日志有三種：
pg_xlog目錄pg_wal 在線重做日志
pg_clog目錄pg_xact 事務日志文件，記錄哪些事務已完成
pg_log 數據庫日志， 記錄host,db,慢SQL信息
歸檔或者流復制發生異常的時候，事務日志會不斷生成，有可能會造成磁盤空間被塞滿，最終導致DB掛掉或者起不來。
遇到這種情況，可以先關閉歸檔或者流復制功能，備份pg_xlog日志到其他地方，不要刪除。然后刪除較早時間的的pg_xlog，有一定空間后再試着啟動Postgres。
postgresql.log只保留一周的日志，進行循環覆蓋
logging_collector = on#啟動日志
log_destination = 'csvlog'#日志格式，值為stderr,csvlog,syslog,and eventlog之一
log_filename = ‘postgresql-%a.log’
log_truncate_on_rotation = on
log_rotation_age = 1d
log_rotation_size = 0
wal日志增長量過大的問題處理，已經配置優化
$ du -sh pg_wal
395G pg_wal
$ pg_controldata
pg_control version number: 1201
Catalog version number: 201909212
Database system identifier: 6831009729171842643
Database cluster state: in production
pg_control last modified: Mon 19 Apr 2021 02:19:49 PM CST
Latest checkpoint location: 341/A0000060
Latest checkpoint's REDO location: 341/A0000028
Latest checkpoint's REDO WAL file: 0000000100000341000000A0
檢查點 341/A0000028已經執行，對應wal日志文件為0000000100000341000000A0，在這之前的日志可清理
1、mv清理到其他盤 2、pg_archivecleanup
例如：
$ ll /data/pgsql_5432/data/pg_wal/ |grep -C 10 0000000100000341000000A0
-rw------- 1 postgres postgres 16777216 Apr 19 14:11 000000010000034100000096
-rw------- 1 postgres postgres 16777216 Apr 19 14:12 000000010000034100000097
-rw------- 1 postgres postgres 16777216 Apr 19 14:13 000000010000034100000098
-rw------- 1 postgres postgres 16777216 Apr 19 14:14 000000010000034100000099
-rw------- 1 postgres postgres 16777216 Apr 19 14:15 00000001000003410000009A
-rw------- 1 postgres postgres 16777216 Apr 19 14:16 00000001000003410000009B
-rw------- 1 postgres postgres 16777216 Apr 19 14:17 00000001000003410000009C
-rw------- 1 postgres postgres 16777216 Apr 19 14:18 00000001000003410000009D
-rw------- 1 postgres postgres 16777216 Apr 19 14:19 00000001000003410000009E
-rw------- 1 postgres postgres 16777216 Apr 19 14:19 00000001000003410000009F
-rw------- 1 postgres postgres 16777216 Apr 19 14:20 0000000100000341000000A0
-rw------- 1 postgres postgres 353 Apr 19 14:56 0000000100000341000000A0.00000028.backup
-rw------- 1 postgres postgres 16777216 Apr 19 14:21 0000000100000341000000A1
-rw------- 1 postgres postgres 16777216 Apr 19 14:22 0000000100000341000000A2
-rw------- 1 postgres postgres 16777216 Apr 19 14:23 0000000100000341000000A3
-rw------- 1 postgres postgres 16777216 Apr 19 14:24 0000000100000341000000A4
-rw------- 1 postgres postgres 16777216 Apr 19 14:25 0000000100000341000000A5
-rw------- 1 postgres postgres 16777216 Apr 19 14:26 0000000100000341000000A6
-rw------- 1 postgres postgres 16777216 Apr 19 14:27 0000000100000341000000A7
-rw------- 1 postgres postgres 16777216 Apr 19 14:28 0000000100000341000000A8
-rw------- 1 postgres postgres 16777216 Apr 19 14:29 0000000100000341000000A9
-rw------- 1 postgres postgres 16777216 Apr 19 14:30 0000000100000341000000AA
0000000100000341000000A0之前的日志，可以進行清理
pg_archivecleanup 0000000100000341000000A0
wal優化配置
shared_buffers = 32GB checkpoint_completion_target = 0.1 checkpoint_timeout = 60min min_wal_size = 4GB max_wal_size = 64GB wal_log_hints = on wal_level = replica wal_keep_segments = 1000
並開啟wal壓縮，
wal_compression = on
參考閱讀:
http://www.postgres.cn/index.php/news/viewone/1/273