Talk:Mfg Libero migration project

This Wiki is read-only for reference purposes to avoid broken links.

Hi Wight (Tim) I m ready to assist in the 2pack understanding. You should not create the minutes page but use this discussion talk page for those points there instead (u can move those here, so that it is intuitional to find stuff. And need not create link.

About your background i wana ask, have u done Compiere before? Anyway i m happy u are here. Welcome and as i said i m happy to help. Teaching is my forte.

Red1 17:14, 22 November 2006 (EST)

Entry needs to be remaned

I propose Kompiere libero migration project, as this Kompiere libero is the official name in Sourceforge.--Khalid HASSANI 18:04, 22 November 2006 (EST)

Notes on project

Oracle seed dump for libero according to Carlos Ruiz: http://svn.sourceforge.net/viewvc/adempiere/tags/release-2.5.3b-initial/data/seed/Compiere.jar

Progress on 12/19/2006

I installed oracle 10.1 on centos 4.3. I compiled kompiere libero in the contributions directory of the adempiere subversion. I then was also able to execute ./install/Compiere2/RUN_setup.sh. It failed on the database saying it could not login with compiere/compiere and that the database had not been imported. I then tried to run the RUN_ImportCompiere.sh script and that failed. I beleive it is failing because the myEnvironment.sh is not set right. the myEnvironment.sh is set by RUN_setup.sh so I feel like the chicken before the egg thing is occuring. I did unzip Compiere.jar in the data see directory of the kompiere libero data directory so I am close to importing/creating the oracle database.

Progress on 12/20/2006

I talked with Karsten a great deal and he helped me to figure out some problems I was having with installing/build libero. What is interesting to understand is that Karsten said I only had to run 2 scripts:

• 310 to 311 migration script
• 311 to 312 migration script

Running these scripts would import the libero database changes into an adempiere 312 oracle install. Karsten also told me that the migration scripts may have some problems as sqlplus can't handle very large lines for some reason. I finally got the kompiere libero server up. When I tried to login I got an error. Victor thinks I have the wrong version of Kompiere libero or Oracle. I have posted in the developer forum about this at:

http://sourceforge.net/forum/forum.php?thread_id=1635853&forum_id=610548

Not exactly - running these scripts will upgrade the libero database to an adempiere 312 preserving :all existing libero customizations. After this make a dump of the migrated database and this will be :and Adempiere+libero seed dmp. And not the scripts have the problems - it's a sqlplus problem ;)

Next step would be to adjust the scripts for postgreSQL, therefore you have to change the datatypes :from oracle to postgresql - there was a document in the wiki about that but I think it got lost in :the backup trouble -- Karsten Thiemann 11:00, 21 December 2006 (GMT+1)

I don't understand what I could have done wrong. Victor says he will handle the migration and it will be done by 12/27/2006.

I finally got in! I added connectionprofile to ad_user and ad_role and got in! Victor said that those fields are from 253d. Victor also said that I don't need to run RUN_ImportCompiere.sh just RUN_DBRestore.sh on the dump he gave me. He said he will verify his dump tomorrow to see if it is correct.

Progress on 1/5/2006

I made some huge progress on Friday. I was talking with Victor about how I could help him. He sent me an oracle database dump for Oracle 10.2 ( ExpDat.mfgupdate.dmp ). I setup oracle 10.1 previously so I just used that. Victor wanted me to assist him with renaming tables/fields for the specific tables dealing with manufacturing and payroll from libero so they could be more easily migrated to Adempiere. He posted in the adempiere developers forum about this at:

http://sourceforge.net/forum/forum.php?thread_id=1644403&forum_id=610548

To accomplish this he gave me two choices. I could use druid database manager tool or another way I can't remember (think it was modify the ddl raw in text form). I told him I would do it the way he preferred. He preferred the druid way. I then imported the specific tables with beginning mpc_ strings and converted them to pp_. I then saved my druid project and sent it to Victor. He did not like the format and asked me to use the format that Trifon used for his druid project at:

http://adempiere.svn.sourceforge.net/viewvc/*checkout*/adempiere/contributions/253b/comxe/_Project-ID-ComXe_Standard/druid/Project-ID-ComXe_Standard.druid

I ran out of time and told Victor I would continue on Monday.

Oracle Notes

  1. . oraenv
  2. lsnrctl stop
  3. lsnrctl start
  4. sqlplus /nolog
  5. conn / as sysdba
  6. shutdown
  7. startup

To import the dump I used the command:

imp userid=system/compiere file=ExpDat.mfgupdate.dmp full=yes

The import did say it had some problems and I was told that the import would most likely fail if I was trying to import a dump from 10.2 into 10.1. All in all though it appeared to work.

Druid Notes

http://sourceforge.net/projects/druid/

  1. java -jar druid-3.8-install.jar
  2. java -jar druid.jar
  3. Install oracle jdbc driver: ojdbc14.jar from within disk druid

Victor also asked to generate sql.

When importing some of the tables, some of the fields were undefined. I do not know why this occurred. I assume it could be because of the import malfunction(s) from 10.2 to 10.1.