記憶體不足導致安裝時報錯ORA-3113（一）

安裝RAC資料庫時報錯ORA-3113。

檢查dbca對應的資訊：

bash-3.00$ cd /data/oracle/cfgtoollogs/dbca/testrac/
bash-3.00$ more CreateDB.log
ORA-27102: out of memory
SVR4 Error: 12: Not enough space
ORA-03113: 通訊通道的檔案結尾

這其實是兩次安裝錯誤的彙總，第一次設定memory_target為2.8G，直接就報錯out of memory，於是將記憶體修改為2G左右，結果出現了這個ORA-3113錯誤。

檢查資料庫的alert檔案：

bash-3.00$ cd /data/oracle/diag/rdbms/testrac/testrac1/trace/
bash-3.00$ tail -100 alert_testrac1.log
System parameters with non-default values:
  processes                = 300
  sessions                 = 472
  nls_language             = "SIMPLIFIED CHINESE"
  nls_territory            = "CHINA"
  memory_target            = 2G
  db_block_size            = 16384
  compatible               = "11.2.0.0.0"
  log_archive_dest_1       = "LOCATION=+DATA"
  log_archive_format       = "%t_%s_%r.dbf"
  db_create_file_dest      = "+DATA"
  db_recovery_file_dest    = "+DATA"
  db_recovery_file_dest_size= 64G
  undo_tablespace          = "UNDOTBS1"
  instance_number          = 1
  remote_login_passwordfile= "EXCLUSIVE"
  db_domain                = ""
  dispatchers              = "(PROTOCOL=TCP) (SERVICE=testracXDB)"
  remote_listener          = ""
  audit_file_dest          = "/data/oracle/admin/testrac/adump"
  audit_trail              = "DB"
  db_name                  = "testrac"
  open_cursors             = 300
  diagnostic_dest          = "/data/oracle"
Cluster communication is configured to use the following interface(s) for this instance
  10.0.0.1
cluster interconnect IPC version:Oracle UDP/IP (generic)
IPC Vendor 1 proto 2
Fri Dec 17 12:43:31 2010
PMON started with pid=2, OS id=20946
Fri Dec 17 12:43:31 2010
VKTM started with pid=3, OS id=20951 at elevated priority
VKTM running at (10)millisec precision with DBRM quantum (100)ms
Fri Dec 17 12:43:31 2010
GEN0 started with pid=4, OS id=20955
Fri Dec 17 12:43:31 2010
DIAG started with pid=5, OS id=20963
Fri Dec 17 12:43:31 2010
DBRM started with pid=6, OS id=20965
Fri Dec 17 12:43:31 2010
PING started with pid=7, OS id=20967
Fri Dec 17 12:43:32 2010
PSP0 started with pid=8, OS id=20972
Fri Dec 17 12:43:32 2010
ACMS started with pid=9, OS id=20977
Fri Dec 17 12:43:32 2010
DIA0 started with pid=10, OS id=20979
Fri Dec 17 12:43:32 2010
LMON started with pid=11, OS id=20987
Fri Dec 17 12:43:34 2010
LMD0 started with pid=12, OS id=20989
Fri Dec 17 12:43:34 2010
RMS0 started with pid=13, OS id=21009
Fri Dec 17 12:43:34 2010
LMHB started with pid=14, OS id=21011
Fri Dec 17 12:43:34 2010
MMAN started with pid=15, OS id=21019
Fri Dec 17 12:43:34 2010
DBW0 started with pid=16, OS id=21021
Fri Dec 17 12:43:34 2010
LGWR started with pid=17, OS id=21023
Fri Dec 17 12:43:35 2010
CKPT started with pid=18, OS id=21028
Fri Dec 17 12:43:35 2010
SMON started with pid=19, OS id=21033
Fri Dec 17 12:43:35 2010
RECO started with pid=20, OS id=21035
Fri Dec 17 12:43:35 2010
MMON started with pid=21, OS id=21043
Fri Dec 17 12:43:35 2010
MMNL started with pid=22, OS id=21045
starting up 1 dispatcher(s) for network address '(ADDRESS=(PARTIAL=YES)(PROTOCOL=TCP))'...
starting up 1 shared server(s) ...
DISM started, OS id=21054
Fri Dec 17 12:44:02 2010
Exception [type: SIGSEGV, Address not mapped to object] [ADDR:0xFFFFFFFF7FFF2000] [PC:0xFFFFFFFF7CC32FA4, clskGetStackTraceBuf()+20] [flags: 0x0, count: 1]
Fri Dec 17 12:44:03 2010
Exception [type: SIGSEGV, Address not mapped to object] [ADDR:0xFFFFFFFF7FFF5F50] [PC:0x10777F280, sdbgrfcfp_convert_fileloc_pathfile()+64] [flags: 0x0, count: 1]
Errors in file /data/oracle/diag/rdbms/testrac/testrac1/trace/testrac1_lmon_20987.trc  (incident=2489):
ORA-07445: ??????: ???? [clskGetStackTraceBuf()+20] [SIGSEGV] [ADDR:0xFFFFFFFF7FFF2000] [PC:0xFFFFFFFF7CC32FA4] [Address not mapped to object] []
Incident details in: /data/oracle/diag/rdbms/testrac/testrac1/incident/incdir_2489/testrac1_lmon_20987_i2489.trc
Errors in file /data/oracle/diag/rdbms/testrac/testrac1/trace/testrac1_psp0_20972.trc  (incident=2465):
ORA-07445: ??????: ???? [sdbgrfcfp_convert_fileloc_pathfile()+64] [SIGSEGV] [ADDR:0xFFFFFFFF7FFF5F50] [PC:0x10777F280] [Address not mapped to object] []
Incident details in: /data/oracle/diag/rdbms/testrac/testrac1/incident/incdir_2465/testrac1_psp0_20972_i2465.trc
Fri Dec 17 12:46:11 2010
Exception [type: SIGSEGV, Address not mapped to object] [ADDR:0xFFFFFFFF7FFF5920] [PC:0x10777FE20, sdbgrfcvp_convert_pathinfo()+480] [flags: 0x0, count: 1]
Fri Dec 17 12:46:13 2010
Errors in file /data/oracle/diag/rdbms/testrac/testrac1/trace/testrac1_lmhb_21011.trc  (incident=2513):
ORA-07445: ??????: ???? [sdbgrfcvp_convert_pathinfo()+480] [SIGSEGV] [ADDR:0xFFFFFFFF7FFF5920] [PC:0x10777FE20] [Address not mapped to object] []
Incident details in: /data/oracle/diag/rdbms/testrac/testrac1/incident/incdir_2513/testrac1_lmhb_21011_i2513.trc
Fri Dec 17 12:46:35 2010
Exception [type: SIGSEGV, Address not mapped to object] [ADDR:0xFFFFFFFF7FFF5E50] [PC:0xFFFFFFFF7ADDA554, _write()] [flags: 0x0, count: 1]
Errors in file /data/oracle/diag/rdbms/testrac/testrac1/trace/testrac1_mmon_21043.trc  (incident=2569):
ORA-07445: ??????: ???? [_write()] [SIGSEGV] [ADDR:0xFFFFFFFF7FFF5E50] [PC:0xFFFFFFFF7ADDA554] [Address not mapped to object] []
Incident details in: /data/oracle/diag/rdbms/testrac/testrac1/incident/incdir_2569/testrac1_mmon_21043_i2569.trc
Fri Dec 17 12:48:32 2010
Trace dumping is performing id=[cdmp_20101217124832]
Fri Dec 17 12:49:04 2010
PMON (ospid: 20946): terminating the instance due to error 481
Instance terminated by PMON, pid = 20946

Oracle出現了大量的ORA-7445錯誤，其實不用查詢metalink，也不用進一步分析詳細錯誤日誌檔案，僅從錯誤的函式上就可以推斷，問題仍然是記憶體不足造成的。

在安裝過程中，曾經在作業系統上執行vmstat和ps –ef時，都出現過無法分配內容的錯誤。其實2G的分配並不多，但是由於這臺Sun480一共只有4G記憶體，還要支援Solaris 10系統，ClusterWare、ASM以及Xmanager和dbca的執行，因此空閒的記憶體只有1.6G左右，從而導致了安裝失敗。

將memory_target設定減小到1.6G以內，系統記憶體不足的錯誤不再出現。