Jean: good to hear. Lucene is a timebomb ticking under Alfresco. Current ECM repositories will grow, exhibit frequent delete and creations of content, and encounter index inconsistencies, leading to downtimes of many days. I lived one this week. It’s easily replicated: just sync with a larger LDAP directory (with a few 1000 users and groups): Alfresco creates and deletes the whole bunch at every sync pass, leading to an rapid increases in useless transactions in alf_transactions.

Alfresco must asap get rid of a full-text indexing engine to index the structured data, and use lucene for what it is meant (full text indexing). indexes are in the database. Should be used. The current trade-off of flexibility in the datamodel and database portability versus structured query performance and reliability is a basic design flaw. Alfresco is allowed to make a few, as so many wonderful choices have been made. But not repairing it is exposing partners and customers to a major risk.

To finish off: I believe a system is scalable to the extent it can be repaired off-business hours, after an incident in production, to be operational again the next morning without data-loss. For Alfresco that means: a few million documents, with not too much meta-data. Given that the -auto option of re-indexing works.

Not sure if you are an Enterprise customer or not, but in 3.1, as an Enterprise Only Feature, a new UI (set of web scripts) was added to do much more sophisticated index management, including partial re-indexing, diagnosis, etc.

It’s not something we recommend you use without guidance from support, but just wanted to make sure you know that we have in fact been working on such tools, in addition to a larger refactoring effort that Jeff mentioned.