AIX不完整的補丁安裝以及Oracle的Bug

今天遇到一個AIX系統BUG，Oracle數據庫無法啟動，甚至無法Nomount，實例在Nomount之后馬上Crash。
報告的錯誤是PMON無法啟動。
告警日志信息摘錄：


PMON started with pid=2, OS id=295272
Sun Dec 2 21:33:20 2007
Errors in file /oracle/admin/oinst/bdump/oinst_pmon_295272.trc:
ORA-07445: exception encountered: core dump [] [] [] [] [] []
PSP0 started with pid=3, OS id=336364
MMAN started with pid=4, OS id=385228
DBW0 started with pid=5, OS id=389154
DBW1 started with pid=6, OS id=360558
DBW2 started with pid=7, OS id=340122
DBW3 started with pid=8, OS id=274592
LGWR started with pid=9, OS id=172072
CKPT started with pid=10, OS id=151578
SMON started with pid=11, OS id=369006
RECO started with pid=12, OS id=315902
CJQ0 started with pid=13, OS id=463090
MMON started with pid=14, OS id=299320
MMNL started with pid=15, OS id=278948
Sun Dec 2 21:33:23 2007
PSP0: terminating instance due to error 472
Instance terminated by PSP0, pid = 336364

最后確定疑似Oracle的一個Bug，BUG Number：6413391
對于這個BUG，Oracle建議的修正是Apply ML06 to AIX 5.3

而客戶的系統目前看來ML06的安裝是不完全的：


instfix -i|grep ML
All filesets for 5.3.0.0_AIX_ML were found.
All filesets for 5300-01_AIX_ML were found.
All filesets for 5300-02_AIX_ML were found.
All filesets for 5300-03_AIX_ML were found.
All filesets for 5300-04_AIX_ML were found.
All filesets for 5300-05_AIX_ML were found.
Not all filesets for 5300-06_AIX_ML were found.

不完全的補丁安裝是危險的，在進行系統升級時一定要慎重，最好將系統運行在經過驗證的穩定的系統版本上，升級過快可能會帶來更多的不穩定因素。