Last week I announced the stable release of pg_chameleon as I felt quite confident the tool worked properly. However the murphy law is always ready to teach us we live in an interesting universe. A couple of days after the release I noticed on a server which data were modified seldom a strange replica issue. For some reason at specific moments of the day the inserts replayed on the postgres tables failed with the primary key violation.
In 2013 I started my pet project pg_chameleon to discharge my frustration only to find the entire thing even more frustrating. It’s fascinating how time and circumstances can fix the personal perspective revamping the passion for development that never left me. This philosophical introduction is to tell I still can’t believe I’ve relased the RC1 of my pet project pg_chameleon. The release is already available on pypi and includes several bug fixes and performance improvement.
After fixing few bugs I decided to release the first pg_chameleon beta. The package details with the changelog are available here http://www.pgdba.co.uk/p/mysql-to-postgresql-replica.html These are the notable changes. Switch to psycopg2 2.7.x - Thanks to Daniele’s work now the requirement installs without need for python or postgresql headers, making the install in virtualenv much simpler Install system wide or in virtualenv follow the same rules. The configuration files and the package files are installed in the python’s site-packages/pg_chameleon.