On Tue, 2007-06-05 at 14:42 +0000, Herve Boulouis wrote: > PostgreSQL version: 8.2.4 > Operating system: IRIX 6.5.26
Thats not yet a supported system, but we should fix this. > Jun 5 15:23:11 6T:origin300 postgres[2009]: [5-1] LOG: database system is > shut down > * 2d start > Jun 5 15:23:17 6T:origin300 postgres[2027]: [1-1] LOG: could not create > IPv6 socket: Address family not supported by protocol family > Jun 5 15:23:17 6T:origin300 postgres[2027]: [2-1] LOG: could not create > socket for statistics collector: Address family not supported by protocol > family > Jun 5 15:23:17 6T:origin300 postgres[2027]: [3-1] LOG: trying another > address for the statistics collector > Jun 5 15:23:17 6T:origin300 postgres[2028]: [4-1] LOG: database system was > shut down at 2007-06-05 15:23:11 GMT > Jun 5 15:23:17 6T:origin300 postgres[2028]: [5-1] LOG: record with zero > length at 0/486718 > * vvv I believe this line is not normal, as there was a clean shutdown > Jun 5 15:23:17 6T:origin300 postgres[2028]: [6-1] LOG: invalid primary > checkpoint record > Jun 5 15:23:17 6T:origin300 postgres[2028]: [7-1] LOG: using previous > checkpoint record at 0/4866C8 > Jun 5 15:23:17 6T:origin300 postgres[2028]: [8-1] LOG: redo record is at > 0/4866C8; undo record is at 0/0; shutdown TRUE > Jun 5 15:23:17 6T:origin300 postgres[2028]: [9-1] LOG: next transaction > ID: 0/593; next OID: 10820 > Jun 5 15:23:17 6T:origin300 postgres[2028]: [10-1] LOG: next MultiXactId: > 1; next MultiXactOffset: 0 > Jun 5 15:23:17 6T:origin300 postgres[2028]: [11-1] LOG: database system > was not properly shut down; automatic recovery in progress > Jun 5 15:23:17 6T:origin300 postgres[2028]: [12-1] LOG: record with zero > length at 0/486718 > Jun 5 15:23:17 6T:origin300 postgres[2028]: [13-1] LOG: redo is not > required > Jun 5 15:23:17 6T:origin300 postgres[2028]: [14-1] LOG: database system is > ready Well, I'd question what's going on with the hardware/filesystem or the settings you've chosen for fsync and wal_fsync. If you shut it down cleanly, yet its flaky when it comes back up that doesn't sound too much like a database problem to me... -- Simon Riggs EnterpriseDB http://www.enterprisedb.com ---------------------------(end of broadcast)--------------------------- TIP 1: if posting/reading through Usenet, please send an appropriate subscribe-nomail command to [EMAIL PROTECTED] so that your message can get through to the mailing list cleanly
