Topic: backup to tape - how long should this take with IDE tape dri

[Edward Kuhn]

I am running an e-smith server with a 20 gig hard drive and a Seagate STT220000A 20 gig IDE tape drive. The backup is set to start at 1:20 am. According to the Backup/Restore data page on the admin interface a snapshot backup to my desktop is 'no more than 15 GB' the hard drive has 18 Gb of data.
I generally do not swap tapes on this server until sometime in the afternoon, end of day. I am seeing from the logs that the tape backup is ending at about 10 am. People are generally working with files starting between 8:30 and 9 am.

First question, is this a reasonably expected length of time for this backup to take? If so, would I see a shorter time if I were to add a scsi interface and a SCSI DDS3 DAT drive to this server?

Second question (although I think I know the answer), should I expect that files that may be in use (read and written to) between 8:30am and the time the backup is complete will either not be backed up or will make the running backup inconsistant.

My situation is that someone accidentally overwrote a particular file on the server and I have been trying to restore this single file from the backup tape. I can do this in a test situation with a newly installed SME5 server however this takes quite some time to complete.

I am looking to one or both of these solutions: add a scsi card and use an external DDS3 drive I have on hand for a faster backup and two, add a second 20 gig hard drive to the server and by using dd, duplicate the primary drive to this second drive each night to provide a quick and easy way to restore an accidentally overwritten/deleted file.

Any comments/other suggestions are welcome.

[Darrell May]

You mention you have 18GB of data, a 20GB HDD and a 20GB TBU (actually 10GB native).  You are at capacity both on your hard drive and tape drive.

Time for some 'pruning' or a hardware upgrade.

Regards,

[Edward Kuhn]

I agree - it's been a chore to keep it within capacity. Actually the server has a separate 4 gig drive with system and an added 20 gig mounted on /home (with about 6 gig available). I am planning to update to the newest version server and am setting up a separate box (single 20 gig HD) so that I can configure and add users, ibays, etc over a one week period. I plan to move some files (management/finance/etc) to this server and leave the second (original server) for projects, work files, etc.

I did manage to recover the file that was overwritten but it took just about an entire day. I'll be adding a second 20 gig drive to the new server and duplicating this disk each evening so there is a quicker way to recover files. Tape is still needed but it is more for complete recovery and off-site storage.

One thing I see is that a directory I set up as a temporary drop box for users who need to share files has collected many gigs of files in directories created for special projects. On the new server I plan to set a cron job that wipes this directory each night. If a project comes that requires shared files I can set up a new ibay for it on request and remove it to cdr when the project is complete (usually 3 days to a week)

I'd also be interested to know what others do in this situation to control usage (other than purchase newer, larger drives and backup)  I'd like this option but for a larger, heavier use server in the future I've been considering some other options that will provide greater redundancy/recoverability albeit at a greater overall cost. I just am not in position to do this yet.