Automatic grooming does not remove enough data

[jacp10]

I have a disk backupset which contains backups of multiple servers/clients. The grooming policy is set to Retrospects defined policy. If the nightly scheduled backup script runs out of room on the backup disk then grooming happens automatically. However this does not always groom out enough space for the backup to complete. If the backup runs out of room after grooming, then rather than groom again, it asks for new media.

 

Is there anyway to indicate to the grooming system how much space it is likely to need to free up in order to complete the backup? It does not seem to calculate this itself (and would have to prescan every client speficied in order to do so). I was wondering if there was some useful interaction with the 'use at most' setting on the backup set, or whether this just effectively results in retrospect viewing the disk as being smaller than it really is, in which case it makes no difference.

[nekr0phage]

Hi,

 

It would be helpful to know which version of Retrospect you are using.

 

If grooming to policy does not allow enough free space on your backup volume, then I personally would recommend using the other grooming option, "Keep only the last X backups." It does not require as much space as the defined policy.

[jacp10]

I'm using 7.5.234 with hotfixes 7.5.8.101.

 

I back nightly so "last n backups" is very different from retrospects defined policy in terms of the number of snapshots that would need to be retained in order to have a backup snapshot from one month ago still on the system. Defined policy is exactly what I want so it would be a pity if I couldn't use it as a resullt of problems with automatic grooming.

 

justin

[nekr0phage]

Justin,

 

So what are the data sizes you're working with. How much space does a nightly incremental take up on average, and how much space do you have available for storage?

[jacp10]

Hi,

 

My main backupset is a drive with 350GB usable capacity. A typical nightly incremental takes up about 15-20GB. A full backup takes about 200GB. I''m currently adding some more space to the backup set as its not big enough to keep the number of snapshots I want, ideally I'd like to follow the defined policy and go back to about 3 months. i.e. daily snapshots for 1 week prior, weekly snapshots for 1 month prior and then a couple of single monthly snapshots.

[nekr0phage]

Okay, so lets do the math here. We'll use your higher end numbers for a worst-case estimate. Your initial backup produces 200GB of data, with 20GB added to that every day. After one week you're already at 300GB for that backup set. Add the minimum of four additional snapshots for each week of the past month and that's another 80GB of data which puts you at 30GB over the advertised (not actual) space of your backup drive.

 

You're trying to put too many eggs in your basket. You need a larger backup drive if you intend to use the defined policy.

[jacp10]

I am putting in more storage. However my point is that to go back to 3 months I need a lot less space if I use the defined policy than if I use the 'keep last n backups' policy. But automatic grooming with the defined policy needs to correctly assess the amount of data it needs to groom out, or be told this by the script, and that doesn't seem possible.

[nekr0phage]

Quote:

---

But automatic grooming with the defined policy needs to correctly assess the amount of data it needs to groom out, or be told this by the script, and that doesn't seem possible.

 

---

 

You're right, it is not possible. According to the description of the defined policy it should "At a minimum, for each source, retain the last backup performed each day for two most recent days." When data is groomed out beyond the normal defined policy (days, weeks, months), the oldest data will be the first to go. In this case, you still will not be getting the 3 month history you desire until you A) increase storage space, or decrease data stored.