How it works GFS

[acrismatic]

Hi all,
We're new on the forum and have some questions about the backup jobs and about GFS configuration on the backup copy jobs.
First of all i try to explain you our initially configuration of the backup jobs:

We have an scenario where we are running daily backup jobs of several VM's. We do a monthly full backup and then we do daily incrementals with different restore points (7 or 30), depending on the VM.
As result we have normally 1 or 2 full backups for a month of a VM and several restore points depending the retention policy. Why do we have 1 or 2 full backup files?

At the end we need to keep 1 full backup of each VM for every month of the year (12 full backups). If i understand correctly on the guide, we can perform it using the Backup Copy Job and GFS. Right?
But i don't know whats is the correctly config for the restore points on this job, and what is the best way to schedule the job.
How it works the field "copy every" on the first screen ? If i only want 1 monthly backup for a VM, 30 day is correct to configure it?
In the third screen on the restore points to keep field. If i only want to keep 1 monthly full backup for a VM, what is the best config for it? 30?

And finally, this kind of jobs are running continuosly? If the answer is yes, How it affects to the bandwith and to the daily Backup Jobs?

Sorry if this questions were answered on another thread.

Thanks in advance.

[foggy]

Hello, yes, all of the questions were already answered multiple times in other threads, quick search could give you those for review. To save you some time, I'm giving some clues below:

Why do we have 1 or 2 full backup files?

This is how retention for forward incremental backup mode works. You cannot delete previous full until the required number of restore points after the latest full is created.

At the end we need to keep 1 full backup of each VM for every month of the year (12 full backups). If i understand correctly on the guide, we can perform it using the Backup Copy Job and GFS. Right?

Right.

How it works the field "copy every" on the first screen ? If i only want 1 monthly backup for a VM, 30 day is correct to configure it?
In the third screen on the restore points to keep field. If i only want to keep 1 monthly full backup for a VM, what is the best config for it? 30?

"Copy every" setting is the interval with which the backup copy job will create its incremental restore points. If you just need monthly backups to be kept as separate full restore points, you need to enable GFS retention and specify the required number of monthly fulls (12). You can set "Copy every" interval and the number of restore points to keep in regular incremental chain to whatever you want, for example 30 days and minimum of 2 restore points.

And finally, this kind of jobs are running continuosly? If the answer is yes, How it affects to the bandwith and to the daily Backup Jobs?

It runs continuously in terms of constantly monitoring restore points newly created by the regular backup job within the specified copy interval. Once the restore point is created, it copies all the blocks required to build next backup copy increment. In case source backup files are locked by the original backup job, backup copy job is suspended and continues once the files are available.

[acrismatic]

Hi Foggy,

I configured the backup copy job to try only with one VM with this settings:

Copy every: 30 days
Add: from backups
Restore points to keep: 2
Keep the following restore points for archival purposes:
- Monthly backup: 12, last day of the month
Data transfer: direct
Schedule: Any time

The job is running now. On the status the job is idle and indicates "No restore points found". That means until the last day of the current month the backup copy job doesn't make full backup, isn't it?
Why we have now two files on the destination folder? *.vbm (1KB) and *.vbk (30,1MB)
With this config, when takes the job the two restore points?

Thanks in advance

[PTide]

Hi,

Copy every: 30 days

By setting this value to 30 you made your Copy Job to check for backups to copy once per 30 days. If you want your backup copy job to pull new backups every day (if any) then you have to set it to "copy every 1 day" and set start time approximately 5 minutes after your backup job start.

That means until the last day of the current month the backup copy job doesn't make full backup, isn't it?

On the first run BCJ creates full backup (.vbk) and meta (.vbm). On each subsequent run BCJ checks for new restore points on source. If any present - changes are taken to destination. The rate how often changes are taken to backup copy location is determined by "copy every". "Restore points not found" means that no restore points to be taken found on source side. What's the size of your VM? 30MB .vbk file means that something went wrong. Could you please delete the backup chain from disk via VBR concole and rerun the job?

With this config, when takes the job the two restore points?

Not sure what you mean. Please clarify.

Thank you.

[acrismatic]

Hi,

I created again the BCJ with the same settings indicated on the latest post and we obtain same result, no restore points. What I'm doing wrong?

As I explained on the first post of the thread, we only want to keep 1 full backup for each VM and month. If i want to create this full backup now, what's the correct config for the BCJ then?

Ah!! The size of this VM is 40GB, of course something went wrong on the BCJ

[PTide]

1. Set "copy every 1 day" starting at Backup Job Start + 5 minutes.
2. Set job as a source
3. Set "restore points to keep" to 2
4. Set "keep monthly backup" to 12
5. After creation right-lick on a job and choose "synchronize now" - that should create full + meta on your target.

[foggy]

Prior to start copying actual data, backup copy job pre-allocates 32MB VBK file you're seeing in the repository. If there is an existing restore point created by the source backup job within the specified copy interval, it should start copying data immediately.

[acrismatic]

Ok, then the incremental backup job for this VM starts today at 21:15h. I'm forcing it to "start" now and runs correctly, creating a new *.vib file of 296.784KB. When this job stops, i went to the BCJ and click on "Sync now". As result, obtains a new *.vib file appears with the same size like the others, 30.844 KB.
Each time i click on Sync Now, happens the same, new *.vib went created with the same size like the *.vbk file.

[PTide]

Assume you have a Backup Job starting at 21:15 each day, Fulls once a month. It has just finished its run at 22:00.

A) BCJ is set to 30 days sync interval today. During first start it will consolidate all restore points for the period of last 30 days and create one .vbk containing the very last state of VM during those 30 days. After that BCJ sits there and waits for a new restore point on source for 30 more days. If any restore point appears then BCJ creates one more RP on target. All subsequent restore points will pulled by backup copy job only during next sync interval which.

B) BCJ is set to 1 day sync interval today. During first start it will consolidate all restore points for the period of last 24 hours and create one .vbk containing the very last state of VM during those 24 hours etc.

With that said, if your last backup is 20 hours old then your sync interval should be > than 20 hours to let your backup copy job grab something. If your last backup is 3 days old then your sync interval should be > 3 days. Please check this article. Please set up another backup copy job from scratch and set the interval in such way that it captures your last incremental. In case the job does not produce .vbk of a full size then please kindy open a case with support and post your case ID here.

Thank you.

[acrismatic]

Hi all,

I created, as Pavel suggested, a new BCJ following your recommendations. By mistake selecting the Backup Repository I selected other backup repository different than the one needed. Doing this, the BCJ starts and runs for a while but, few minutes later, it stops giving a warning.
After this, I deleted the BCJ and the files created on the backup repository in order to configure a new BCJ with the correct repository. Now the same error of yesterday is happening: the BCJ starts but states idle and doesn't copy any data.
The repository configured as a target destination for the BCJ is the same repository of the Backup jobs, could be this the reason of these errors?

[PTide]

Hi,

If so then that's an unexpected behaviour - please open a case with support and post case ID here.

Thank you.

[acrismatic]

Hi Pavel,

The case number is 01005838.

[foggy]

The repository configured as a target destination for the BCJ is the same repository of the Backup jobs, could be this the reason of these errors?

Definitely, it is. You cannot use the same repository as a target for both backup and backup copy jobs, there's a built-in mechanism preventing backup copy job from looking restore points in the repository that is specified as backup copy job target (not much sense in such configuration).

[acrismatic]

Ok,

Thanks Alex, I suspected it. I created a new partition on this volume exclusively to these jobs and they work correctly.
Now I close the support case also.

Thanks alot for your patience with the new users

[foggy]

No problem. This one is typical misconfiguration issue that has just slipped my mind until you mentioned it.