ORA-27091 and ORA-27072 on AIX 5.3 ML12
I was wondering whether anyone had seen issues like this, specifically if the server their database was on was AIX 5.3 and had been upgraded to ML 12.
I have a 9.2.0.8 database (yes, I know - but this has also happened to a 9.2.0.6 database and a 10.2.0.3 database) and my users saw instances of the following error on the client-side:
ORA-01115: IO error reading block from file 3 (block # 1123)
ORA-01110: data file 3: '/opt/oracle/oradata01/cheese/cheesedata_01.dbf'
ORA-27091: skgfqio: unable to queue I/O
ORA-27072: skgfdisp: I/O error IBM AIX RISC System/6000 Error: 5: I/O error
They were seeing this every minute overnight during a batch run. The batch run is nothing too crazy and hasn't changed in terms of volume, etc for what sounds like eternity. The datafile that this is occurring on is always the same, but the block is always different. After a few hours, it calms down and stops sending these messages. At no time do we get any inkling that anything is wrong in either the trace dests or the alert.log on the server side.
I think it's related to MOS 559697.1, but the maxreqs is 8192. On the other databases it has affected, it is set to double this.
Anyone else seen this when moving to AIX 5.3 ML12? It happened since that move and IBM are claiming it's an Oracle issue. Oracle Support don't seem to know what the issue yet either...?
Anyone got any thoughts (other than 'upgrade' or 'use Linux' :))