GFS Retention Policy

[wolflord]

Hi All,

I've been trying to set up GFS backup copy jobs for some of my VMs. The idea is to have a weekly GFS retention point that occurs every Sunday.

After letting some time go by, I notice that all my 'Weekly' retention points are showing the date for every Monday as opposed to every Sunday.

Anyone know what I'm doing wrong? How can I get the weekly GFS jobs to run on a Sunday? Why are they running on Mondays?

Example1:




This is pretty much the same situation for all the GFS jobs..

[v.Eremin]

Couldn't it happen that someone changed a weekly schedule to Monday at some point of time and reverted it back recently? Honestly, I cannot think of any other reason of the experienced behaviour.

Anyway, might be a good idea to investigate it further with support team assistance.

Thanks.

[ds2]

Just a idea. Colud it be caused by the sheduling times. Daily 3am and weekly at 11 am? And how long is the window for the job?

As far as i know weekly, monthly, quaterly and yearliy Restorepoints always starts at 0 am. You create restorepoints at 3am on so it is preformed in the next run on monday.

[foggy]

Backup copy creates backup file right when the interval starts, so each restore point has the date corresponding to the interval start time and date when it was created. Since, in backup copy GFS mechanism, the regular full backup becomes GFS restore point once it reaches the day specified in GFS settings (Sunday 11am in your case), it is always the restore point created on previous Monday.

[ChrisGundryCEGA]

Hi There,

I have a daily job that runs every day and backs up to Repo1 and retains 7 restore points.
I then have a GFS job that is setup as a secondary target for the above job and stores its backups in Repo2. This job is configured to 'copy every 7 day' and retain 2 restore points, 5 weekly points, 5 monthly points. The scheudle is configured so the weekly backup is Friday at 23:00, monthly is the last day of the month.

I have just gone to restore a month end backup for something and instead of it being on the 31/03/2017 it is on the 25/03/2017. This has WM next to it so is being used as a weekly and monthly. As far as I am concerned I should have the 25th as a W and then the 31st as a M on top of the 25th...? I have checked a few others and I can see the same issue, the monthly backup that is being retained isn't the last day of the month. The 30/04/2017 should be the end of April but the backups that have been retained are 28-29/04/2017 not 30th. 24-25/02/2017 has been retained as the February month end instead of the 28/02/2017.

Can anyone advise why this might be happening? Is it because of the 'copy every 7 day' cycle it is currently on? Means that it can only use the Friday point (when the 7 day cycle is set to cycle through) to create GFS points from? If so does that mean we should change it so it runs daily and on most days it will copy the backups but won't create a GFS point, unless it is a Friday or a month end day?

Many thanks!

[foggy]

Hi Chris, GFS restore point is created once the full backup in the regular backup copy chain reaches the day specified in GFS settings (in your case, last day of the month for the monthly restore point).

[ChrisGundryCEGA]

Hi Foggy,

Sorry, I am not sure I am quite following that last bit? Are you able to explain it differently?
Are you saying that the CopyJob is getting to the 31st and because it only has 2 points, plus the weekly points it is using the weekly point as it is closer to the 31st than the latest '2' points that it has?

When we set this up we discussed with Veeam support and we were advised that this way of doing it would work, so that is why we continued with it. But obviously it isn't giving us monthly backups that were from the day we told it to use, which isn't what we need.

I need the month end backup to actually be at the month end, not on a random day up to a week away from it. So what should I do to achieve this? Should I leave it set at 2 restore points, but set it to do a daily copy? Then when the 31st comes around it will have copied the source backup from last night copied over and it will use that? Or should I do something else?

I am just thinking that by doing that it will be copying data every day, but only using it on a Friday for the weekly and on the last day of the month for the monthly. But on any other day it has copied the data and won't be using it for creation of a GFS point. And by doing it every day it will be doing a merge process every day on the 'normal' restore points...? This repo is stored on ReFS so that is less of an issue but still...

Thanks

Chris

[foggy]

Please disregard the last sentence in my post above, it was a momentary lapse of reason. I've removed it to avoid confusion of future readers. Indeed, the monthly full should be put aside once the full backup in the regular chain reaches the last day of the month and should have the date either of that day (in case it falls on the day when the job cycle starts) or the day the next job cycle starts. I'd recommend to ask support to review the setup once again.

[ChrisGundryCEGA]

I posted this topic because I wanted to get some feedback and suggestions from people that do this. I have had this post merged into another topic now, where no one will find it. I don't really want to read through 200 posts to find any possible relevant information in the topic. I have read several pages and not found anything that helps me at the moment... Very annoying.

Are saying to set it to run daily and also set it to retain 31 restore points? Can you please write out what you are suggesting? Maybe with an example?

If you are suggesting the above then I don't really want to do that as it will retain 31 restore points, copying one each day, taking up network traffic, storage space and also do a daily merge and shuffle process taking up IO. I just want it to run on the Friday and take a copy of the source backup from that day. Run on the month end and take a backup on that day. Would it change things if I set it to create the backups using "read entire backup from source backup"? Would that make it run on the required day and only copy data on that day? I really don't want to have it run and copy data every day, store that for a month just to get 6 restore points (5 weekly and a monthly), then repeat that forever...

I tend to avoid raising support requests for things like this because I find the support team don't tend to be as good at understanding what we are trying to achieve when it comes to things like copy jobs. I guess that is why the original setup was suggested to us and as it now turns out, it doesn't work. I was hoping for some clarification on the forum instead. Unfortunately at the moment I am still just as confused as before.

[foggy]

The only thing I've suggested is to verify the job configuration with the help of support, since what you're seeing doesn't look expected.