Backup Copy job using copy full retention

[skrause]

So I just wanted to make sure I was not off base in my thinking.

We run a backup copy job on a very large file server with an extremely high change rate (it is where our SQL backups live) once a week to a different location where we keep 5 restore points. Since there is such a high change rate, our weekly copy of changed blocks is about 60-70% of the total VM size every week which makes the merge take quite a long time.

We just purchased additional backup storage at the secondary location and I was planning on making some changes to how I set up the copy jobs to make things function more efficiently and was planning on just having it run a copy full every week and my thought is that if I use a copy full every week, then the job will just delete the oldest restore point instead of doing a merge.

Is this a correct assumption?

[foggy]

In case of active full backup copies it will act as a forward incremental backup job, deleting the old points once the new chain reaches the required retention.

[skrause]

So looking at the document you linked, plus trying to configure the job, there is no way to do just once a week do a full copy without running some kind of daily job with a retention chain?

Having to run a daily copy of several hundred GB just to be able to change the retention to not use transforms is kind of annoying.

[foggy]

You can select the "Read from source" option in GFS settings, set the copy interval to 7 days, enable weeklies and in this case the chain will consist of fulls only, regardless of the regular retention setting.

[skrause]

Ok cool. I will try that.

[WRS2200]

Thank you for the post. I am in a similar position. I will attempt this as well.

[skrause]

One more question, in this scenario, will the most recent point still be locked until the next copy interval or can it be used immediately after completion as a source for a tape job?

[JaxIsland7575]

Once the backup copy job finishes, you can have a scheduled tape job run off the same source or destination. The Backup Copy job runs continuous but you can still use the data while the copy job is idle.

[skrause]

The issue is which restore point it uses. With regular backup copy jobs, the most recent restore point is locked until the next copy interval. So in cases of a 1 week interval, the tape job uses the restore point from the week previous.

If I use GFS in this manner, I am hoping I can use the most recent restore point immediately to make scheduling a GFS tape job to run automatically every month easier.

[JaxIsland7575]

I am not familiar with the restore point being locked. I have a backup copy job that is set to run on Saturday evening, then Sunday morning it takes the data from the Saturday evening backup copy job and runs a tape job against it. This captures the most recent data. Maybe foggy can clarify but I was not aware of the file lock on a backup copy job.

[skrause]

So I just set up a test tape job using my fresh GFS backup copy job with a 7 day interval as a source and I got the following message:

8/15/2016 9:49:28 AM :: Job started at 8/15/2016 9:49:27 AM
8/15/2016 9:49:29 AM :: Processing tasks
8/15/2016 9:49:31 AM :: No new files to backup per job settings
8/15/2016 9:49:31 AM :: Job finished at 8/15/2016 9:49:31 AM

There is 1 restore point on disk currently (it ran yesterday). So it looks like even when it is in "copy full restore point"/GFS mode it still holds the last restore point hostage until the next copy interval starts.

[JaxIsland7575]

I may be a bit confused when you say that the last restore point is held hostage. The job finished so now you should be able to run a tape job against that backup file that was copied prior.

[foggy]

The behavior is expected, tape job waits for the interval to expire, to guarantee that there will be no more writes into the latest restore point. Here's a bit more on that.

[skrause]

You mention in a post in the thread you linked that you guys would be investigating a way to change the behavior of the flag tape jobs use in the future. Any progress on that?

For SLA purposes I need to have a tape job that uses the restore point at the end of the month. Since my tape library is connected directly to the repository server where my backup copys are living, I would prefer to be able to use the backup copy as a source for a tape job that runs on a specific day every month than to have to pull all that data over the wire from my primary datacenter again as it will be much faster when it does not need to go over my throttled connection.

[foggy]

You mention in a post in the thread you linked that you guys would be investigating a way to change the behavior of the flag tape jobs use in the future. Any progress on that?

Currently the behavior cannot be changed.