fix(deploy): 在 migration 后再次重置 CDC offset

问题：migration 会启动容器执行迁移，导致 CDC consumer
自动启动并消费消息。在数据库重建后启动服务时，消息
已经被消费完毕。

解决方案：在 migration 后增加 Step 7，停止容器并
再次重置 CDC offset，确保最终启动时能重新消费。

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

backend/services/deploy-mining.sh

@@ -746,16 +746,16 @@ full_reset() {
     fi
 
     echo ""
-    log_step "Step 1/7: Stopping 2.0 services..."
+    log_step "Step 1/8: Stopping 2.0 services..."
     for service in "${MINING_SERVICES[@]}"; do
         service_stop "$service"
     done
 
-    log_step "Step 2/7: Waiting for Kafka consumers to become inactive..."
+    log_step "Step 2/8: Waiting for Kafka consumers to become inactive..."
     log_info "Waiting 15 seconds for consumer group session timeout..."
     sleep 15
 
-    log_step "Step 3/7: Resetting CDC consumer offsets..."
+    log_step "Step 3/8: Resetting CDC consumer offsets..."
     # Reset offsets BEFORE migrations (which may start containers)
     for group in "${CDC_CONSUMER_GROUPS[@]}"; do
         log_info "Resetting consumer group: $group"
@@ -792,16 +792,61 @@ full_reset() {
         fi
     done
 
-    log_step "Step 4/7: Dropping 2.0 databases..."
+    log_step "Step 4/8: Dropping 2.0 databases..."
     db_drop
 
-    log_step "Step 5/7: Creating 2.0 databases..."
+    log_step "Step 5/8: Creating 2.0 databases..."
     db_create
 
-    log_step "Step 6/7: Running migrations..."
+    log_step "Step 6/8: Running migrations..."
     db_migrate
 
-    log_step "Step 7/7: Starting 2.0 services..."
+    # Stop any containers that were started during migration
+    log_step "Step 7/8: Stopping containers and resetting CDC offsets again..."
+    log_info "Migration may have started CDC consumers, stopping them now..."
+    for service in "${MINING_SERVICES[@]}"; do
+        docker compose -f "$COMPOSE_FILE" --env-file "$ENV_FILE" stop "$service" 2>/dev/null || true
+    done
+
+    log_info "Waiting 20 seconds for consumer groups to become inactive..."
+    sleep 20
+
+    # Reset CDC offsets again after migration
+    for group in "${CDC_CONSUMER_GROUPS[@]}"; do
+        log_info "Resetting consumer group: $group"
+        local reset_success=false
+        local retry_count=0
+        local max_retries=3
+
+        while [ "$reset_success" = false ] && [ $retry_count -lt $max_retries ]; do
+            if docker ps --format '{{.Names}}' 2>/dev/null | grep -q "^${KAFKA_CONTAINER}$"; then
+                if docker exec "$KAFKA_CONTAINER" kafka-consumer-groups --bootstrap-server localhost:9092 \
+                    --group "$group" \
+                    --reset-offsets \
+                    --to-earliest \
+                    --all-topics \
+                    --execute 2>&1 | grep -q "NEW-OFFSET"; then
+                    log_success "CDC offsets reset for $group"
+                    reset_success=true
+                else
+                    retry_count=$((retry_count + 1))
+                    if [ $retry_count -lt $max_retries ]; then
+                        log_warn "Consumer group still active, waiting 10s (retry $retry_count/$max_retries)..."
+                        sleep 10
+                    fi
+                fi
+            else
+                log_warn "Kafka container '$KAFKA_CONTAINER' not found"
+                break
+            fi
+        done
+
+        if [ "$reset_success" = false ]; then
+            log_warn "Could not reset offsets for $group after $max_retries attempts"
+        fi
+    done
+
+    log_step "Step 8/8: Starting 2.0 services..."
     for service in "${MINING_SERVICES[@]}"; do
         service_start "$service"
     done