imp 命令参数
index=n 不导索引
commit=y 以buffer数量为单位提交.如果多数据量小使用N.这样提交会以表为单位 .保证数据一致性
buffer=1000000
imp userid=\'/ as sysdba\' full=y rows=n indexes=n Constraints=y grants=y buffer=10240000 ignore=y commit=y feedback=10000 file=exp_meta_001.dmp, exp_meta_002.dmp, exp_meta_003.dmp log=imp_f.log statistics=None
imp 导完 没反应
imp的session 显示空闲等待 single-task message 也不能用alter system kill session终止 只能用kill -9
PURPOSE
-------
Getting 10046 Trace for Export and Import Utilities
SCOPE & APPLICATION
-------------------
DBA's and Support Analysts
Getting 10046 Trace for Export and Import Utilities
----------------------------------------------------
1] Run command export and let the program prompt you for the options.
$ exp system/manager
2] Open another window and attach the debugger as described below.
Run the following SQL statement to find the session that is running the export.
$ sqlplus system/manager
sql> select sid,program from
v$session where username = 'SYSTEM'
SID PROGRAM
---------- ------------------------------------------------
10 exp@machine01.com (TNS V1-V3) <-- Session running export command
sql> select s.sid, p.pid, p.spid
from v$session s, v$process p
where s.paddr = p.addr and s.sid = 10;
SID PID SPID
---------- ---------- ---------
10 15 2619
SPID from the previous query is equivalent to OSPID (operating System process).
3] Generate a trace file to Proccess ID 2619
Note:- Use SQLPLUS and connect as sysdba in 9i and higher.
$ svrmgrl <enter>
svrmgr> connect internal
svrmgr> oradebug setospid 2619
svrmgr> oradebug Event 10046 trace name context forever, level 12;
4] Switch back to the exp/imp windows and enter the parameters and then start the operation.
e.g
FULL=y
FILE=file.dmp (or the Parameters used by your export)
5] Set the Trace off once you get the Required information or the error.
svrmgr> oradebug Event 10046 trace name context off;
This will generate a trace file in the udump directory with the Process ID appended.
RELATED DOCUMENTS
-----------------
Troubleshooting a slow running import (IMP).
Solution
Consider splitting the import into two separate processes, by using the following Import parameters:
1. Run mport with INDEXFILE to build a file of index creation commands
2. Then run import with:
BUFFER -> set as large as you can
COMMIT=Y -> so don't run out of rollback
INDEXES = N -> Build indexes after importing data
3. Consider increasing SORT_AREA_SIZE to avoid/reduce sort data going to disk, but be aware that it is a per-user parameter. Run the index creation script generated from INDEXFILE above