STOPPED Job While Migrating Non-CDB to PDB via Autoupgrade

Hi,

Lately, I started to learn about autoupgrade which is the new upgrade tool for almost anything. Many thanks to Mike Dietrich for amazing blog posts ( https://mikedietrichde.com ). His blog is a real treasure.

I am about to migrate some non-cdbs to pdbs and before doing that I wanted to learn my options about how I can achieve that. I will write a blog post about three ways of migrating non-cdb to pdb soon. one of them, of course, autoupgrade and I almost trying to do it for last 10 days! All my efforts has ended with a STOPPED autoupgrade job at “compatibility check” step and it drove me crazy. You must know that for 10 days I tries tens of different combinations to achieve that and every each of them failed! Finally (just a mistake) I realized what happened. Let me explain:

I don’t want to go into details about autoupgrade non-cdb to pdb operation in this post, so I will just share my config file and command to run autoupgrade.

Here is the test case:

Oracle Database EE 19.14 on Oracle Linux 8 (virtual machine on Windows 11 host).

Latest autopgrade.jar is downloaded from Oracle Support.

Oracle Home is the same for both cdb (sid: cdb) and non-cdb (sid: orcl) databases.

here is my autoupgrade config file:

then run the autoupgrade to migrade orcl db into a CDB as a pdb:

I have read so many documents and blog posts about it and every time I run this command, it started Command Line Interface (CLI) of autoupgrade and when I query my job with lsj command, I got this result:

This is instant. I type lsj immediately when I reach to cli and I saw the upgrade job is STOPPED (status is running but operation is stopped and job is not working). So I check the log files of course and here is the error:

so an ArrayIndexOutOfBoundsException is thrown. Also, in noncdb_to_pdb_orcl.log file I got this:

This makes me think that there might be a compatibility problem or a violation but actually there were none! I spend days to understand this 🙂

After tens of autoupgrade tests, reading hundereds of log file results, with some luck and coincidence (and my hard work of course)  something got my attention. in log files I saw same familiar codes of mine! Where did I use those codes and why are they in autoupgrade logs? They are my “glogin.sql” file codes. for those who are not familiar to glogin.sql it is a initial script file for sqlplus tool in $ORACLE_HOME/sqlplus/admin directory. I use glogin.sql to set some default values like assigning db name in prompt or set SID as terminal header so I can see my current sqlplus SID all the time etc. but non of them is causing this but the guilty one is:

yes! I got “set timing on” in my glogin.sql file so I always get execution time of my sql scripts and somehow autoupgrade does not run a “set timing off” command and this cause some unexpected results. as you know, set timing on adds execution time to end of of your commands outout. probably, some scripts are assume it as a result of a query.  I remove it and re-run the autoupgrade tool. everything work smooth and perfect. database is migrated as a pdb.

So,  I spent 10 days because of “set timing on” but at least I learned a lot and solved a problem of mine, I am proud of it 🙂 I hope this helps whoever uses glogin.sql file and set timing on in it.

Wish you healthy days.

Peace at the Home, Peace in the World (Mustafa Kemal ATATÜRK) .