RAC节点异常重启导致信息丢失

 

今天遇到一个奇怪的现象，RAC数据库的一个节点1异常重启，如下所示：
10:29:45 数据库重启

Beginning log switch checkpoint up to RBA [0x1a5.2.10], SCN: 9064367067124
Tue Aug  9 10:27:25 2011
Thread 1 advanced to log sequence 421 (LGWR switch)
  Current log# 1 seq# 421 mem# 0: /zyxdb/zyxdbredo_u01/zyxdb/redo01a.log
  Current log# 1 seq# 421 mem# 1: /zyxdb/zyxdbundo_u01/zyxdb/redo01b.log
Tue Aug  9 10:27:42 2011
Beginning log switch checkpoint up to RBA [0x1a6.2.10], SCN: 9064367068148
Tue Aug  9 10:27:42 2011
Thread 1 advanced to log sequence 422 (LGWR switch)
  Current log# 2 seq# 422 mem# 0: /zyxdb/zyxdbredo_u01/zyxdb/redo02a.log
  Current log# 2 seq# 422 mem# 1: /zyxdb/zyxdbundo_u01/zyxdb/redo02b.log
Tue Aug  9 10:28:29 2011
Completed checkpoint up to RBA [0x1a1.2.10], SCN: 9064367053926
Tue Aug  9 10:29:45 2011
Error: KGXGN aborts the instance (6)
Tue Aug  9 10:29:45 2011
USER: terminating instance due to error 481
Tue Aug  9 10:29:45 2011
System state dump is made for local instance
System State dumped to trace file /u01/oracle/admin/zyxdb/bdump/zyxdb1_diag_852126.trc

 

这是当时我用tail -f alert_zyxdb1.log的时候，屏幕上显示的内容。

但是系统重启后，信息却变了。这块信息丢失了。

 

Thread 1 advanced to log sequence 420 (LGWR switch)
  Current log# 10 seq# 420 mem# 0: /zyxdb/zyxdbredo_u01/zyxdb/redo10a.log
  Current log# 10 seq# 420 mem# 1: /zyxdb/zyxdbundo_u01/zyxdb/redo10b.log
Tue Aug  9 10:27:21 2011
Completed checkpoint up to RBA [0x1a0.2.10], SCN: 9064367047041
Tue Aug  9 10:27:25 2011
Beginning log switch checkpoint up to RBA [0x1a5.2.10], SCN: 9064367067124
Tue Aug  9 10:27:25 2011
Thread 1 advanced to log sequence 421 (LGWR switch)
  Current log# 1 seq# 421 mem# 0: /zyxdb/zyxdbredo_u01/zyxdb/redo01a.log
  Current log# 1 seq# 421 mem# 1: /zyxdb/zyxdbundo_u01/zyxdb/redo01b.log
Tue Aug  9 10:27:42 2011
Beginning log switch checkpoint up to RBA [0x1a6.2.10], SCN: 9064367068148
Tue Aug  9 10:27:42 2011
Thread 1 advanced to log sequence 422 (LGWR switch)
  Current log# 2 seq# 422 mem# 0: /zyxdb/zyxdbredo_u01/zyxdb/redo02a.log
  Current log# 2 seq# 422 mem# 1: /zyxdb/zyxdbundo_u01/zyxdb/redo02b.log
Tue Aug  9 10:28:29 2011
Completed checkpoint up to RBA [0x1a1.2.10], SCN: 9064367053926
Tue Aug  9 10:42:23 2011
Starting ORACLE instance (normal)
sskgpgetexecname failed to get name
LICENSE_MAX_SESSION = 0
LICENSE_SESSIONS_WARNING = 0
Interface type 1 en1 10.192.14.0 configured from OCR for use as a cluster interconnect
Interface type 1 en0 10.192.39.0 configured from OCR for use as  a public interface
Picked latch-free SCN scheme 3
Autotune of undo retention is turned on.
LICENSE_MAX_USERS = 0
SYS auditing is disabled
ksdpec: called for event 13740 prior to event group initialization
Starting up ORACLE RDBMS Version: 10.2.0.4.0.
System parameters with non-default values:
  processes                = 4000
  sessions                 = 4405
  timed_statistics         = TRUE
  instance_groups          = node1
  sga_max_size             = 107374182400
  __shared_pool_size       = 12884901888
  shared_pool_size         = 1288490188
 
 
  从Tue Aug  9 10:28:29 2011  直接跳到了 Tue Aug  9 10:42:23 2011。
 
 
怀疑数据还没来得及写磁盘AIX就重启了。可能文件这些信息来自缓存。
信息的丢失给问题诊断带来了困难。