Table number changing

C

casorohi

Member
Hi guys need your help. I am doing DB upgrade from V9 to 102B on linux 64bit, my problem is table numbers are chanced while creating a new DB. Below are steps I am doing
1. First upgrade the DB to V10 through conv910 ( to avail the multithread dump)
2. Created a new empty/template DB with same df in above converted DB and add more storages area for Type-II (empty/template DB contains old DB area as well new area for type II storage)
3. Moved the table to new area (Type-II) through tablemove in empty/template DB
4. Dumped the .df files from empty/template DB
5. Create a new Type-II DB and load the .df from empty/template DB
6. I found the new DB created in step 6, the table numbers are changed – means I have to compile the code
The idea is to do dump and Load for type-II, another thing while doing binary all the records are not dumped, means a table have 250000 records, but only 21000 + records are dump and no error/alert in dump log file.

Any idea/ways how to keep the same table number.
Thanks
 
You need to recompile the code anyway when you upgrade from any Progress V9 to any OpenEdge 10 version. There's no way around that.

Before doing the migration you should do an index check - to me it sounds like your database suffers some sort of index corruption. Generally, when doing binary dump & load table number are not relevant. During the binary load all indexes on the table are deactivated - therefore you need to make sure that you rebuild the indexes afterward.

What are the exact commands you are using to dump & load?

Heavy Regards, RealHeavyDude.
 
Maybe I don't understand completely, but shouldn't you dump and load the sequences also? In my database it's not an issue, but I remember some warnings about missing sequences and lines that got replaced, leaving your database inconsistent.

And I once had a dump loading proces that was missing a lot of records because I didn't have sufficient rights, stupid as it might sound, that is something to check aswell.
 
Why are there two "create a new db" steps?

The easy way to add new storage areas to a db that you are about to load:

1) dump .df
2) grep -v " AREA" dbname.df > new.df
3) load new.df into an empty shell
4) use a script & proutil tablemove to assign tables and indexes to new areas (the db is empty so this is very fast)
5) proceed to load data
 
A difference is record counts is usually because of a problem. There should be error messages somewhere -- but if you don't log output to a file they may be lost.

Or it might be that a second .bd file was created and you only loaded one of them.

Or it could be something procedural on your end -- maybe you overwrote the file with a shorter one somehow.
 
You must log in or register to reply here.
Back
Top