fix: 优化 promtail 配置，调整日志解析阶段和标签提取

Author: z0z0r4

promtail/config.yaml

@@ -9,74 +9,73 @@ clients:
   - url: http://loki:3100/loki/api/v1/push
 
 scrape_configs:
-  - job_name: access
-    static_configs:
-      - targets:
-          - localhost
-        labels:
-          job: nginx-access-log
-          __path__: /var/log/nginx/access.log
-    pipeline_stages:
-      - match:
-          selector: '{job="nginx-access-log"}'
-          stages:
-            - json: # 解析 JSON 格式的日志
-                expressions:
-                  remote_addr:
-                  request_uri:
-                  verb:
-                  httpversion:
-                  response:
-                  body_bytes_sent:
-                  referrer:
-                  user_agent:
-                  http_x_forwarded_for:
-                  server_name:
-                  request_time:
-                  upstream_response_time:
-                  upstream_addr:
-                  realpath_root:
-                  cookie:
-                  nginx_version:
-                  scheme:
-                  time:
+- job_name: access
+  static_configs:
+  - targets:
+      - localhost
+    labels:
+      job: nginx-access-log
+      __path__: /var/log/nginx/access.log
+  pipeline_stages:
+  - match:
+      selector: '{job="nginx-access-log"}'
+      stages:
+      - json:  # 日志为json格式 选择需要的字段
+          expressions:
+            remote_addr:
+            request_uri:
+            verb:
+            httpversion:
+            response:
+            body_bytes_sent:
+            referrer:
+            user_agent:
+            http_x_forwarded_for:
+            server_name:
+            request_time:
+            upstream_response_time:
+            upstream_addr:
+            realpath_root:
+            cookie:
+            nginx_version:
+            scheme:
+            time:
+      - labels:
+          verb:
+          response:
+          server_name:
+          scheme:
+      - timestamp:
+          format: '2006-01-02T15:04:05+08:00'
+          source:  time
+          location: Asia/Shanghai
 
-            - labels: # 只选择低基数字段作为标签
-              verb:
-              response:
-              server_name:
-              scheme:
-            - timestamp:
-              format: "2006-01-02T15:04:05+08:00"
-              source: time
-              location: Asia/Shanghai
-
-  - job_name: nginx-error
-    static_configs:
-      - targets:
-          - localhost
-        labels:
-          job: nginx-error-log
-          __path__: /var/log/nginx/error.log
-    pipeline_stages:
-      - match:
-          selector: '{job="nginx-error-log"}'
-          stages:
-            - regex: # 使用正则选择要提取的字段
-                expression: '^(?P<timestamp>\d{4}/\d{2}/\d{2} \d{2}:\d{2}:\d{2}) \[(?P<level>\S+)\] (?P<pid>\d+)#(?P<tid>\d+): \*(?P<cid>\d+) (?P<msg>.*) client: (?P<client_ip>[^,]+), server: (?P<server_name>[^,]+), request: "(?P<request_method>\S+) (?P<request_path>\S+) (?P<request_protocol>\S+)", upstream: "(?P<upstream>[^"]+)", host: "(?P<host>[^"]+)"$'
-            - labels:
-                level:
-                pid:
-                tid:
-                cid:
-                client_ip:
-                server_name:
-                request_method:
-                request_path:
-                request_protocol:
-                upstream:
-                host:
-            - timestamp:
-                format: "2006/01/02 15:04:05"
-                source: time # 正则提取的字段
-                location: Asia/Shanghai
+- job_name: nginx-error
+  static_configs:
+  - targets:
+      - localhost
+    labels:
+      job: nginx-error-log
+      __path__: /var/log/nginx/error.log
+  pipeline_stages:
+  - match:
+      selector: '{job="nginx-error-log"}'
+      stages:
+      - regex:  # 使用正则选择要提取的字段
+          expression: '^(?P<timestamp>\d{4}/\d{2}/\d{2} \d{2}:\d{2}:\d{2}) \[(?P<level>\S+)\] (?P<pid>\d+)#(?P<tid>\d+): \*(?P<cid>\d+) (?P<msg>.*) client: (?P<client_ip>[^,]+), server: (?P<server_name>[^,]+), request: "(?P<request_method>\S+) (?P<request_path>\S+) (?P<request_protocol>\S+)", upstream: "(?P<upstream>[^"]+)", host: "(?P<host>[^"]+)"$'
+      - labels:
+          level:
+          pid:
+          tid:
+          cid:
+          client_ip:
+          server_name:
+          request_method:
+          request_path:
+          request_protocol:
+          upstream:
+          host:
+      - timestamp:
+          format: "2006/01/02 15:04:05"
+          source: time  # 正则提取的字段
+          location: Asia/Shanghai