Automatic Cloning with two NDMP Groups (both blocking each other)

You can check with script if backup is finished and if drives are idle and if they are run it. Put that to run every hour or so and that's would be one way of doing it.


With automatic cloning it would get delayed too. Started cloning would wait for drive to become available and carry on when ready.

Usually environments are set to have backups over the night and cloning over the day. Not sure if that can be applied in your case.


In that case you still need two devices as before and two groups could overlap again. You would have 1 target instead of two, but still 2 sources.

Right.
But I suppose scripts can only run at a certain time via cron. The advantage of automatical cloning is that it would start right after the backup is finished and only if its finished.

In cases when backup gets delayed by missing media for instance, the cloning scripts might fail since the drives are still busy. An extra manual start would be required.

Thats why I prefer the automatic cloning.

What if I would only use one clone storage node. Like setting the clone storage node for both filers to explicit netapp2

client netapp1: clone storage node - "netapp2"
client netapp2: clone storage node - "netapp2"

Cheers

What you wish is easily controlled via script. That means that you would probably have cloning script where first you clone data from one filer and after from another one. With automatic cloning you won't be able to achieve that (unless time between groups is increaced in such manner that cloning which runs after is done by the time next group should kick in).

Well, one approach would to check if devices are available - if they are then check mediadb for saveset without clone (for specific clients or group unless you clone everything) and then clone those. That can be done with single line script.


I see that as major problem with automatic. I wish my backups not to be limited by the fact they can't be written to tape because drive is being used by cloning process. Also, with script you can sort first savesets according not spanning tapes thus achieving faster cloning results. Not sure about current NW versions, but that was not the case with automatic cloning before.

Ok. When I use different groups during the week and weekends I guess I also have to run different scripts according to the Networker schedule calendar to check what group has finished, do I?


Right, thats clear, but at least it starts right after and doesnt depend when the script is started next.


Yes, would work except for the full backup wich happens monthly. This would take then longer than 8 hours. But this happens on the weekend anyway.


Ok, makes sence. I thought there could be a logic inside networker to realize that only 2 tapes are available and therefore only one cloning session starts by another.

Ok, I think I have two options:

- Using cloning scripts
- Using one group so that it backs up one filer by another which will take twice as much since they not backup in parallel to the devices)

Just stagger the group start time and schedule to achive differnt finish time for the group and so for cloning, do you not have sufficient number of devices for cloning avallable to the server ?
Why not merge both filder or clusters in one group that way the auto cloning dont need four devices ( assuming two are reading, two are writing)

If it is not too late, a slightly different approach than those already suggested would be to do the following:
- use the savepnpc command for one of the groups
- use the post backup script to cause that group to pause after it finishes, such as by adding a sleep for an hour or two

This would allow time for Legato to unload the drive so the other group could use it to start its clone. And when the post backup 'pause' script for this group finally finishes, the group would wait for the two tape drives to become available and then run its automatic clone.

This would allow you to start the backups at the regular time and would allow automatic cloning (which I prefer to scripted cloning, but can't use in my environment because of the quantity of backup groups to be backed up during each backup window).

Let me know if you would like some more detail on setting up this idea.

Good Luck.

Thats it!
I will use the savepnpc command amd write a post backup script. This script will check for active running Groups. If there is only its own group in active running state the script will finish.

Thanks

We have exact same problems with cloning hanging with two NDMP client , one clustered Netapp Filer, when you are done with the script , can you post or email me if possible so i can implement it also.