如何跟踪dbca工具，Tracing the Database Configuration Assistant (DBCA)

Comments

1. admin says

   2010/08/17 at 12:16

   DBCA APPEARS TO HANG AFTER CLICKING FINISH BUTTON
   Applies to:

   Oracle Database Configuration Assistant – Version: 10.2.0.1 to 10.2.0.4
   Information in this document applies to any platform.
   Symptoms

   Trying to create a database using the DBCA
   After configuring all options successfully, clicking finish does nothing.

   There is no progress, and no errors are returned

   Looking at the dbca trace.log does not show any errors, just the following:
   [AWT-EventQueue-0] [11:20:29:828] [DBCAWizard.onFinish:1139] m_bFinishClicked: true
   [AWT-EventQueue-0] [11:20:30:2] [DBCAWizard.onFinish:1139] m_bFinishClicked: true
   [AWT-EventQueue-0] [11:20:35:126] [SQLEngine.done:1958] Done called

   In 10.2 trace information should be written automatically to the following location

   $ORACLE_HOME/cfgtoollogs/dbca/trace.log

   In 11g, the location of trace.log has changed to:

   $ORACLE_BASE/cfgtoollogs/dbca//trace.log

   Cause

   In most cases, it appears that an X client is being used to run the dbca, like Xming.

   Solution

   To use Xming, you have to install the complete font package.

   Optionally, run dbca locally from the console, using a different X client or using VNC.

   Reflections X and Exceed are also known to work.

   回复
2. admin says

   2010/08/17 at 12:18

   DBCA Tips and Pitfalls in a Windows RAC 9i Environment

   Tips And Pitfalls Using DBCA in a Windows RAC 9i Environment:
   ===========================================================

   Under normal circumstances, if the Database Configuration Assistant (DBCA) is
   invoked via the Oracle Universal Installer (OUI) and/or independently
   following successful install, then none of the following documented steps
   should be necessary.

   However, in environments where DBCA should happen to fail, here are some tips
   and troubleshooting actions that may be followed to get DBCA to perform the
   desired database creation.

   It should be noted that manual creation of a database – even in a clustered
   environment – is still a very valid option for creating your database and
   should be considered against debugging DBCA as a way of meeting a stringent
   deadline. In which case Note.137288.1 Manual Database Creation in
   Oracle9i (Single Instance and RAC) details the process to be followed.

   Topics Covered in this Article:
   ===============================

   1. Run the Global Services Daemon (GSD) in the foreground
   2. Trace DBCA during database creation
   3. Cleaning up after a failed DBCA create cluster database attempt
   4. DBCA will fail if Oracle Net is not configured properly
   5. spfile.ora does not point to a logical partition
   6. What is happening when DBCA seems to hang around 95-100%
   7. Miscellaneous configuration which may prevent DBCA from succeeding
   + Sample Configuration Files

   **********************************************************************

   1. Run the Global Services Daemon (GSD) in the Foreground
   ———————————————————

   The DBCA requires that the GSD be run in the foreground on each cluster
   node in order to function properly.

   1. Start the GSD on ALL cluster nodes in the foreground before invoking the
   DBCA.

   If the GSD is already running:

   – Stop the GSD using commandline: gsdctl stop
   OR
   – Use the control panel services applet

   THEN run ‘gsd.bat’ from the command line and leave this window open,
   undisturbed (gsd.bat is found in $ORACLE_HOME\bin)

   Repeat on all nodes.

   The output on each node will display the name of the node where the gsd.bat
   was first invoked and it will repeat that name over and over in a ping-like
   fashion.

   Do not close the window from which you invoke gsd.bat until the DBCA is
   completely finished and you have exited the DBCA properly.

   Example output of gsd.bat command (assuming gsd.bat has been invoked from
   NODE1 first):

   Example, from NODE1:

   E:\>gsd.bat
   “E:\Program Files\Oracle\jre\1.1.8\bin\jre” -DTRACING.ENABLED=false
   -DTRACING.LEVEL=2 -classpath
   (etc…)
   NODE1READY
   )VNODE1

   Example, from NODE2:

   E:\>gsd.bat
   “E:\Program Files\Oracle\jre\1.1.8\bin\jre” -DTRACING.ENABLED=false
   -DTRACING.LEVEL=2 -classpath
   (etc…)
   NODE2READY
   )VNODE1

   2. Trace DBCA During Database Creation
   —————————————

   Generating a DBCA trace file, in the event the DBCA should fail, will save
   time on senseless guessing as to what caused the failure.

   To generate a trace file for the DBCA it is necessary to edit the dbca.bat
   file. The dbca.bat file can be found in $ORACLE_HOME\bin and can be edited
   with a tool such as notepad. Remember to make a copy of the original dbca.bat
   before making any changes to this file.

   Note: This information is specific to Windows. If you are seeking
   instructions for tracing the DBCA on UNIX, please see
   Note.188134.1 Tracing the Database Configuration Assistant
   (DBCA).

   1. Use an editing tool such as Notepad or PFE to edit. Tracing information
   should be added to line 40 of the dbca.bat.

   Line 40 should now appear as follows:

   “E:\Program Files\Oracle\jre\1.1.8\BIN\JRE” -DTRACING.ENABLED=true
   -DTRACING.LEVEL=2 -DORACLE_HOME=”%OH%” …

   2. To indicate where the trace file should be created, invoke DBCA from the
   command line using the following syntax:

   E:\> e:\oracle\ora920\bin\dbca.bat > e:\dbcatrace.txt

   When DBCA is run, the file ‘dbcatrace.txt’ will be generated in the directory
   indicated (in this case, e:\).

   3. Cleaning up after a failed DBCA create cluster database attempt
   ——————————————————————-

   As with standalone databases, the DBCA will read the registry first to
   determine if a particular instance name is already in use. If the SID is
   known either in the registry or in the SRVM repository then the DBCA will
   not allow you to reuse that particular SID name on the second create
   database attempt.

   1. Use the ORADIM utility at the command line to remove the OracleServiceSID
   from the Windows registry. Issue this command on each node, remembering
   to edit the SID name for each node)

   The oradim utility is found in the $ORACLE_HOME\bin directory.

   Example:

   Assume failed DBCA cluster database attempt where SID prefix =
   ‘RACNT’

   From Node1:

   oradim -delete -sid ‘RACNT1’

   From Node2:

   oradim -delete -sid ‘RACNT2’

   2. Cleaning up the srvm repository: If this is the first cluster database
   to be added to the cluster, the srvm repository may be reinitialized.
   This should only be performed if this is the only cluster database known
   to the SRVM repository. This step will be necessary if the DBCA create
   database attempt has failed and you wish to use the same SID when
   attempting DBCA create database a second time.

   A. The GSD must first be stopped on all cluster nodes. Use gsdctl stop
   or stop the OracleGSDService from the control panel services applet.

   B. To reinitialize the repository, issue the command:

   srvconfig -init -f

   When DBCA is invoked again, it will no longer report the presence of database
   RACNT and it will allow this name to be used again.

   Note: It is also possible to trace GSD – see Note.178683.1 Tracing GSD,
   SRVCTL, GSDCTL, and SRVCONFIG for details.

   4. DBCA will fail if Oracle Net is not configured properly
   ————————————————————

   One of the tasks the DBCA will perform will be to configure your spfile and/or
   init.ora file parameters for this cluster database. Some parameters – such as
   ‘LOCAL_LISTENER’ will require that the proper net configuration is in place in
   order to allow the database to open. This is the reason that an improper net
   configuration could cause the DBCA to fail.

   As explained in Note.188555.1 ORA-119, ORA-132, and ORA-1078 When Using DBCA
   to Create a RAC Database, one of the requirements for the DBCA to run properly
   is that the Net Configuration Assistant (NETCA) must first have been run
   successfully.

   Therefore it is strongly advised that you do not manually edit the listener.ora
   and tnsnames.ora file information, but rather you leave the necessary
   configuration steps to the NETCA.

   In the event that these files have been edited manually since installation time,
   below are some troubleshooting tips that may help to get the DBCA to create the
   desired database.

   Net requirements before invoking the DBCA in a RAC environment are explained
   here in three categories:

   A. listener.ora configuration
   B. tnsnames.ora configuration
   C. init.ora configuration

   A. LISTENER.ORA

   Listener.ora is usually configured at installation time when the OUI
   invokes the Net Configuration Assistant (NETCA). At this time, a
   very simple listener.ora file is generated and stored in
   $ORACLE_HOME\network\admin.

   If for any reason the NETCA is not run successfully at installation
   time, it is necessary for the DBCA that this tool be run independently
   thereafter.

   When using DBCA to create your cluster database it is absolutely
   unnecessary – AND ILL ADVISED – to manually edit the listener.ora file.

   See Sample Configuration Files ‘A1’ here below for example listener.ora
   files before and after DBCA is run.

   B. TNSNAMES.ORA

   Although not clearly documented, in order to run the DBCA successfully
   in your RAC environment, proper configuration of the tnsnames.ora file
   is critical.

   As with the listener.ora information it would be ill advised to edit
   the tnsnames.ora file manually as this may cause the database not to
   open and the DBCA to fail!

   The best way to begin is with the most simple tnsnames.ora information
   possible and to allow DBCA to put in the information it will need to
   create the instances, etc.

   See Sample Configuration Files ‘B1′ here below for example tnsnames.ora
   files before and after DBCA is run.

   C. INIT.ORA (net configuration)

   The DBCA will configure the SPFILE to contain the following parameters
   that correspond directly to the necessary parameters in the tnsnames.ora
   file:

   RACNT1.local_listener=’LISTENER_RACNT1′
   RACNT2.local_listener=’LISTENER_RACNT2′

   *.remote_listener=’LISTENERS_RACNT’

   Note: Each of these init.ora entry values must correspond to a
   tnsnames.ora configured service entry.

   5. spfile.ora does not point to a logical partition
   ————————————————————-

   While running Database Configuration Assistant (DBCA) to create a database
   with datafiles to be stored on Oracle Cluster File System (OCFS) in an
   Oracle 9i Real Application Clusters environment on Windows the following
   message is encountered:

   c:\oracle\ora92\database\spfileRACNT.ora does not point to a logical partition.

   The correct way to invoke DBCA in a RAC on Windows environment to create a
   database that will have its datafile storage on an OCFS partition is as
   documented in:

   Note:178882.1 : Step-By-Step Install of RAC with OCFS on Windows 2000

   invoke DBCA from the command line as instructed by Note:178882.1 :

   Step-By-Step Install of RAC with OCFS on Windows 2000

   dbca -datafileDestination O:\oradata

   If you don’t issue this command exactly, you will get the error as described
   above.

   A. Possible cause #1:
   ==================

   Failure to capitalize the drive letter …

   dbca -datafileDestination o:\oradata

   instead of:

   dbca -datafileDestination O:\oradata

   B. Possible cause #2:
   ==================
   Failure to include backslash,

   dbca -datafileDestination O:oradata

   instead of:

   dbca -datafileDestination O:\oradata

   Be sure to use the following syntax to invoke DBCA in a RAC on Windows
   environment to create a database that will have its datafile storage on an
   OCFS partition EXACTLY as documented:

   dbca -datafileDestination O:\oradata

   6. What is happening when DBCA seems to hang around 95-100%
   ———————————————————–

   Do not panic. What appears to be a ‘hang’ may actually be a ‘pause’.

   It is not uncommon if the DBCA seems to ‘pause’ at 95-100% completion. At the
   end of the DBCA create cluster database process additional instances are
   created on the remote nodes and additional log threads are assigned to these
   instances.

   When the DBCA reaches 100% completion, the Password Management screen should
   appear. Make certain that the password management screen is not just simply
   ‘hiding’ behind the DBCA applet. The DBCA window with status 100% will not go
   away until the Password Management screen has been exited.

   Other possible causes of probable DBCA hangs are covered in
   Note.213416.1 RAC Troubleshooting Windows NT/2000 Service Hangs

   7. Miscellaneous configuration that may prevent DBCA from succeeding
   ——————————————————————–

   1. Note.223554.1 Automatic Startup of the Intelligent Agent Fails in RAC
   Environment.

   2. Be sure that your OracleGSDService is configured to run under a user
   account that is a member of the ORA_DBA group. By default, the
   OracleGSDService is configured to run under the Local System account
   this change must be made manually. Please see the Windows Administrator
   guide for more information.

   3. There are srvm and dbca updates packaged wintin the CFS bundle for
   Windows. If you are using these executables from the CD installation,
   please obtain the CFS bundle from Oracle Tech Net whether you plan to
   use RAW or OCFS datafiles:

   Oracle Real Application Clusters, Cluster File System for Windows NT/2000
   http://otn.oracle.com/software/products/oracle9i/htdocs/winsoft.html

   4. Do not delete all control files in the Storage screen of DBCA. If you
   want to change the File Name or Directory from the default, simply edit
   the names that are given. Removing all control files listed will have
   the result that you will be unable to add any new control file names on
   this screen. If you have done so, then when attempting to proceed to
   the next window of the DBCA you will not be allowed since at least two
   controlfiles must be specified in order to have DBCA create a database
   for you.

   5. The DBCA template may indicate that the datfafiles will be created
   with AUTOEXTEND = true. However, if the datafiles will be created on
   raw devices, this will not be the case. After database creation, it
   will be necessary to alter the datafiles to trun on AUTOEXTEND if this
   is desired.

   To check the AUTOEXTENSIBLE status of the datafiles in the database,
   perform the following query:

   SQL> select file_name, autoextensible from dba_data_files;

   To set the AUTOEXTEND flag to ‘ON’ for particular or all datafiles,
   the following command must be issued:

   SQL> alter database datafile ‘\\.\TOOLS01.DBF’ autoextend on
   maxsize 80M;

   Note: Setting datafiles stored on raw devices to AUTOEXTEND in a Windows
   environment is not recommended.

   ********************************************************************************
   If you have chosen not to create a database during the Oracle Universal Installer
   installation of your RDBMS / RAC option, then you must invoke NETCA and create
   listeners for your cluster nodes before invoking DBCA to create your RAC
   database.

   Simply invoke NETCA, choose cluster configuration, choose all cluster nodenames
   from the list, ‘Add Listener’ and leave the default name of LISTENER.

   Assuming node names =
   racnode1
   racnode2

   Database name =
   racdb

   Your LISTENER.ORA file should look like this one BEFORE and AFTER
   running DBCA in this environment:

   *****

   LISTENER =
   (DESCRIPTION_LIST =
   (DESCRIPTION =
   (ADDRESS_LIST =
   (ADDRESS = (PROTOCOL = TCP)(HOST = racnode1)(PORT = 1521))
   )
   )
   )

   SID_LIST_LISTENER =
   (SID_LIST =
   (SID_DESC =
   (SID_NAME = PLSExtProc)
   (ORACLE_HOME = C:\oracle\ora92)
   (PROGRAM = extproc)
   )
   (SID_DESC =
   (ORACLE_HOME = C:\oracle\ora92)
   (SID_NAME = racdb1)
   )
   )

   Once the listeners have been created, you can run DBCA which will create your database
   as well as update your SRVM repository with information regarding this database
   AND update your tnsnames.ora files and initialization parameter files as
   follows:

   *****
   *****
   *****
   TNSNAMES.ORA file:
   *****
   # TNSNAMES.ORA Network Configuration File: C:\oracle\ora92\network\admin\tnsnames.ora
   # Generated by Oracle configuration tools.

   racdb =
   (DESCRIPTION =
   (ADDRESS_LIST =
   (ADDRESS = (PROTOCOL = TCP)(HOST =racnode1)(PORT = 1521))
   (ADDRESS = (PROTOCOL = TCP)(HOST = racnode2)(PORT = 1521))
   (LOAD_BALANCE = yes)
   )
   (CONNECT_DATA =
   (SERVER = DEDICATED)
   (SERVICE_NAME = racdb)
   )
   )

   LISTENER_racdb2 =
   (ADDRESS = (PROTOCOL = TCP)(HOST = racnode2)(PORT = 1521))

   LISTENER_racdb1 =
   (ADDRESS = (PROTOCOL = TCP)(HOST = racnode1)(PORT = 1521))

   racdb1_STARTUP =
   (DESCRIPTION =
   (ADDRESS = (PROTOCOL = TCP)(HOST = racnode1)(PORT = 1521))
   (CONNECT_DATA =
   (SID = racdb1)
   )
   )

   racdb2 =
   (DESCRIPTION =
   (ADDRESS = (PROTOCOL = TCP)(HOST = racnode2)(PORT = 1521))
   (CONNECT_DATA =
   (SERVICE_NAME = racdb)
   (INSTANCE_NAME = racdb2)
   )
   )

   racdb1 =
   (DESCRIPTION =
   (ADDRESS = (PROTOCOL = TCP)(HOST = racnode1)(PORT = 1521))
   (CONNECT_DATA =
   (SERVICE_NAME = racdb)
   (INSTANCE_NAME = racdb1)
   )
   )

   INST1_HTTP =
   (DESCRIPTION =
   (ADDRESS_LIST =
   (ADDRESS = (PROTOCOL = TCP)(HOST = racnode1)(PORT = 1521))
   )
   (CONNECT_DATA =
   (SERVER = SHARED)
   (SERVICE_NAME = MODOSE)
   (PRESENTATION = http://HRService)
   )
   )

   racdb2_STARTUP =
   (DESCRIPTION =
   (ADDRESS = (PROTOCOL = TCP)(HOST = racnode2)(PORT = 1521))
   (CONNECT_DATA =
   (SID = racdb2)
   )
   )

   LISTENERS_racdb =
   (ADDRESS_LIST =
   (ADDRESS = (PROTOCOL = TCP)(HOST = racnode1)(PORT = 1521))
   (ADDRESS = (PROTOCOL = TCP)(HOST = racnode2)(PORT = 1521))
   )

   *****
   *****
   *****
   Initializiation parameters were:

   RACDB1.LOCAL_LISTENER=LISTENER_racdb1
   RACDB2.LOCAL_LISTENER=LISTENER_racdb2
   *.REMOTE_LISTENER=LISTENERS_racdb
   *****
   *****
   *****
   Subsequent runnings of DBCA to create additional Real Application Clusters
   databases require no manual intervention for the NET files.
   Simply run DBCA to create the additional databases and your tnsnames.ora
   file as well as your initializiation parameter file will be updated as
   appropriate.

   回复
3. admin says

   2010/08/17 at 12:18

   PRKS-1011 And CRS-0210 creating an ASM based RAC database using DBCA
   Applies to:

   Oracle Server – Enterprise Edition – Version: 10.1.0.2 and later [Release: 10.1 and later ]
   IBM AIX on POWER Systems (64-bit)
   IBM AIX Based Systems (64-bit)
   new install of Oracle 10g RAC.
   Symptoms

   dbca is failing with the following errors while creating a RAC database using ASM :
   For node 1:
   PRKS-1011 – Failed to check status of ASM instance “+ASM1” on node “oraclust1”,
   [CRS-0210: Could not find resource ora.oraclust1.ASM1.asm.]

   For node 2:
   PRKS-1011 – Failed to check status of ASM instance “+ASM2” on node “oraclust2”,
   [CRS-0210: Could not find resource ora.oraclust2.ASM2.asm.]

   After turning on dbca tracing ( -DTRACING.ENABLED=true -DTRACING.LEVEL=2)
   the dbca trace shows the following erros when it tries to register the ASM instance with CRS:
   Thread-36] [19:14:48:332] [RuntimeExec.runCommand:71] Calling Runtime.exec() with the command
   [Thread-36] [19:14:48:332] [RuntimeExec.runCommand:73]
   /oracle/app/oracle/OraHome_1/bin/crs_register
   [Thread-36] [19:14:48:332] [RuntimeExec.runCommand:73] ora.oraclust1.ASM1.asm
   [Thread-76] [19:14:48:385] [StreamReader.run:61] In StreamReader.run
   [Thread-36] [19:14:48:386] [RuntimeExec.runCommand:127] runCommand: Waiting for the process
   [Thread-75] [19:14:48:386] [StreamReader.run:61] In StreamReader.run
   [Thread-75] [19:14:52:373] [StreamReader.run:65] OUTPUT>2004-07-30 19:14:52.337: ALERT
   [READMESSAGE:1] Receive message header caa_clsrecv ret 11
   [Thread-75] [19:14:52:376] [StreamReader.run:65] OUTPUT>2004-07-30 19:14:52.375: ALERT
   [REQUEST_REPLY_:1] Error reading response IOException : Didn’t receive header part of message
   [Thread-75] [19:14:52:376] [StreamReader.run:65] OUTPUT>(File: caa_Message.cpp, line: 695)
   [Thread-76] [19:14:52:377] [StreamReader.run:65] ERROR>CRS-0184: Cannot communicate with the CRS daemon.
   [Thread-76] [19:14:52:386] [StreamReader.run:65] ERROR>
   [Thread-36] [19:14:52:404] [RuntimeExec.runCommand:129] runCommand: process returns 184
   [Thread-36] [19:14:52:405] [RuntimeExec.runCommand:143] RunTimeExec: output>
   [Thread-36] [19:14:52:405] [RuntimeExec.runCommand:146] 2004-07-30 19:14:52.337: ALERT
   [READMESSAGE:1] Receive message header caa_clsrecv ret 11
   [Thread-36] [19:14:52:405] [RuntimeExec.runCommand:146] 2004-07-30 19:14:52.375: ALERT
   [REQUEST_REPLY_:1] Error reading response IOException : Didn’t receive header part of message
   [Thread-36] [19:14:52:405] [RuntimeExec.runCommand:146] (File: caa_Message.cpp, line: 695)
   [Thread-36] [19:14:52:406] [RuntimeExec.runCommand:151] RunTimeExec: error>
   [Thread-36] [19:14:52:406] [RuntimeExec.runCommand:154] CRS-0184: Cannot communicate with the
   CRS daemon.
   [Thread-36] [19:14:52:406] [RuntimeExec.runCommand:154]

   Cause

   These messages indicate that the CRS daemon crashed while registering the resource.

   The default ulimits on AIX – even for the root user – are set to:
   time(seconds) unlimited
   file(blocks) 2097151
   data(kbytes) 131072
   stack(kbytes) 32768
   memory(kbytes) 32768
   coredump(blocks) 2097151
   nofiles(descriptors) 2000

   These limits (data, stack or memory) appear to be too low (for the CRS daemon to successfully register the ASM instances.
   Please note that the CRS daemon does start with these ulimits but will fail later.
   Solution

   After setting the ulimits to unlimited on the root account, dbca could successfully register the ASM instances with CRS:
   time(seconds) unlimited
   file(blocks) unlimited
   data(kbytes) unlimited
   stack(kbytes) 2097152
   memory(kbytes) unlimited
   coredump(blocks) 2097151
   nofiles(descriptors) unlimited

   This needs to done on all nodes of the cluster. The CRS daemon needs to be restarted (e.g. via reboot).

   On AIX the default ulimits and ulimits for individual users are set in /etc/security/limits. Add an entry for the root user using the following values* (these values should be set for root and Oracle as per Bug:7330414):

   root:
   fsize = -1
   core = 2097151
   cpu = -1
   data = 524288
   rss = 524288
   stack = 524288
   stack_hard = 524288
   nofiles = 2000

   Verify the current settings – after logging out and back in as root – with
   # ulimit -a

   For more information see the AIX man page for ‘ulimit’.

   回复