I've been reading through the backup & restore documentation here http://wiki.alfresco.com/wiki/Backup_and_Restore and it makes reference to lucene indexes. I can't seem to find these or the backup-lucene-indexes directory referred to in my Alfresco Community 4 installation.
Ah cool. So by that rationale then, I don't need to do anything special to 'sync' the database with the indexes. I can just run pg_dump to dump the database, backup the contentstore and I'm done then?
Would you stop the tomcat service too to prevent inconsistencies (i.e somebody uploading a document while the database backup is in progress)?
Does the SOLR indexing happen automatically, and where are those indexes? - I'm trying to put together my own documentation here 😉
The database backup needs to be a consistent snapshot so unless you have a tool that can do that then the easiest way is to shutdown. The good news is that pg_dump seems to be such a tool.
The lucene/solr indexes are optional since they are derived data but may speed up a restore. In particular versions of alfresco prior to 4 would need to complete the lucene re-index before restarting which can take ages on large repositories - hence the optional step to back up and restore the lucene indexes onto the new install.
Alfresco 4 is a lot less reliant upon the solr indexes being up to date and will eventually catch up, but I'm sure there are solr data that could be usefully backed up and restored. Of course if you are also using lucene on Alfresco 4 then the old considerations will also apply.
Our backup process is a bit lethargic however and we occasionally have users connect at strange hours. I thought it might be best to shutdown tomcat before running the backup just to make doubly sure that noone changes anything during the database backup. I'd assume if this happened, the database and the actual files loaded on the system might be inconsistent in a potential restore situation.