Oracle8 BLOCKDUMPs
~~~~~~~~~~~~~~~~~~			Oracle7 BLOCKDUMPs:  <>

    A 'blockdump' allows you to dump Oracles SYMBOLLIC interpretation of
    a block on the database. For indexes you may also want to look at 
    <>

Steps:
~~~~~~
1. Determine the file and decimal block number of the block you want to dump.

2. Use the steps below to dump the block substituting the filename/number
   and block number/range:

 *From an SQL Session:


	ALTER SYSTEM DUMP DATAFILE {'filename'}|{filenumber}

	        |---BLOCK MIN {blockno} BLOCK MAX {blockno}|-->
	        |                                          |
	        |----BLOCK {blockno}-----------------------|


   Eg: ALTER SYSTEM DUMP DATAFILE 1 BLOCK 5586;
       ALTER SYSTEM DUMP DATAFILE 1 BLOCK MIN 5585 BLOCK MAX 5586;
       ALTER SYSTEM DUMP DATAFILE '/u01/oradata/MYDB/system01.dbf' BLOCK 98;

 *NOTE:
	- This command returns SUCCESS even when it sometimes fails. The
	  failure message is written to the trace file. 
	  Eg: Dumping an invalid block number returns 'Statement processed'.

	- If you omit a BLOCK (or block range) the entire file is dumped !!
	  (Up to <> )

	- To use a {filenumber} the database MUST be OPEN and the file ONLINE.
	  {filenumber} should be the absolute file number (not a relative no.).

	- If you use a 'filename' you can dump any datafile with the same
	  blocksize, even one from a different database. The session you dump
	  from must be at least a 'nomount' instance. 
	  Eg: startup nomount
	      alter system dump datafile '/other/db/datafile.dbf' block 88;

	- This command returns SUCCESS even when it sometimes fails. The
	  failure message is written to the trace file.

3. The trace should be in USER_DUMP_DEST (default $ORACLE_HOME/rdbms/log).
   (Even if USER_DUMP_DEST is set then in some cases output will still
    go to $ORACLE_HOME/rdbms/log so check both locations)

4. It is sometimes useful to get an OS blockdump also.

 *Disk dump (if block wont blockdump):           Interpreting: <>

       VMS:    DUMP/BLOCKS=(start:XXXX,end:YYYY)/out=dump.out dbfile.dbf
                VMS blocks=512 bytes. XXXX=OS block number

       UNIX:   dd if=dbfile.dbf bs=2k skip={block} count=1 | od -x > dump.out
                                   ^^block size
                skip=n blocks to skip
		It is sometimes useful to dump the block BEFORE and AFTER the
		one required. Eg: skip={block}-1 count=3
      
       NT:     There is no tool shipped as standard. Suggestions include:
                MKS Toolkit (includes the 'dd' command as above)
                FileView (Shareware)
                HEdit    (Shareware as referenced from Microsoft WWW Home Page)



.