We had a servercrash the other night and while none of the datadisks was affected, the db-server (7.1.3) is not starting again:
Dec 16 14:48:44 alpha postgres[12432]: [1] DEBUG: database system shutdown was interrupted at 2004-12-16 14:46:30 CET
Dec 16 14:48:44 alpha postgres[12432]: [2] DEBUG: CheckPoint record at (8, 836622368)
Dec 16 14:48:44 alpha postgres[12432]: [3] DEBUG: Redo record at (8, 836622368); Undo record at (0, 0); Shutdown FALSE
Dec 16 14:48:44 alpha postgres[12432]: [4] DEBUG: NextTransactionId: 635329587; NextOid: 3492600
Dec 16 14:48:44 alpha postgres[12432]: [5] DEBUG: database system was not properly shut down; automatic recovery in progre
ss...
Dec 16 14:48:44 alpha postgres[12432]: [6] DEBUG: redo starts at (8, 836622432)
Dec 16 14:48:44 alpha postgres[12432]: [7] DEBUG: ReadRecord: record with zero len at (8, 837544592)
Dec 16 14:48:44 alpha postgres[12432]: [8] DEBUG: redo done at (8, 837544556)
Dec 16 14:48:44 alpha postgres[12432]: [9] FATAL 2: XLogFlush: request is not satisfied
/usr/local/pgsql/bin/postmaster: Startup proc 13588 exited with status 512 - abort
I did a new initdb and recovered from the backup, but there is one minor database that was not in the backup for weeks (blame me), so I would be very happy if there is way to start postgres on the old datadir and retrieve the data.
Isnt there any way to let postgres ignore the whole XLog and just give what it has? The table in questions sure was not updated while the server crashed.
thnx, peter
-- mag. peter pilsl goldfisch.at IT-management tel +43 699 1 3574035 fax +43 699 4 3574035 [EMAIL PROTECTED]
---------------------------(end of broadcast)---------------------------
TIP 9: the planner will ignore your desire to choose an index scan if your
joining column's datatypes do not match
