運維筆記--postgresql佔用CPU問題定位sql
場景描述:數據庫
業務系統訪問變慢,登錄服務器查看系統負載並不高,而後查看佔用CPU較高的進程,發現是鏈接數據庫的幾個進程佔用系統資源較多。服務器
處理方式:運維
查找出佔用系統內存&CPU排名前10的進程:[或者用top命令查看] ---這裏須要注意,若是用了容器,須要進入容器內部查看相應的進程。oop
ps aux|head -1;ps aux|grep -v PID|sort -rn -k +3|head
切換到postgres用戶,執行psql,進入數據庫終端:指定上述命令找到的系統進程號post
SELECT procpid, START, now() - START AS lap, current_query FROM ( SELECT backendid, pg_stat_get_backend_pid (S.backendid) AS procpid, pg_stat_get_backend_activity_start (S.backendid) AS START,pg_stat_get_backend_activity (S.backendid) AS current_query FROM (SELECT pg_stat_get_backend_idset () AS backendid) AS S) AS S WHERE current_query <> '<IDLE>' and procpid=15874 ORDER BY lap DESC;
定位到SQL,確認該SQL完成的業務查詢功能,查看執行計劃,增長索引or 修改代碼。ui
SELECT "******_edoc_queue".id FROM "******_edoc_queue" WHERE (("*******_edoc_queue"."edoc_id" = '521300000004TCS60515001FV-960157.pdf') AND ("*****_edoc_queue"."active" = true)) ORDER BY "*****_edoc_queue"."id"
查詢該條SQL的執行計劃:(Postgresql使用explain analyze + sql語法的格式)spa
postgres=# \c ***你的實際模式schema You are now connected to database "stbg" as user "postgres". stbg=# explain analyze SELECT "cus_center_new_edoc_queue".id FROM "cus_center_new_edoc_queue" WHERE (("cus_center_new_edoc_queue"."edoc_id" = '521300000008TCS60417066FV-960101.pdf') AND ("cus_center_new_edoc_queue"."active" = true)) ORDER BY "cus_center_new_edoc_queue"."id"; ---獲得以下執行計劃: stbg= BY "cus_center_new_edoc_queue"."id";ter_new_edoc_queue"."active" = true)) ORDER QUERY PLAN ----------------------------------------------------------------------------- Sort (cost=21044.85..21044.85 rows=1 width=4) (actual time=109.905..109.905 ro ws=0 loops=1) Sort Key: id Sort Method: quicksort Memory: 25kB -> Seq Scan on cus_center_new_edoc_queue (cost=0.00..21044.84 rows=1 width=4) (actual time=109.880..109.880 rows=0 loops=1) Filter: (active AND ((edoc_id)::text = '521300000008TCS60417066FV-960101.pdf'::text)) Rows Removed by Filter: 583348 Planning time: 0.468 ms Execution time: 109.952 ms (8 rows) ----能夠看出執行查詢時間:109.952 ms