Thread
-
A problem with WAL
John Summerfield <summer@os2.ami.com.au> — 2001-08-29T02:05:13Z
I've read the documentation for PostgreSQL 7.1.3 and I don't see how to tell when I can remove these: [root@dugite data]# ls pg_xlog/00000000000000* | wc -l 168 [root@dugite data]# I'd have more if this hadn't happened: [summer@possum summer]$ cat pglog 2001-08-29 04:39:56 [21803] DEBUG: XLogWrite: new log file created - consider increasing WAL_FILES 2001-08-29 04:42:28 [21803] DEBUG: XLogWrite: new log file created - consider increasing WAL_FILES 2001-08-29 04:45:47 [21803] DEBUG: XLogWrite: new log file created - consider increasing WAL_FILES 2001-08-29 04:48:32 [4945] FATAL 2: ZeroFill(/var/lib/pgsql/data/pg_x log/xlogtemp.4945) failed: No such file or directory Server process (pid 4945) exited with status 512 at Wed Aug 29 04:48:33 2001 Terminating any active server processes... 2001-08-29 04:48:33 [21803] NOTICE: Message from PostgreSQL backend: The Postmaster has informed me that some other backend died abnormally and possibly corrupted shared memory. I have rolled back the current transaction and am going to terminate your database system connection and exit. Please reconnect to the database system and repeat your query. 2001-08-29 04:48:33 [21722] NOTICE: Message from PostgreSQL backend: The Postmaster has informed me that some other backend died abnormally and possibly corrupted shared memory. I have rolled back the current transaction and am going to terminate your database system connection and exit. Please reconnect to the database system and repeat your query. Server processes were terminated at Wed Aug 29 04:48:33 2001 Reinitializing shared memory and semaphores 2001-08-29 04:48:34 [4946] DEBUG: database system was interrupted at 2001-08-29 04:48:32 WST 2001-08-29 04:48:34 [4946] DEBUG: CheckPoint record at (0, 3099970512) 2001-08-29 04:48:34 [4946] DEBUG: Redo record at (0, 3099659748); Undo record at (0, 294566376); Shutdown FALSE 2001-08-29 04:48:34 [4946] DEBUG: NextTransactionId: 9951; NextOid: 829728 2001-08-29 04:48:34 [4946] DEBUG: database system was not properly shut down; automatic recovery in progress... 2001-08-29 04:48:34 [4946] DEBUG: redo starts at (0, 3099659748) 2001-08-29 04:48:34 [4946] DEBUG: ReadRecord: record with zero len at (0, 3099987408) 2001-08-29 04:48:34 [4946] DEBUG: redo done at (0, 3099987344) 2001-08-29 04:48:36 [4946] FATAL 2: ZeroFill(/var/lib/pgsql/data/pg_x log/xlogtemp.4946) failed: No such file or directory /usr/bin/postmaster: Startup proc 4946 exited with status 512 - abort [summer@possum summer]$ The problem arises when PG consumes all available disk space. The problems I see 1) Documentation 2) Performance - there's no obvious need to use all this space 3) Performance - there's no (or insufficient) removal of old logs. 4) Recovery. I see no means of recovering short of either adding disk or deleting the entire database. -- Cheers John Summerfield Microsoft's most solid OS: http://www.geocities.com/rcwoolley/ Note: mail delivered to me is deemed to be intended for me, for my disposition.