今天開發須要刪除AWS Redshift的一個歸檔表,可是直接drop發生了以下報錯,發現有其餘對象依賴這個表,致使沒法直接刪除,可是報錯沒有直接顯示是什麼對象依賴它,多是有視圖,外鍵等約束
html
analyticsprod=# DROP TABLE IF EXISTS facts.auto_events_bk_20170223;node
ERROR: cannot drop table auto_events_bk_20170223 because other objects depend on it數據庫
HINT: Use DROP ... CASCADE to drop the dependent objects too.bash
經過\d+ auto_events_bk_20170223 查看錶結構,發現並無其餘表的外鍵約束,因此排除;ide
而後經過\dv+命令在schema:facts下查看視圖,可是發現沒有任何view存在,可是view可能存在其餘schema或db,spa
那麼問題來了,如何在整個數據庫中找到與這個表有依賴的的view?htm
這裏AWS Redshift已經給咱們提供了一個好的方法,參考連接:https://docs.aws.amazon.com/redshift/latest/dg/r_DROP_TABLE.html對象
建立一個視圖find_depend:開發
create view find_depend as select distinct c_p.oid as tbloid, n_p.nspname as schemaname, c_p.relname as name, n_c.nspname as refbyschemaname, c_c.relname as refbyname, c_c.oid as viewoid from pg_catalog.pg_class c_p join pg_catalog.pg_depend d_p on c_p.relfilenode = d_p.refobjid join pg_catalog.pg_depend d_c on d_p.objid = d_c.objid join pg_catalog.pg_class c_c on d_c.refobjid = c_c.relfilenode left outer join pg_namespace n_p on c_p.relnamespace = n_p.oid left outer join pg_namespace n_c on c_c.relnamespace = n_c.oid where d_c.deptype = 'i'::"char" and c_c.relkind = 'v'::"char";
查詢視圖找到與表auto_events_bk_20170223相關的view:get
analyticsprod=# select * from facts.find_depend where schemaname='facts' and name not in ('find_depend') order by name; tbloid | schemaname | name | refbyschemaname | refbyname | viewoid --------+------------+-------------------------+-----------------+-------------+--------- 879566 | facts | auto_events_bk_20170223 | public | map_routes2 | 965969 879566 | facts | auto_events_bk_20170223 | public | map_routes | 966155 879566 | facts | auto_events_bk_20170223 | public | map_edges | 1019697 (3 rows) analyticsprod=# set search_path to public; analyticsprod=# \dv List of relations schema | name | type | owner --------+-----------------------+------+------- public | map_edges | view | fengw public | map_edges_group | view | fengw public | map_route_edges_group | view | fengw public | map_routes | view | fengw public | map_routes2 | view | fengw public | map_routes_group | view | fengw (6 rows)
最後若是判斷這些相關的view沒有用的話,能夠經過命令drop table auto_events_bk_20170223 cascade刪除表,包括視圖一同刪除。