SQuaLe Changelog

New in version 0.1.12

February 27th, 2009
  • This new release introduces the possibility to delay commits in the Oracle workers. This can be useful when dealing with massive inserts use cases (raw logging of web traffic to a database) where committing at every query would seriously damage the database performance. You can configure for each worker the "commmit-every" parameter which will indicate to commit after a certain amount of insert/update/delete queries.
  • WARNING: Configure this parameter to a value different from 0 will mean that you are doing insert/delete/update on some worker and you don't know when the commit will happen. This is definitely not for people who want to do transactions !

New in version 0.1.11 (September 26th, 2008)

  • This release implements a feature where workers are putting back jobs that they can't execute because of connection failure into the joblist. This way other workers which have a valid connection are able to process them. This is particularly useful if you use SQuaLe to load balance traffic across multiple database servers and want to handle transparent fail over when a server needs to go down. For now this works only on PostgreSQL and MySQL as Oracle has no such thing as mysql_ping and you need to actually do the query and analyze all the error codes to realize the connection was lost.