Преглед на файлове

处理线上数据合并效率,解决合并重复公告的问题,调整提取的消费策略

luojiehua преди 2 години
родител
ревизия
18688319ba
променени са 2 файла, в които са добавени 4 реда и са изтрити 4 реда
  1. 1 1
      BaseDataMaintenance/maintenance/dataflow.py
  2. 3 3
      BaseDataMaintenance/maintenance/dataflow_mq.py

+ 1 - 1
BaseDataMaintenance/maintenance/dataflow.py

@@ -2865,7 +2865,7 @@ class Dataflow_dumplicate(Dataflow):
 
             set_docid = set(_docids.split(","))
             if save==1:
-                set_docid.add(docid)
+                set_docid.add(str(docid))
             else:
                 if str(docid) in set_docid:
                     set_docid.remove(str(docid))

+ 3 - 3
BaseDataMaintenance/maintenance/dataflow_mq.py

@@ -539,8 +539,8 @@ class Dataflow_ActivteMQ_extract(Dataflow_extract):
 
         self.industy_url = "http://127.0.0.1:15000/industry_extract"
 
-        self.extract_interfaces = [["http://127.0.0.1:15030/content_extract",15],
-                                   ["http://192.168.0.115:15030/content_extract",7]
+        self.extract_interfaces = [["http://127.0.0.1:15030/content_extract",20],
+                                   ["http://192.168.0.115:15030/content_extract",10]
                                    ]
 
 
@@ -560,7 +560,7 @@ class Dataflow_ActivteMQ_extract(Dataflow_extract):
         self.block_url = RLock()
         self.url_count = 0
 
-        self.comsumer_count = 30
+        self.comsumer_count = 40
         self.list_extract_comsumer = []
         for _i in range(self.comsumer_count):
             listener_extract = self.ExtractListener(getConnect_activateMQ(),self.comsumer_handle)