Kbase 19150: Clarifications on Use of Binary Dump and Load
| Autor |
Progress Software Corporation - Progress |
| Acesso |
Público |
| Publicação |
30/07/2007 |
|
Status: Verified
GOAL:
Discussion of the two methods to perform a binary dump and load of your Progress database
GOAL:
Methods to perform binary dump and load?
GOAL:
What are the different methods to perform binary dump and load?
GOAL:
Binary dump
GOAL:
Binary Load
FACT(s) (Environment):
Progress 8.X
Progress 9.X
OpenEdge 10
FIX:
Both processes to perform binary dump and loads dump records, they do not dump database blocks.
That means that database fragmentation is eliminated and that indexes must be rebuilt. This process is equivalent to a dictionary dump followed by bulkload and idxbuild.
Both processes require that the source and target table physical field layout match exactly (make sure the fields have the same Cyclical Redundancy Check [CRC]). You can obtain the CRC of a table from the _file record, _crc field.
If, for example, you have database (A) and database (B) and want to list the non-matching CRCs on identically named tables, you could run the following:
FOR EACH a._file NO-LOCK WHERE a._file._file-number > 0,
FIRST b._file NO-LOCK WHERE b._file._file-name = a._file._file-name AND
b._file._crc <> a._file._crc:
DISPLAY a._file._file-name
a._file._crc
b._file._crc.
END.
You should not perform this process unless you are 100.00001% sure that both source and target databases have matching physical field structure. If you dump and load the definitions of a table, you get different physical field layouts if there were any deleted fields in the definitions.
However, please note the following from the v9 Database Administration Guide and Reference, specifically the reference section for proutil load:
For example, when you dump a table, the PROUTIL utility calculates a CRC value for the table and stores it in the header of the binary dump file. When you load the table, PROUTIL matches the CRC value stored in the header with the CRC value of the target table. The values must match or the load is rejected.
You can load binary dump files created with a previous version of the PROUTIL DUMP utility, because the current version of PROUTIL LOAD uses the CRC value established when the file was originally dumped. Consequently, the OpenEdge database maintains backwards compatibility.
However, you cannot use PROUTIL LOAD from Version 8.3 or earlier to load a binary dump file created using the Version 9.0 or later of the PROUTIL DUMP utility. The earlier versions of PROUTIL DUMP and LOAD did not use CRC values to establish the criteria for loading, but instead used other mechanisms,...
---------------------------------------------------------
The upshot is that you cannot binary load with v8 any dump file created with v9, while the reverse applies - a file created with a v8 dump can be loaded with a v9 load.
The two methods of binary dump and load are:
- dbrpr dump/load (the old way):
This requires additional considerations. It is not supported by Progress and is not officially documented. You should not use this method unless you fully understand the concept of field relative position (rpos).
Dump with the following steps:
1) proutil -C dbrpr.BR>
2) Menu option #1.
3) Choose the dump option.
4) Load with proutil -C dbrpr, option "Load RM Dump File".
5) Do an idxbuild afterward.
Indexes are not disabled, so without an idxbuild, you cannot find the loaded records except via RECID reads.
- proutil -C dump/proutil -C load (the supported method):
This method is documented and supported, and initially was available in Version 8.2. (Earlier versions do not disable indexes so you must idxbuild.)
The PROUTIL method is recommended for all large dump/loads because it is significantly faster than dictionary dump/loads. Keep in mind that proutil -C dump/load accepts normal database connection parameters.
Progress recommends that you use -RO (read-only) when you dump and -1 -i (single-user
o-crash recovery) when you load.
The -RO (client) parameter should not be used to connect to a database that is started as a server. Supply a -B parameter to allocate a medium size cache during the dump.
Load is a little faster without a -B because it is a sequential operation..