[SOLVED] Understanding prune options?

[wasteground]

Hi!

I am trying to figure out how the prune scheduling works a little better, since I am seeing some results I don't expect (and thus, I am not sure if this is a bug, or just my misunderstanding!).

My schedule is:
last: 1, hourly: 24, daily: 7, weekly: 4, monthly: 12, yearly: 1

My intention with this is:
- always keep at least one backup
- keep 24 hourly backups (if they exist) - for the previous 24 hours
- keep one daily backup for 7 days
- keep one weekly backup for 4 weeks
- keep one monthly backup for 1 year
- keep one annual backup

What I'm seeing is that basically everything is being kept currently for the last 3 weeks since backups began on 1st August, and it seems to be related to my "keep-hourly" setting. That seems to confuse things and cause more backups than I expect to be held. There isn't a lot in the documentation (yet) to explain exactly how this works with examples, but based on my schedule and my intention/understanding from the documentation, I don't think this should be the behaviour.

As an example, I tried to manually run this schedule against a VM which has backed up daily at 05:00 since 1st August, and every backup is kept using the above schedule, nothing is pruned. If I reduce the "keep-hourly" to 1, only backups since 13th August are kept, plus 1st August, and 9th August - which make perfect sense as the monthly and weekly backups I want to keep.

Every time I increment "keep-hourly" by 1, an additional backup day is kept (so setting it to 2, I get backups since 12th August, setting it to 3, I get backups since 11th August, etc).

I can just not keep hourly backups anyway, since right now I'm only doing daily backups it doesn't really matter, but I'd like to make sure I understand it correctly since in future I probably will want some backups running more frequently than once per day.

The only way I can make sense of this is that keep-hourly only makes sense to be set greater than 0 if you're actually doing backups on an hourly basis. Otherwise, it seems that it will just keep the last "keep-hourly" backups for all time, even if those are created on a daily schedule. Maybe my expectation is that it only operates on backups made in the past 24 hours, but this is not the behaviour intended?

Thanks!

 

[t.lamprecht]

Hi!

Did you already checked:
https://pbs.proxmox.com/docs/administration-guide.html#pruning-and-removing-backups

The, for your misunderstanding, relevant part should be:

Each option only covers backups within its time period. The next option does not take care of already covered backups. It will only consider older backups.

So, if keep-last touches some backup snapshots, the keep-hourly option will not even check those and start only after that. I.e., those options are to be seen as accumulative.

In addition, keep hourly does not goes over X hours in the past and marks the latest one in such an hour intervall, but rather it goes over all snapshots, and marks the latest in such an hour interval - if there's no backup in one hour-interval it does not count this, again, it is accumulative.

So setting only keep-last to 1 is the same as setting only "keep-hour"ly to 1 (or any other setting, if really just a single one is set).
Setting keep-last AND keep-hourly to one will always keep two backups of a group.

Hope that helps a bit.

 

[wasteground]

Ah! Yep, that does indeed help. I read the docs, but I did not interpret that particular piece correctly, makes much more sense (and I guess that means my assumption on my misunderstanding was correct). Thanks for the clarification!