Mal: Thanks for the offer! Actually most of the space (66GB of the 180 used) is actually /home/backups -- which includes nightly full dumps of the ASM database - about 1GB a night. Another large portion of the data (44GB) is /home/confluence/data/backups. Each daily confluence backup is about 500MB. Each month or so I usually go into /home/backups, move one file from each previous month (so June 1, July 1, etc) into the "monthly" directory and delete all but the current months' backups. Then I do the same in /home/confluence/data/backups (mv ./*_01.zip monthly/) and delete the rest except for the current month. In this manner I can always revert a screwed up DB to the last few days/weeks, or at least to the last month since 2005. Each night the server mounts an identical drive to the one shown in the system and does a full rsync from the production drive to the backup drive, then unmounts it. Each 1st of the month this process includes deleting on the backup drive any file which was deleted on the main drive. So, in this manner we always have nightly saves of all our DB data covering one month, plus monthly saves covering X years -- and replicated to a spare disk in case the main disk fails, gets erased, etc. I consider the above described manner (nightly rsync to an unmounted disk) better than a RAID 1 for our purposes, because we have only 2 disk slots on the server -- and if they're both mounted to a RAID 1 then a controller failure could corrupt both disks simultaneously, leaving us with no backups to the data. Check out the server now, after I've gone through all the processes above (took about 2 mins) - we're at 52% used with 110GB avail. Mal Raff wrote:
At the moment it looks like there's almost 50GB out of almost 250GB left. I'm happy to put in some money to purchase another disk to add to that... large disks are remarkably cheap these days. Might need it if people start using the site to store images, rather than public access commercial sites.
Also wondering about the value of having at least RAID 1 (mirror) to make sure the site essentially never 'goes down' because of the very infrequent disk failures.... as well as having an image backup of the system.
Just thinking ahead out loud... you can certainly choose to ignore ;-)
-m-