporting_events_rollups_schema.up.sql"))
require.NoError(t, err)
tests := []struct {
name, values, want string
rows int64
}{
{"invalid_period", "(1, 'inbox', '42', 'conversations_count', 1, 'not-a-date')", "cannot convert to date", 1},
{"invalid_dimension", "(1, 'inbox', 'vip', 'conversations_count', 1, '2026-08-21')", "cannot convert to bigint", 1},
{"converted_key_collision", "(1, 'inbox', '042', 'conversations_count', 1, '2026-08-21'), (1, 'inbox', '42', 'conversations_count', 2, '2026-08-21')", "unique-key collision", 2},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
schema := "migration_48_rejection_" + tt.name
t.Cleanup(func() { _ = db.Exec("DROP SCHEMA IF EXISTS " + schema + " CASCADE").Error })
require.NoError(t, db.Exec("DROP SCHEMA IF EXISTS "+schema+" CASCADE").Error)
require.NoError(t, db.Exec("CREATE SCHEMA "+schema).Error)
require.NoError(t, db.Exec("SET search_path TO "+schema).Error)
require.NoError(t, db.Exec(`
CREATE TABLE reporting_events_rollups (
id SERIAL PRIMARY KEY, account_id INTEGER NOT NULL, dimension VARCHAR(50) NOT NULL,
dimension_value VARCHAR(255) NOT NULL, metric_name VARCHAR(50) NOT NULL,
value DOUBLE PRECISION NOT NULL, value_in_business_hours DOUBLE PRECISION,
period VARCHAR(50) NOT NULL, created_at TIMESTAMPTZ DEFAULT NOW(),
updated_at TIMESTAMPTZ DEFAULT NOW(), deleted_at TIMESTAMPTZ,
UNIQUE(account_id, dimension, dimension_value, metric_name, period)
);
INSERT INTO reporting_events_rollups
(account_id, dimension, dimension_value, metric_name, value, period)
VALUES `+tt.values).Error)
require.ErrorContains(t, db.Exec(string(up48)).Error, tt.want)
require.NoError(t, db.Exec("ROLLBACK").Error)
var rows int64
require.NoError(t, db.Table(schema+".reporting_events_rollups").Count(&rows).Error)
assert.Equal(t, tt.rows, rows)
require.NoError(t, db.Exec("SET search_path TO public").Error)
require.NoError(t, db.Exec("DROP SCHEMA "+schema+" CASCADE").Error)
})
}
}
@@ -64,6 +64,11 @@ func (r *DirectUploadRepo) UpdateStatus(ctx context.Context, id uint, status mod
return r.db.WithContext(ctx).Model(&model.DirectUpload{}).Where("id = ?", id).Update("status", status).Error
}
// Delete permanently removes a staged upload after its object is gone.
func (r *DirectUploadRepo) Delete(ctx context.Context, id uint) error {
return r.db.WithContext(ctx).Unscoped().Delete(&model.DirectUpload{}, id).Error
}
// FindExpired retrieves all direct uploads that have passed their expiry timestamp
// and are still in pending status.
func (r *DirectUploadRepo) FindExpired(ctx context.Context, before time.Time) ([]model.DirectUpload, error) {
@@ -83,4 +88,4 @@ func (r *DirectUploadRepo) BatchDeleteExpired(ctx context.Context, before time.T
Where("status = ? AND expires_at < ?", model.DirectUploadStatusPending, before).
Delete(&model.DirectUpload{})
return result.RowsAffected, result.Error
}
}
+136 -5
View File
@@ -26,9 +26,15 @@ import (
"github.com/gochat/gochat/internal/model"
"github.com/gochat/gochat/internal/repository"
"github.com/gochat/gochat/internal/security"
"github.com/gochat/gochat/internal/worker"
applogger "github.com/gochat/gochat/pkg/logger"
)
const (
TaskTypeCleanupExpiredUploads = "storage:cleanup_expired_uploads"
uploadCleanupInterval = 24 * time.Hour
)
// UploadService handles file uploads for both account-level and widget direct uploads.
type UploadService struct {
directUploadRepo *repository.DirectUploadRepo
@@ -846,14 +852,139 @@ func randomStorageKey() string {
return hex.EncodeToString(buf)
}
// CleanupExpiredUploads removes expired direct upload records and their files.
// CleanupExpiredUploads removes each object before its record. Failed object
// deletions leave the row intact so the durable job can retry safely.
func (s *UploadService) CleanupExpiredUploads(ctx context.Context) (int64, error) {
count, err := s.directUploadRepo.BatchDeleteExpired(ctx, time.Now())
now := time.Now()
uploads, err := s.directUploadRepo.FindExpired(ctx, now)
if err != nil {
return 0, fmt.Errorf("failed to cleanup expired uploads: %w", err)
return 0, fmt.Errorf("find expired uploads: %w", err)
}
applogger.L().Infof("Cleaned up %d expired direct uploads", count)
return count, nil
var count int64
var cleanupErr error
for i := range uploads {
if err := s.removeUploadFile(uploads[i].FileURL); err != nil {
cleanupErr = errors.Join(cleanupErr, fmt.Errorf("remove upload %s: %w", uploads[i].UploadUUID, err))
continue
}
if err := s.directUploadRepo.Delete(ctx, uploads[i].ID); err != nil {
cleanupErr = errors.Join(cleanupErr, fmt.Errorf("delete upload %s: %w", uploads[i].UploadUUID, err))
continue
}
count++
}
orphans, err := s.cleanupOrphanedUploadFiles(ctx, now.Add(-uploadCleanupInterval))
count += orphans
cleanupErr = errors.Join(cleanupErr, err)
applogger.L().Infof("Cleaned up %d expired or orphaned uploads", count)
return count, cleanupErr
}
// cleanupOrphanedUploadFiles reconciles only roots owned by UploadService.
// The grace period keeps an in-flight file-to-row write from racing cleanup.
func (s *UploadService) cleanupOrphanedUploadFiles(ctx context.Context, cutoff time.Time) (int64, error) {
if s.accessDB == nil {
return 0, nil
}
references := map[string]struct{}{}
var urls []string
for _, column := range []string{"file_url", "thumb_url"} {
urls = nil
if err := s.accessDB.WithContext(ctx).Model(&model.DirectUpload{}).
Where(column+" LIKE ?", "/uploads/%").Pluck(column, &urls).Error; err != nil {
return 0, fmt.Errorf("load direct upload references: %w", err)
}
for _, url := range urls {
references[url] = struct{}{}
}
}
if s.accessDB.Migrator().HasTable(&model.User{}) {
urls = nil
if err := s.accessDB.WithContext(ctx).Model(&model.User{}).
Where("avatar_url LIKE ?", "/uploads/account/%").Pluck("avatar_url", &urls).Error; err != nil {
return 0, fmt.Errorf("load avatar references: %w", err)
}
for _, url := range urls {
references[url] = struct{}{}
}
}
localPath := "./uploads"
if s.cfg != nil && s.cfg.Storage.LocalPath != "" {
localPath = s.cfg.Storage.LocalPath
}
var removed int64
var reconcileErr error
for _, subdir := range []string{"account", "widget_direct"} {
root := filepath.Join(localPath, subdir)
err := filepath.Walk(root, func(path string, info os.FileInfo, walkErr error) error {
if walkErr != nil {
if errors.Is(walkErr, os.ErrNotExist) {
return nil
}
return walkErr
}
if !info.Mode().IsRegular() || info.ModTime().After(cutoff) {
return nil
}
relative, err := filepath.Rel(localPath, path)
if err != nil {
return err
}
fileURL := "/uploads/" + filepath.ToSlash(relative)
if _, ok := references[fileURL]; ok {
return nil
}
if err := os.Remove(path); err != nil {
if errors.Is(err, os.ErrNotExist) {
return nil
}
return err
}
removed++
return nil
})
if err != nil && !errors.Is(err, os.ErrNotExist) {
reconcileErr = errors.Join(reconcileErr, fmt.Errorf("reconcile %s uploads: %w", subdir, err))
}
}
return removed, reconcileErr
}
func (s *UploadService) removeUploadFile(fileURL string) error {
relative := strings.TrimPrefix(fileURL, "/uploads/")
if relative == fileURL || relative == "" || filepath.IsAbs(relative) || strings.HasPrefix(filepath.Clean(relative), "..") {
return fmt.Errorf("invalid upload path %q", fileURL)
}
localPath := s.cfg.Storage.LocalPath
if localPath == "" {
localPath = "./uploads"
}
err := os.Remove(filepath.Join(localPath, relative))
if errors.Is(err, os.ErrNotExist) {
return nil
}
return err
}
// RegisterUploadCleanupJobs runs orphan cleanup daily through the existing
// durable, idempotent worker queue.
func RegisterUploadCleanupJobs(wp *worker.WorkerPool, svc *UploadService) {
wp.Register(TaskTypeCleanupExpiredUploads, func(ctx context.Context, _ *model.BackgroundJob) error {
_, cleanupErr := svc.CleanupExpiredUploads(ctx)
_, enqueueErr := EnqueueUploadCleanup(ctx, wp, time.Now().Add(uploadCleanupInterval))
return errors.Join(cleanupErr, enqueueErr)
})
}
func EnqueueUploadCleanup(ctx context.Context, wp *worker.WorkerPool, at time.Time) (*model.BackgroundJob, error) {
bucket := at.UTC().Truncate(uploadCleanupInterval).Unix()
return wp.Enqueue(ctx, TaskTypeCleanupExpiredUploads, nil,
worker.WithQueue("low"),
worker.WithScheduledAt(at),
worker.WithMaxAttempts(10),
worker.WithIdempotencyKey(fmt.Sprintf("storage:cleanup_expired_uploads:%d", bucket)),
)
}
// --- MIME detection helpers (reuse patterns from widget_theme_service.go) ---
@@ -5,12 +5,14 @@ import (
"encoding/json"
"mime/multipart"
"net/http"
"net/http/httptest"
"os"
"path/filepath"
"strings"
"testing"
"time"
"github.com/gin-gonic/gin"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"gorm.io/driver/sqlite"
@@ -145,7 +147,8 @@ func setupUploadServiceTest(t *testing.T) (*gorm.DB, *UploadService, string) {
directUploadRepo := repository.NewDirectUploadRepo(db)
uploadService := NewUploadService(directUploadRepo, cfg).
WithWidgetAuth(repository.NewInboxRepo(db), repository.NewContactInboxRepo(db))
WithWidgetAuth(repository.NewInboxRepo(db), repository.NewContactInboxRepo(db)).
WithAccessDB(db)
return db, uploadService, tmpDir
}
@@ -195,6 +198,25 @@ func TestUploadService_AccountUpload_Success(t *testing.T) {
assert.GreaterOrEqual(t, len(files), 1, "file should be saved on disk")
}
func TestUploadSharedStorageSurvivesReplicaRecreation(t *testing.T) {
_, svc, storagePath := setupUploadServiceTest(t)
content := []byte("shared attachment")
upload, err := svc.AccountUpload(context.Background(), 1, AccountUploadRequest{
FileHeader: createTestFileHeader(t, "shared.txt", content),
})
require.NoError(t, err)
for range 2 {
replica := gin.New()
replica.StaticFS("/uploads", gin.Dir(storagePath, false))
response := httptest.NewRecorder()
request := httptest.NewRequest(http.MethodGet, upload.FileURL, nil)
replica.ServeHTTP(response, request)
assert.Equal(t, http.StatusOK, response.Code)
assert.Equal(t, content, response.Body.Bytes())
}
}
func TestUploadService_AccountUpload_NoAccountID(t *testing.T) {
_, svc, _ := setupUploadServiceTest(t)
@@ -459,7 +481,7 @@ func TestUploadService_PrivateAttachmentAccessIsTenantAndSessionScoped(t *testin
}
func TestUploadService_CleanupExpiredUploads(t *testing.T) {
db, svc, _ := setupUploadServiceTest(t)
db, svc, tmpDir := setupUploadServiceTest(t)
// Insert a pending-but-expired upload record (pending status with past expires_at)
// BatchDeleteExpired deletes pending uploads where expires_at < now
@@ -476,6 +498,9 @@ func TestUploadService_CleanupExpiredUploads(t *testing.T) {
ExpiresAt: time.Now().Add(-1 * time.Hour), // already past expiry
}
require.NoError(t, db.Create(expiredUpload).Error)
expiredPath := filepath.Join(tmpDir, "account", "1", "old_file.png")
require.NoError(t, os.MkdirAll(filepath.Dir(expiredPath), 0o755))
require.NoError(t, os.WriteFile(expiredPath, []byte("expired"), 0o600))
// Insert a non-expired upload (pending status with future expires_at)
activeUpload := &model.DirectUpload{
@@ -491,13 +516,55 @@ func TestUploadService_CleanupExpiredUploads(t *testing.T) {
ExpiresAt: time.Now().Add(24 * time.Hour),
}
require.NoError(t, db.Create(activeUpload).Error)
failedUpload := &model.DirectUpload{
UploadUUID: "failed-uuid-789", AccountID: 1, Status: model.DirectUploadStatusPending,
Source: model.DirectUploadSourceAccount, OriginalName: "invalid.png", FileType: "image",
MimeType: "image/png", FileSize: 100, FileURL: "https://example.test/invalid.png",
ExpiresAt: time.Now().Add(-time.Hour),
}
require.NoError(t, db.Create(failedUpload).Error)
count, err := svc.CleanupExpiredUploads(context.Background())
require.NoError(t, err)
require.ErrorContains(t, err, "invalid upload path")
assert.Equal(t, int64(1), count, "should clean up 1 expired upload")
assert.NoFileExists(t, expiredPath)
// Verify active upload still exists
var remaining model.DirectUpload
require.NoError(t, db.Where("upload_uuid = ?", "active-uuid-456").First(&remaining).Error)
assert.Equal(t, model.DirectUploadStatusPending, remaining.Status)
remaining = model.DirectUpload{}
require.NoError(t, db.Where("upload_uuid = ?", "failed-uuid-789").First(&remaining).Error)
}
func TestUploadService_ReconcilesCrashWindowOrphans(t *testing.T) {
db, svc, storagePath := setupUploadServiceTest(t)
dir := filepath.Join(storagePath, "account", "1")
require.NoError(t, os.MkdirAll(dir, 0o755))
orphan := filepath.Join(dir, "crash-window.png")
referenced := filepath.Join(dir, "referenced.png")
fresh := filepath.Join(dir, "in-flight.png")
for _, path := range []string{orphan, referenced, fresh} {
require.NoError(t, os.WriteFile(path, testPNG, 0o600))
}
old := time.Now().Add(-uploadCleanupInterval - time.Hour)
require.NoError(t, os.Chtimes(orphan, old, old))
require.NoError(t, os.Chtimes(referenced, old, old))
require.NoError(t, db.Create(&model.DirectUpload{
UploadUUID: "referenced", AccountID: 1, Status: model.DirectUploadStatusPending,
Source: model.DirectUploadSourceAccount, OriginalName: "referenced.png", FileType: "image",
MimeType: "image/png", FileSize: int64(len(testPNG)), FileURL: "/uploads/account/1/referenced.png",
ExpiresAt: time.Now().Add(time.Hour),
}).Error)
count, err := svc.CleanupExpiredUploads(context.Background())
require.NoError(t, err)
assert.Equal(t, int64(1), count)
assert.NoFileExists(t, orphan)
assert.FileExists(t, referenced)
assert.FileExists(t, fresh)
count, err = svc.CleanupExpiredUploads(context.Background())
require.NoError(t, err)
assert.Zero(t, count)
}
@@ -1,22 +1,5 @@
-- Revert reporting_events_rollups to the original (incorrect) init schema structure.
-- WARNING: this restores the broken column names (dimension, dimension_value, metric_name,
-- value, value_in_business_hours, period) that do not match the GORM model.
DROP TABLE IF EXISTS reporting_events_rollups;
CREATE TABLE IF NOT EXISTS reporting_events_rollups (
id SERIAL PRIMARY KEY,
account_id INTEGER NOT NULL,
dimension VARCHAR(50) NOT NULL,
dimension_value VARCHAR(255) NOT NULL,
metric_name VARCHAR(50) NOT NULL,
value DOUBLE PRECISION NOT NULL,
value_in_business_hours DOUBLE PRECISION DEFAULT 0,
period VARCHAR(50) NOT NULL,
created_at TIMESTAMP WITH TIME ZONE DEFAULT NOW(),
updated_at TIMESTAMP WITH TIME ZONE DEFAULT NOW(),
deleted_at TIMESTAMP WITH TIME ZONE,
UNIQUE(account_id, dimension, dimension_value, metric_name, period)
);
CREATE INDEX idx_reporting_events_rollups_deleted_at ON reporting_events_rollups(deleted_at);
DO $$
BEGIN
RAISE EXCEPTION 'migration 000048 is irreversible; restore the pre-upgrade backup instead of destroying rollup data';
END
$$;
@@ -1,18 +1,67 @@
-- Fix reporting_events_rollups table structure to match GORM model
-- The init schema (000001) created columns: dimension, dimension_value, metric_name,
-- value, value_in_business_hours, period — but the GORM model and all Go code use:
-- date, dimension_type, dimension_id, metric, count, sum_value, sum_value_business_hours.
-- This migration drops and recreates the table with the correct schema.
-- The table is empty in practice (rollups are computed on-demand), so data loss is negligible.
-- Convert the legacy rollup schema without dropping its data. The explicit
-- transaction makes lock/statement timeout failures leave the old table intact.
BEGIN;
LOCK TABLE reporting_events_rollups IN ACCESS EXCLUSIVE MODE;
DROP TABLE IF EXISTS reporting_events_rollups;
DO $$
DECLARE
invalid_periods BIGINT;
invalid_dimensions BIGINT;
invalid_counts BIGINT;
oversized_dimensions BIGINT;
collision_groups BIGINT;
BEGIN
SELECT count(*) INTO invalid_periods
FROM reporting_events_rollups
WHERE NOT pg_input_is_valid(period, 'date');
IF invalid_periods > 0 THEN
RAISE EXCEPTION 'migration 000048: % rollup period value(s) cannot convert to date', invalid_periods
USING HINT = 'Repair or remove the reported legacy rows explicitly, then retry the migration.';
END IF;
CREATE TABLE reporting_events_rollups (
SELECT count(*) INTO invalid_dimensions
FROM reporting_events_rollups
WHERE NOT pg_input_is_valid(dimension_value, 'bigint');
IF invalid_dimensions > 0 THEN
RAISE EXCEPTION 'migration 000048: % rollup dimension_value(s) cannot convert to bigint', invalid_dimensions
USING HINT = 'Map non-numeric legacy dimensions explicitly, then retry the migration.';
END IF;
SELECT count(*) INTO invalid_counts
FROM reporting_events_rollups
WHERE metric_name LIKE '%count%' AND NOT pg_input_is_valid(value::text, 'bigint');
IF invalid_counts > 0 THEN
RAISE EXCEPTION 'migration 000048: % count value(s) cannot convert losslessly to bigint', invalid_counts
USING HINT = 'Repair fractional, non-finite, or out-of-range count values, then retry the migration.';
END IF;
SELECT count(*) INTO oversized_dimensions
FROM reporting_events_rollups
WHERE length(dimension) > 20;
IF oversized_dimensions > 0 THEN
RAISE EXCEPTION 'migration 000048: % dimension value(s) exceed the new 20-character limit', oversized_dimensions
USING HINT = 'Shorten or explicitly map oversized dimensions, then retry the migration.';
END IF;
SELECT count(*) INTO collision_groups
FROM (
SELECT 1
FROM reporting_events_rollups
GROUP BY account_id, period::date, dimension, dimension_value::BIGINT, metric_name
HAVING count(*) > 1
) collisions;
IF collision_groups > 0 THEN
RAISE EXCEPTION 'migration 000048: % unique-key collision group(s) appear after type conversion', collision_groups
USING HINT = 'Merge or choose one legacy row in each collision group explicitly, then retry the migration.';
END IF;
END $$;
CREATE TABLE reporting_events_rollups_v2 (
id SERIAL PRIMARY KEY,
account_id INTEGER NOT NULL,
date DATE NOT NULL,
dimension_type VARCHAR(20) NOT NULL,
dimension_id INTEGER NOT NULL,
dimension_id BIGINT NOT NULL,
metric VARCHAR(50) NOT NULL,
count BIGINT NOT NULL DEFAULT 0,
sum_value DOUBLE PRECISION NOT NULL DEFAULT 0,
@@ -23,7 +72,36 @@ CREATE TABLE reporting_events_rollups (
UNIQUE(account_id, date, dimension_type, dimension_id, metric)
);
INSERT INTO reporting_events_rollups_v2 (
id, account_id, date, dimension_type, dimension_id, metric, count,
sum_value, sum_value_business_hours, created_at, updated_at, deleted_at
)
SELECT
id,
account_id,
period::date,
dimension,
dimension_value::BIGINT,
metric_name,
CASE WHEN metric_name LIKE '%count%' THEN value::BIGINT ELSE 1 END,
value,
COALESCE(value_in_business_hours, 0),
created_at,
updated_at,
deleted_at
FROM reporting_events_rollups;
SELECT setval(
pg_get_serial_sequence('reporting_events_rollups_v2', 'id'),
COALESCE((SELECT MAX(id) FROM reporting_events_rollups_v2), 1),
EXISTS (SELECT 1 FROM reporting_events_rollups_v2)
);
DROP TABLE reporting_events_rollups;
ALTER TABLE reporting_events_rollups_v2 RENAME TO reporting_events_rollups;
ALTER INDEX reporting_events_rollups_v2_pkey RENAME TO reporting_events_rollups_pkey;
CREATE INDEX idx_reporting_events_rollups_deleted_at ON reporting_events_rollups(deleted_at);
CREATE INDEX idx_reporting_events_rollups_account_id ON reporting_events_rollups(account_id) WHERE deleted_at IS NULL;
CREATE INDEX idx_reporting_events_rollups_date ON reporting_events_rollups(date) WHERE deleted_at IS NULL;
CREATE INDEX idx_reporting_events_rollups_dimension ON reporting_events_rollups(dimension_type, dimension_id) WHERE deleted_at IS NULL;
COMMIT;
@@ -1,52 +1,4 @@
DELETE FROM custom_attribute_definitions
WHERE attribute_model = 'conversation_attribute'
AND attribute_name IN (
'swt_source_url',
'swt_source_search_term',
'swt_source_purchase_term',
'swt_source_keyword_id',
'swt_source_channel',
'swt_source_realtime_location',
'swt_source_region',
'swt_source_ad_account_id',
'swt_source_wakeable',
'swt_baidu_conversation_type',
'swt_baidu_agent_name',
'swt_baidu_ssid'
);
UPDATE conversations
SET
custom_attributes = COALESCE(custom_attributes, '{}'::jsonb)
- 'swt_source_url'
- 'swt_source_description'
- 'swt_source_search_term'
- 'swt_source_purchase_term'
- 'swt_source_keyword_id'
- 'swt_source_channel'
- 'swt_source_realtime_location'
- 'swt_source_region'
- 'swt_source_ad_account_id'
- 'swt_source_wakeable'
- 'swt_baidu_conversation_type'
- 'swt_baidu_agent_name'
- 'swt_baidu_ssid',
updated_at = CURRENT_TIMESTAMP
WHERE COALESCE(custom_attributes, '{}'::jsonb) ?| ARRAY[
'swt_source_url',
'swt_source_description',
'swt_source_search_term',
'swt_source_purchase_term',
'swt_source_keyword_id',
'swt_source_channel',
'swt_source_realtime_location',
'swt_source_region',
'swt_source_ad_account_id',
'swt_source_wakeable',
'swt_baidu_conversation_type',
'swt_baidu_agent_name',
'swt_baidu_ssid'
];
-- Historical kind=8 system messages are backfilled from the Connector SQLite
-- store because PostgreSQL intentionally does not retain the raw encoded payload.
-- Keep the legacy source attributes during the expand/contract window. New
-- connector versions emit source information as system messages, while an old
-- application digest still needs these definitions and values after rollback.
SELECT 1;
+57 -49
View File
@@ -1,54 +1,62 @@
#!/bin/bash
# GoChat Database Backup Script
# Reference: Chatwoot uses pg_dump for backup in production deployments
# Usage: ./scripts/db_backup.sh [env]
#!/usr/bin/env bash
# Create one encrypted, off-site copy containing PostgreSQL, attachments, and
# the Connector's online SQLite backup.
set -euo pipefail
umask 077
ENV="${1:-production}"
TIMESTAMP=$(date +%Y%m%d_%H%M%S)
BACKUP_DIR="${GOCHAT_BACKUP_DIR:-/var/backups/gochat}"
DSN="${GOCHAT_DATABASE_DSN:-}"
RETENTION_DAYS="${GOCHAT_BACKUP_RETENTION_DAYS:-30}"
dsn=${GOCHAT_DATABASE_DSN:?GOCHAT_DATABASE_DSN is required}
storage=${GOCHAT_STORAGE_PATH:?GOCHAT_STORAGE_PATH is required}
connector=${GOCHAT_CONNECTOR_BACKUP_FILE:?GOCHAT_CONNECTOR_BACKUP_FILE is required}
backup_dir=${GOCHAT_BACKUP_DIR:-/var/backups/gochat}
offsite_dir=${GOCHAT_BACKUP_OFFSITE_DIR:?GOCHAT_BACKUP_OFFSITE_DIR is required}
passphrase_file=${GOCHAT_BACKUP_PASSPHRASE_FILE:?GOCHAT_BACKUP_PASSPHRASE_FILE is required}
retention_days=${GOCHAT_BACKUP_RETENTION_DAYS:-30}
version=${GOCHAT_VERSION:-unknown}
timestamp=$(date -u +%Y%m%dT%H%M%SZ)
if [[ -z "$DSN" ]]; then
echo "[$(date)] ERROR: GOCHAT_DATABASE_DSN is not set"
exit 1
fi
# Parse DSN: postgres://user:password@host:port/dbname?sslmode=...
DB_USER=$(echo "$DSN" | sed 's|.*://||; s|:.*||')
DB_PASS=$(echo "$DSN" | sed 's|.*://[^:]*:||; s|@.*||')
DB_HOST=$(echo "$DSN" | sed 's|.*@||; s|:.*||')
DB_PORT=$(echo "$DSN" | sed 's|.*@.*:||; s|/.*||')
DB_NAME=$(echo "$DSN" | sed 's|.*/||; s|\?.*||')
mkdir -p "${BACKUP_DIR}"
BACKUP_FILE="${BACKUP_DIR}/${DB_NAME}_${TIMESTAMP}.sql.gz"
echo "[$(date)] Starting backup of ${DB_NAME} on ${DB_HOST}:${DB_PORT}"
# pg_dump with compression — mirrors Chatwoot backup approach
PGPASSWORD="${DB_PASS}" pg_dump \
-h "${DB_HOST}" \
-p "${DB_PORT}" \
-U "${DB_USER}" \
-d "${DB_NAME}" \
--format=custom \
--compress=9 \
| gzip > "${BACKUP_FILE}"
BACKUP_SIZE=$(du -h "${BACKUP_FILE}" | cut -f1)
echo "[$(date)] Backup complete: ${BACKUP_FILE} (${BACKUP_SIZE})"
# Prune old backups beyond retention period
find "${BACKUP_DIR}" -name "*.sql.gz" -mtime +"${RETENTION_DAYS}" -delete
echo "[$(date)] Pruned backups older than ${RETENTION_DAYS} days"
# Verify backup integrity
gunzip -t "${BACKUP_FILE}" && echo "[$(date)] Backup integrity verified" || {
echo "[$(date)] ERROR: Backup integrity check failed!"
rm -f "${BACKUP_FILE}"
exit 1
[[ -d "$storage" ]] || { echo "storage path does not exist: $storage" >&2; exit 1; }
[[ -z "$(find "$storage" -type l -print -quit)" ]] || { echo "storage must not contain symbolic links" >&2; exit 1; }
[[ -f "$connector" ]] || { echo "connector backup does not exist: $connector" >&2; exit 1; }
[[ -r "$passphrase_file" ]] || { echo "backup passphrase file is not readable" >&2; exit 1; }
server_major=$(($(psql "$dsn" -Atqc "SHOW server_version_num") / 10000))
client_major=$(pg_dump --version | awk '{print $NF}' | cut -d. -f1)
[[ "$client_major" == "$server_major" ]] || {
echo "pg_dump major $client_major must match PostgreSQL major $server_major" >&2
exit 1
}
install -d -m 0700 "$backup_dir" "$offsite_dir"
[[ "$(realpath "$backup_dir")" != "$(realpath "$offsite_dir")" ]] || {
echo "off-site directory must use a different path/failure domain" >&2
exit 1
}
work_dir=$(mktemp -d "$backup_dir/.backup.XXXXXX")
trap 'rm -rf "$work_dir"' EXIT
stage=$work_dir/snapshot
mkdir "$stage"
pg_dump --dbname "$dsn" --format=custom --compress=9 --file "$stage/postgres.dump"
pg_restore --list "$stage/postgres.dump" >/dev/null
tar -C "$storage" -cf "$stage/attachments.tar" .
cp "$connector" "$stage/connector.db"
created_at_epoch=$(date -u +%s)
{
echo "created_at=$timestamp"
echo "created_at_epoch=$created_at_epoch"
echo "version=$version"
echo "rpo_target_seconds=86400"
} >"$stage/manifest"
(cd "$stage" && sha256sum postgres.dump attachments.tar connector.db manifest >SHA256SUMS)
bundle=$backup_dir/gochat-$timestamp.tar.enc
tar -C "$stage" -cf - . | openssl enc -aes-256-cbc -pbkdf2 -salt -pass "file:$passphrase_file" -out "$bundle"
(cd "$backup_dir" && sha256sum "$(basename "$bundle")" >"$(basename "$bundle").sha256")
openssl enc -d -aes-256-cbc -pbkdf2 -pass "file:$passphrase_file" -in "$bundle" | tar -tf - >/dev/null
cp "$bundle" "$bundle.sha256" "$offsite_dir/"
find "$backup_dir" "$offsite_dir" -maxdepth 1 -type f -name 'gochat-*.tar.enc*' -mtime "+$retention_days" -delete
echo "backup=$bundle offsite=$offsite_dir/$(basename "$bundle") version=$version created_at=$timestamp"
+70
View File
@@ -0,0 +1,70 @@
#!/usr/bin/env bash
# Restore an encrypted GoChat backup into an empty database/storage target and
# print auditable RPO/RTO and business-integrity evidence.
set -euo pipefail
umask 077
bundle=${1:?usage: db_restore.sh /path/to/gochat-*.tar.enc}
dsn=${GOCHAT_DATABASE_DSN:?GOCHAT_DATABASE_DSN is required}
storage=${GOCHAT_STORAGE_PATH:?GOCHAT_STORAGE_PATH is required}
connector=${GOCHAT_CONNECTOR_DB_PATH:?GOCHAT_CONNECTOR_DB_PATH is required}
passphrase_file=${GOCHAT_BACKUP_PASSPHRASE_FILE:?GOCHAT_BACKUP_PASSPHRASE_FILE is required}
started_at=$(date -u +%s)
[[ -f "$bundle" ]] || { echo "backup bundle does not exist: $bundle" >&2; exit 1; }
[[ -f "$bundle.sha256" ]] || { echo "backup checksum does not exist: $bundle.sha256" >&2; exit 1; }
(cd "$(dirname "$bundle")" && sha256sum -c "$(basename "$bundle").sha256")
server_major=$(($(psql "$dsn" -Atqc "SHOW server_version_num") / 10000))
client_major=$(pg_restore --version | awk '{print $NF}' | cut -d. -f1)
[[ "$client_major" == "$server_major" ]] || {
echo "pg_restore major $client_major must match PostgreSQL major $server_major" >&2
exit 1
}
table_count=$(psql "$dsn" -Atqc "SELECT count(*) FROM pg_tables WHERE schemaname = 'public'")
[[ "$table_count" == "0" ]] || { echo "target database is not empty" >&2; exit 1; }
install -d -m 0700 "$storage" "$(dirname "$connector")"
[[ -z "$(find "$storage" -mindepth 1 -maxdepth 1 -print -quit)" ]] || { echo "target storage is not empty" >&2; exit 1; }
[[ ! -e "$connector" ]] || { echo "target connector database already exists" >&2; exit 1; }
work_dir=$(mktemp -d "$(dirname "$bundle")/.restore.XXXXXX")
trap 'rm -rf "$work_dir"' EXIT
archive=$work_dir/snapshot.tar
openssl enc -d -aes-256-cbc -pbkdf2 -pass "file:$passphrase_file" -in "$bundle" -out "$archive"
while IFS= read -r path; do
case "$path" in
/*|../*|*/../*|*/..) echo "unsafe archive path: $path" >&2; exit 1 ;;
esac
done < <(tar -tf "$archive")
if tar -tvf "$archive" | awk '{type = substr($1, 1, 1); if (type != "-" && type != "d") found = 1} END {exit(found ? 0 : 1)}'; then
echo "backup archive contains unsupported entry types" >&2
exit 1
fi
tar -C "$work_dir" -xf "$archive"
rm "$archive"
(cd "$work_dir" && sha256sum -c SHA256SUMS)
pg_restore --list "$work_dir/postgres.dump" >/dev/null
pg_restore --exit-on-error --no-owner --no-privileges --dbname "$dsn" "$work_dir/postgres.dump"
while IFS= read -r path; do
case "$path" in
/*|../*|*/../*|*/..) echo "unsafe attachment path: $path" >&2; exit 1 ;;
esac
done < <(tar -tf "$work_dir/attachments.tar")
if tar -tvf "$work_dir/attachments.tar" | awk '{type = substr($1, 1, 1); if (type != "-" && type != "d") found = 1} END {exit(found ? 0 : 1)}'; then
echo "attachment archive contains unsupported entry types" >&2
exit 1
fi
tar -C "$storage" -xf "$work_dir/attachments.tar"
install -m 0600 "$work_dir/connector.db" "$connector"
migration_version=$(psql "$dsn" -Atqc "SELECT COALESCE(MAX(version), 0) FROM schema_migrations WHERE NOT dirty")
accounts=$(psql "$dsn" -Atqc "SELECT count(*) FROM accounts WHERE deleted_at IS NULL")
attachments=$(psql "$dsn" -Atqc "SELECT count(*) FROM attachments WHERE deleted_at IS NULL")
backup_epoch=$(awk -F= '$1 == "created_at_epoch" {print $2}' "$work_dir/manifest")
version=$(awk -F= '$1 == "version" {print substr($0, index($0, "=") + 1)}' "$work_dir/manifest")
finished_at=$(date -u +%s)
echo "restore=ok version=$version migration=$migration_version accounts=$accounts attachments=$attachments rpo_seconds=$((started_at-backup_epoch)) rto_seconds=$((finished_at-started_at))"
+8 -2
View File
@@ -50,6 +50,9 @@ RUN CGO_ENABLED=0 GOOS=linux GOARCH=amd64 \
-X github.com/gochat/gochat/internal/config.BuildDate=${BUILD_DATE}" \
-o /gochat-worker ./cmd/gochat/
RUN CGO_ENABLED=0 GOOS=linux GOARCH=amd64 \
go build -ldflags="-s -w" -o /migrate ./cmd/migrate/
# ========== Production Stage ==========
FROM alpine:3.21
@@ -60,17 +63,20 @@ LABEL org.opencontainers.image.version="${VERSION}" \
org.opencontainers.image.revision="${COMMIT_SHA}" \
org.opencontainers.image.created="${BUILD_DATE}"
# Install runtime dependencies
RUN apk --no-cache add ca-certificates tzdata curl && addgroup -S gochat && adduser -S gochat -G gochat
# Install runtime and recovery dependencies.
RUN apk --no-cache add bash ca-certificates tzdata curl openssl postgresql16-client && \
addgroup -S gochat && adduser -S gochat -G gochat
WORKDIR /app
# Copy binary and configs from builder
COPY --from=builder /gochat /app/gochat
COPY --from=builder /gochat-worker /app/gochat-worker
COPY --from=builder /migrate /app/migrate
COPY --chown=gochat:gochat --from=frontend-builder /app/frontend/dist /app/frontend/dist
COPY backend/configs/ /app/configs/
COPY backend/migrations/ /app/migrations/
COPY backend/scripts/db_backup.sh backend/scripts/db_restore.sh /app/scripts/
ENV GOCHAT_FRONTEND_DIST=/app/frontend/dist
+70 -1
View File
@@ -8,7 +8,7 @@ x-gochat-environment: &gochat-environment
GOCHAT_SERVER_MODE: release
GOCHAT_SERVER_CORS_ALLOWED_ORIGINS: ${GOCHAT_SERVER_CORS_ALLOWED_ORIGINS:?set production CORS origins}
GOCHAT_DATABASE_DSN: ${GOCHAT_DATABASE_DSN:-postgres://${POSTGRES_USER:-gochat}:${POSTGRES_PASSWORD:?set POSTGRES_PASSWORD}@postgres:5432/${POSTGRES_DB:-gochat_production}?sslmode=disable}
GOCHAT_DATABASE_RUN_MIGRATIONS: "true"
GOCHAT_DATABASE_RUN_MIGRATIONS: "false"
GOCHAT_DATABASE_MIGRATIONS_PATH: /app/migrations
GOCHAT_REDIS_DSN: redis://:${REDIS_PASSWORD:?set REDIS_PASSWORD}@redis:6379
GOCHAT_SEARCH_ENGINE: meilisearch
@@ -126,6 +126,75 @@ services:
memory: 512M
cpus: "1.0"
migrate:
image: *gochat-image
profiles: ["ops"]
depends_on:
postgres:
condition: service_healthy
entrypoint: ["/app/migrate"]
command: ["up"]
restart: "no"
environment:
<<: *gochat-environment
GOCHAT_DATABASE_DSN: postgres://${POSTGRES_USER:-gochat}:${POSTGRES_PASSWORD:?set POSTGRES_PASSWORD}@postgres:5432/${POSTGRES_DB:-gochat_production}?sslmode=disable&lock_timeout=5000&statement_timeout=900000
GOCHAT_DATABASE_RUN_MIGRATIONS: "false"
backup:
image: *gochat-image
user: "0:0"
profiles: ["ops"]
depends_on:
postgres:
condition: service_healthy
entrypoint: ["/app/scripts/db_backup.sh"]
restart: "no"
environment:
GOCHAT_DATABASE_DSN: postgres://${POSTGRES_USER:-gochat}:${POSTGRES_PASSWORD:?set POSTGRES_PASSWORD}@postgres:5432/${POSTGRES_DB:-gochat_production}?sslmode=disable
GOCHAT_STORAGE_PATH: /source/storage/uploads
GOCHAT_CONNECTOR_BACKUP_FILE: /source/connector/${GOCHAT_CONNECTOR_BACKUP_NAME:-latest.db}
GOCHAT_BACKUP_DIR: /backup/local
GOCHAT_BACKUP_OFFSITE_DIR: /backup/offsite
GOCHAT_BACKUP_PASSPHRASE_FILE: /run/secrets/backup-passphrase
GOCHAT_BACKUP_RETENTION_DAYS: ${GOCHAT_BACKUP_RETENTION_DAYS:-30}
GOCHAT_VERSION: ${GOCHAT_IMAGE_REF}
volumes:
- gochat_storage:/source/storage:ro
- shangwutong_backups:/source/connector:ro
- ${GOCHAT_BACKUP_DIR:-./backups/local}:/backup/local
- type: bind
source: ${GOCHAT_BACKUP_OFFSITE_DIR:?set an existing external off-site mount point}
target: /backup/offsite
bind:
create_host_path: false
- ${GOCHAT_BACKUP_PASSPHRASE_FILE:-./.secrets/backup-passphrase}:/run/secrets/backup-passphrase:ro
restore:
image: *gochat-image
user: "0:0"
profiles: ["ops"]
depends_on:
postgres:
condition: service_healthy
entrypoint: ["/app/scripts/db_restore.sh"]
command: ["/backup/offsite/${GOCHAT_RESTORE_BUNDLE:-missing.tar.enc}"]
restart: "no"
environment:
GOCHAT_DATABASE_DSN: postgres://${POSTGRES_USER:-gochat}:${POSTGRES_PASSWORD:?set POSTGRES_PASSWORD}@postgres:5432/${POSTGRES_DB:-gochat_production}?sslmode=disable
GOCHAT_STORAGE_PATH: /restore/storage/uploads
GOCHAT_CONNECTOR_DB_PATH: /restore/connector/connector.db
GOCHAT_BACKUP_PASSPHRASE_FILE: /run/secrets/backup-passphrase
volumes:
- gochat_storage:/restore/storage
- shangwutong_data:/restore/connector
- type: bind
source: ${GOCHAT_BACKUP_OFFSITE_DIR:?set an existing external off-site mount point}
target: /backup/offsite
read_only: true
bind:
create_host_path: false
- ${GOCHAT_BACKUP_PASSPHRASE_FILE:-./.secrets/backup-passphrase}:/run/secrets/backup-passphrase:ro
shangwutong:
image: ${SHANGWUTONG_IMAGE_REF:?set SHANGWUTONG_IMAGE_REF to an immutable image digest}
restart: always
+39 -1
View File
@@ -12,7 +12,7 @@ if (($#)); then
compose_args=(--env-file "$env_file" "${compose_args[@]}")
fi
required=(GOCHAT_IMAGE_REF SHANGWUTONG_IMAGE_REF GOCHAT_SERVER_CORS_ALLOWED_ORIGINS POSTGRES_PASSWORD REDIS_PASSWORD MEILI_MASTER_KEY GOCHAT_JWT_SECRET)
required=(GOCHAT_IMAGE_REF SHANGWUTONG_IMAGE_REF GOCHAT_SERVER_CORS_ALLOWED_ORIGINS POSTGRES_PASSWORD REDIS_PASSWORD MEILI_MASTER_KEY GOCHAT_JWT_SECRET GOCHAT_BACKUP_OFFSITE_DIR GOCHAT_BACKUP_OFFSITE_SOURCE GOCHAT_BACKUP_OFFSITE_FSTYPE)
for name in "${required[@]}"; do
value=${!name:-}
if [[ -z $value || ${value^^} == *CHANGE_ME* ]]; then
@@ -21,6 +21,44 @@ for name in "${required[@]}"; do
fi
done
if [[ $GOCHAT_BACKUP_OFFSITE_DIR != /* ]] || [[ ! -d $GOCHAT_BACKUP_OFFSITE_DIR ]]; then
echo "GOCHAT_BACKUP_OFFSITE_DIR must be an existing external mount point" >&2
exit 1
fi
offsite_dir=$(realpath -e -- "$GOCHAT_BACKUP_OFFSITE_DIR")
if [[ $offsite_dir == / ]]; then
echo "GOCHAT_BACKUP_OFFSITE_DIR must not be /" >&2
exit 1
fi
if ! offsite_info=$(findmnt -M "$offsite_dir" -n -o SOURCE,FSTYPE,MAJ:MIN); then
echo "GOCHAT_BACKUP_OFFSITE_DIR must be an existing external mount point" >&2
exit 1
fi
read -r offsite_source offsite_type offsite_device <<< "$offsite_info"
if [[ $offsite_source != "$GOCHAT_BACKUP_OFFSITE_SOURCE" || $offsite_type != "$GOCHAT_BACKUP_OFFSITE_FSTYPE" ]]; then
echo "GOCHAT_BACKUP_OFFSITE_DIR mount source/type does not match the approved values" >&2
exit 1
fi
case $offsite_type in
tmpfs | devtmpfs | ramfs)
echo "GOCHAT_BACKUP_OFFSITE_DIR must not use an in-memory filesystem" >&2
exit 1
;;
esac
local_probe=${GOCHAT_BACKUP_DIR:-./backups/local}
if [[ $local_probe != /* ]]; then
local_probe=$script_dir/$local_probe
fi
while [[ ! -e $local_probe && $local_probe != / ]]; do
local_probe=$(dirname -- "$local_probe")
done
read -r local_source local_device < <(findmnt -T "$local_probe" -n -o SOURCE,MAJ:MIN)
if [[ $offsite_source == "$local_source" || $offsite_device == "$local_device" ]]; then
echo "GOCHAT_BACKUP_OFFSITE_DIR must use a different source/device than GOCHAT_BACKUP_DIR" >&2
exit 1
fi
if ((${#GOCHAT_JWT_SECRET} < 32)); then
echo "GOCHAT_JWT_SECRET must be at least 32 characters" >&2
exit 1
+96
View File
@@ -0,0 +1,96 @@
#!/usr/bin/env bash
set -euo pipefail
root=$(cd "$(dirname "${BASH_SOURCE[0]}")/../.." && pwd)
tmp=$(mktemp -d)
trap 'rm -rf "$tmp"' EXIT
mkdir -p "$tmp/bin" "$tmp/local" "$tmp/offsite"
cat > "$tmp/bin/findmnt" <<'EOF'
#!/usr/bin/env bash
if [[ $1 == -M && $2 == "$GOCHAT_BACKUP_OFFSITE_DIR" ]]; then
printf '%s %s %s\n' "$TEST_OFFSITE_SOURCE" "$TEST_OFFSITE_FSTYPE" "$TEST_OFFSITE_DEVICE"
elif [[ $1 == -T ]]; then
printf '%s %s\n' "$TEST_LOCAL_SOURCE" "$TEST_LOCAL_DEVICE"
else
exit 1
fi
EOF
cat > "$tmp/bin/docker" <<'EOF'
#!/usr/bin/env bash
if [[ -n ${TEST_MUTABLE_IMAGE:-} ]]; then
echo 'pgvector/pgvector:pg16'
else
printf '%s\n' \
'gochat@example.invalid/gochat@sha256:0000000000000000000000000000000000000000000000000000000000000000' \
'gochat@example.invalid/connector@sha256:1111111111111111111111111111111111111111111111111111111111111111'
fi
EOF
chmod +x "$tmp/bin/findmnt" "$tmp/bin/docker"
export PATH="$tmp/bin:$PATH"
export GOCHAT_IMAGE_REF='gochat@example.invalid/gochat@sha256:0000000000000000000000000000000000000000000000000000000000000000'
export SHANGWUTONG_IMAGE_REF='gochat@example.invalid/connector@sha256:1111111111111111111111111111111111111111111111111111111111111111'
export GOCHAT_SERVER_CORS_ALLOWED_ORIGINS=https://chat.example.test
export POSTGRES_PASSWORD=ci-postgres-secret
export REDIS_PASSWORD=ci-redis-secret
export MEILI_MASTER_KEY=ci-meili-secret-16
export GOCHAT_JWT_SECRET=ci-smoke-jwt-secret-at-least-32-characters
export GOCHAT_BACKUP_DIR="$tmp/local"
export GOCHAT_BACKUP_OFFSITE_DIR="$tmp/offsite"
export GOCHAT_BACKUP_OFFSITE_SOURCE='backup.example.test:/gochat'
export GOCHAT_BACKUP_OFFSITE_FSTYPE=nfs4
export TEST_OFFSITE_SOURCE=$GOCHAT_BACKUP_OFFSITE_SOURCE
export TEST_OFFSITE_FSTYPE=$GOCHAT_BACKUP_OFFSITE_FSTYPE
export TEST_OFFSITE_DEVICE=0:42
export TEST_LOCAL_SOURCE=/dev/sda1
export TEST_LOCAL_DEVICE=8:1
run_preflight() {
"$root/deploy/docker/preflight.sh" > "$tmp/output" 2>&1
}
expect_failure() {
local name=$1 expected=$2
if run_preflight; then
echo "preflight accepted $name" >&2
exit 1
fi
grep -F "$expected" "$tmp/output" >/dev/null
}
GOCHAT_BACKUP_OFFSITE_DIR=/
expect_failure 'the root filesystem' 'must not be /'
GOCHAT_BACKUP_OFFSITE_DIR=/tmp
TEST_OFFSITE_SOURCE=tmpfs
TEST_OFFSITE_FSTYPE=tmpfs
GOCHAT_BACKUP_OFFSITE_SOURCE=tmpfs
GOCHAT_BACKUP_OFFSITE_FSTYPE=tmpfs
expect_failure 'tmpfs' 'must not use an in-memory filesystem'
GOCHAT_BACKUP_OFFSITE_DIR=$tmp/offsite
TEST_OFFSITE_SOURCE='backup.example.test:/gochat'
TEST_OFFSITE_FSTYPE=nfs4
GOCHAT_BACKUP_OFFSITE_SOURCE=$TEST_OFFSITE_SOURCE
GOCHAT_BACKUP_OFFSITE_FSTYPE=$TEST_OFFSITE_FSTYPE
TEST_OFFSITE_DEVICE=$TEST_LOCAL_DEVICE
expect_failure 'the local backup device' 'must use a different source/device'
TEST_OFFSITE_DEVICE=0:42
GOCHAT_BACKUP_OFFSITE_SOURCE='other.example.test:/gochat'
expect_failure 'an unapproved mount source' 'does not match the approved values'
GOCHAT_BACKUP_OFFSITE_SOURCE=$TEST_OFFSITE_SOURCE
MEILI_MASTER_KEY=too-short
expect_failure 'a short Meilisearch key' 'must be at least 16 bytes'
MEILI_MASTER_KEY=ci-meili-secret-16
TEST_MUTABLE_IMAGE=1
export TEST_MUTABLE_IMAGE
expect_failure 'a mutable production image' 'must be pinned to a sha256 digest'
unset TEST_MUTABLE_IMAGE
run_preflight
grep -F 'production preflight passed' "$tmp/output" >/dev/null
echo 'preflight tests passed'
+130 -47
View File
@@ -1,65 +1,148 @@
# GoChat Rolling Upgrade Strategy
# Reference: Chatwoot deployment uses zero-downtime upgrade pattern
# Production backup, restore, and upgrade runbook
## Overview
Production web and worker processes never migrate on startup. The Compose file
also requires `GOCHAT_IMAGE_REF` to be an immutable `image@sha256:digest`; tags such
as `latest` are not an acceptable rollback record.
GoChat follows a blue-green deployment strategy for production upgrades,
ensuring zero downtime during version transitions.
The bundled `gochat_storage` volume is shared and durable for replicas on one
Docker host. Multi-host replicas must provision that volume with a shared
volume driver/filesystem; never use separate node-local volumes.
## Upgrade Process
## Recovery objectives
### Step 1: Pre-flight Checks
1. Verify new Docker image is built and pushed: `docker pull gochat/gochat:${NEW_VERSION}`
2. Run database migrations on a staging environment first
3. Verify backward compatibility of migrations (new code must work with old schema)
4. Check feature flags — new features should be disabled by default
- RPO: 24 hours. Run the encrypted backup at least daily and alert if the latest
off-site bundle is older than 24 hours.
- RTO: 4 hours. Rehearse a clean-environment restore quarterly and record the
script's `rpo_seconds`, `rto_seconds`, image digest, migration version, account
count, and attachment count.
- Local and off-site backup directories must be different mounts/failure
domains. The bundle is AES-256 encrypted; keep the passphrase file in the
secret manager, never beside either backup copy.
## Daily encrypted backup
Set these operator-owned paths before any Compose command:
### Step 2: Database Migration
```bash
# Run migrations BEFORE deploying new code
# Migrations must be backward-compatible
docker compose -f docker-compose.prod.yml exec gochat /app/gochat migrate up
export GOCHAT_IMAGE_REF='ghcr.io/gochat/gochat@sha256:<digest>'
export GOCHAT_BACKUP_DIR='/mnt/backup-local/gochat'
export GOCHAT_BACKUP_OFFSITE_DIR='/mnt/gochat-offsite'
export GOCHAT_BACKUP_OFFSITE_SOURCE='backup.example.com:/gochat'
export GOCHAT_BACKUP_OFFSITE_FSTYPE='nfs4'
export GOCHAT_BACKUP_PASSPHRASE_FILE='/run/secrets/gochat-backup-passphrase'
export GOCHAT_CONNECTOR_BACKUP_NAME="connector-$(date -u +%Y%m%dT%H%M%SZ).db"
```
### Step 3: Blue-Green Deployment (Docker Compose)
Backup/restore are audited one-shot containers and run as root only to read or
rebuild Docker volumes; web and worker remain non-root.
`GOCHAT_BACKUP_OFFSITE_DIR` must be an existing external mount point provisioned
outside this Compose project. Set its approved source and filesystem type to the
exact values reported by `findmnt -M "$GOCHAT_BACKUP_OFFSITE_DIR"`; the preflight
rejects `/`, in-memory filesystems, unapproved mount metadata, and the local
backup device.
Create a consistent online Connector backup, then create and verify the single
encrypted bundle containing PostgreSQL, attachments, and that Connector copy:
```bash
# 1. Deploy new version as "green" alongside "blue" (current)
docker compose -f docker-compose.prod.yml up -d --no-deps gochat-green
# 2. Wait for health check to pass
curl -f http://gochat-green:3000/health
# 3. Switch traffic (update nginx/upstream config)
# nginx: switch upstream from blue to green
# 4. Drain old connections on blue
# Wait 30s for in-flight requests to complete
# 5. Stop blue
docker compose -f docker-compose.prod.yml stop gochat
docker compose -f deploy/docker/docker-compose.prod.yml exec shangwutong \
shangwutong backup --output "/backup/$GOCHAT_CONNECTOR_BACKUP_NAME"
docker compose -f deploy/docker/docker-compose.prod.yml --profile ops run --rm backup
```
### Step 4: Verification
1. Smoke test: hit /health endpoint
2. Check logs for errors: `docker compose logs gochat --since 5m`
3. Verify metrics: Prometheus dashboard should show normal traffic
4. Monitor for 15 minutes before finalizing
Schedule those two commands daily. Preserve their final `backup=... offsite=...`
line as audit evidence. A failed dump, archive checksum, decrypt/list check, or
off-site copy exits non-zero and must alert.
## Clean-environment restore rehearsal
Use an isolated host/project with empty volumes and an empty PostgreSQL database.
Do not point this procedure at the live project.
### Step 5: Rollback (if needed)
```bash
# Docker Compose rollback
docker compose -f docker-compose.prod.yml exec gochat /app/gochat migrate down ${N}
docker compose -f docker-compose.prod.yml up -d --no-deps gochat-${OLD_VERSION}
export COMPOSE_PROJECT_NAME=gochat-restore-$(date +%Y%m%d)
export GOCHAT_RESTORE_BUNDLE='gochat-<timestamp>.tar.enc'
docker compose -f deploy/docker/docker-compose.prod.yml --profile ops run --rm restore
```
## Migration Compatibility Rules
The restore refuses a non-empty database, attachment directory, or Connector DB.
It verifies the encrypted bundle and internal checksums before restoring, then
prints the RPO/RTO, image version, migration version, accounts, and attachments.
Afterward start web/worker with the recorded digest and verify `/health`, one
attachment URL, and Connector `readyz`.
- Migrations MUST be additive only in production (add columns, never remove)
- Column removals require a 2-phase migration: soft-remove then hard-remove
- New columns should have defaults or be nullable
- Renames require a 3-phase migration: add new → copy data → remove old
## Upgrade preflight
## Worker Upgrade
1. Record the running digest as `OLD_IMAGE`; pull and record `NEW_IMAGE` by
digest. Never derive rollback state from a mutable tag.
2. Complete the backup above and restore it in the isolated environment.
3. Stop writes or enter the maintenance window. Migration 000079 takes
`SHARE ROW EXCLUSIVE` locks; migration 000048 takes `ACCESS EXCLUSIVE` on the
rollup table. The migration job uses a 5-second lock timeout and 15-minute
statement timeout, so contention fails instead of waiting indefinitely.
4. Capture these pre-migration checks:
Workers drain naturally: set a shutdown deadline, let in-flight jobs finish,
then stop. New workers pick up queued jobs from Redis.
```sql
SELECT count(*) AS rollups FROM reporting_events_rollups;
SELECT count(*) AS swt_source_rows FROM conversations
WHERE COALESCE(custom_attributes, '{}'::jsonb) ?| ARRAY['swt_source_url','swt_source_search_term'];
SELECT account_id, (config->>'assistant_id')::bigint, count(*)
FROM agent_bots
WHERE bot_type = 'captain' AND config->>'assistant_id' ~ '^[0-9]+$'
GROUP BY 1, 2 HAVING count(*) > 1;
```
While migrating, observe lock waits with:
```sql
SELECT pid, wait_event_type, wait_event, clock_timestamp() - query_start AS elapsed, query
FROM pg_stat_activity
WHERE datname = current_database() AND state <> 'idle';
```
## One-shot migration and application rollout
Run exactly one named migration service before changing web/worker. Its output is
the audit log; `golang-migrate` is idempotent and PostgreSQL serializes competing
migration runners, but the deployment pipeline must contain only this one step.
```bash
export GOCHAT_IMAGE_REF="$NEW_IMAGE"
time docker compose -f deploy/docker/docker-compose.prod.yml --profile ops \
up --abort-on-container-exit --exit-code-from migrate migrate
docker compose -f deploy/docker/docker-compose.prod.yml logs migrate
docker compose -f deploy/docker/docker-compose.prod.yml up -d --no-deps gochat worker
curl -fsS http://127.0.0.1:3000/health
```
Post-migration checks:
```sql
SELECT version, dirty FROM schema_migrations;
SELECT count(*) AS rollups FROM reporting_events_rollups;
SELECT count(*) AS duplicate_bot_bindings FROM (
SELECT account_id, captain_assistant_id FROM agent_bots
WHERE captain_assistant_id IS NOT NULL GROUP BY 1, 2 HAVING count(*) > 1
) duplicates;
SELECT count(*) AS orphaned_inbox_bindings FROM agent_bot_inboxes b
LEFT JOIN agent_bots a ON a.id = b.agent_bot_id WHERE a.id IS NULL;
```
Migration 000048 preserves every legacy rollup row, 000076 retains legacy
Connector attributes for old-image compatibility, and 000079 repairs references
before deduplication. Any count mismatch or dirty version stops the rollout.
## Rollback
Rollback the application only, using the recorded digest:
```bash
export GOCHAT_IMAGE_REF="$OLD_IMAGE"
docker compose -f deploy/docker/docker-compose.prod.yml up -d --no-deps gochat worker
curl -fsS http://127.0.0.1:3000/health
```
Production `migrate down` and negative `migrate steps` are blocked. Never run a
destructive schema down during an application rollback. If the new schema itself
is unusable, stop web/worker and restore the verified pre-upgrade bundle into a
clean database and volumes.