Switching from cold to hot backup

[mpw]

Hello,

I used to do cold backups, as it was easier to configure. But the alfresco-data has serveral GB now and the down time has become quite long. So I want to switch to hot backups.

I'm a little scared to do something wrong. Ofc I read the Wiki:
https://wiki.alfresco.com/wiki/Backup_and_Restore#Hot_Backup_Procedure

- Do I have to create the lucene indexes before each backup, or does Alfresco do this periodically automatically? It's marked as optional, what are the disadvantages of not doing it?
- What kind of command can I use to hot backup a postgresql-database? Are there smarter ways than just dumping it?

Right now, I basically stop alfresco. And then I just save the data directory and dump the database:

sudo -u postgres pg_dump -Fc alfresco >> $date/$date-datenbank.dump 2>> $log
tar cjpf $date/$date-alf_data.tar.bz2 /var/lib/tomcat7/alf_data &>> $log
‍‍‍‍

Do I have to synchronize the backup of the database in any way with the alfresco backup? Or does the lucene index care about that?

Greetings from Germany
MPW

[mlagneaux]

Hi,

Lucene indexes backup is created every night by an Alfresco job. If you're hot backup doesn't include indexes backup, you'll have to fully reindex the Alfresco repository when restoring. That can be very long. If you restore indexes backup, at startup, Alfresco will only index transactions done between the index backup and your hot backup.
I always use a dump to backup my database and I don't know if there is a smarter way to do that.
Your database backup has to be done before the Alfresco filesytem backup. Alfresco has no problem with file of the filesystem having no reference in the database. But database records pointing to files that are not existing in the filesystem are problematic.

[mpw]

Thanks for your reply.

So all I have to do is:

- Assuring that Alfresco creates that index regularly
- Not stopping tomcat before a backup anymore

That sounds easier than that wiki article describes it.

Thank you!