Scheduling GFS monthly for tape of copy jobs

[pkelly_sts]

[Edit: Case #02081448]

I have 2 sites, each site being largely similar in setup in they they both have a tape library & plenty of B2D capacity so I have what I think is a good strategy to copy each sites backup copy to the local tape library as well so I have Local backup, local copy to tape, copy to remote site, remote site copy to tape, e.g. the backup exists 4 times, 2 sites, 2 different media.

This works reasonably well except for GFS backups as I'd like to also keep a GFS monthly/yearly set on each site as well, this means the tapes on each site can simply stay in the library & auto-rotate & I know I'm still protected from disk/array failure on each site.

However, it seems to me that GFS tape logic really doesn't fit this model and I've never been happy with what I end up with (I've just opened another case related to /last/ months backup ,27-01-17, being written to todays GFS tape, I'm just waiting for the logs to upload before I get the case number which I'll post here as a comment).

I think part of the problem is that I've come in this morning to see a GFS tape job running & processing already, which is I assumed was copying *last nights* overnight copy data, but is actually doing the above job from last month.

Anyhow, what is expected behaviour here if it wasn't doing last months? If I'd copied a job from last night, which completed copying to local storage after midnight last night (so after the GFS job starts waiting) then would you expect /that/ data (e.g. Thursday's backup) to be written to tape?

I really really ONLY wnat Friday data written to GFS tapes, we already know there's no way of stopping GFS jobs starting up a midnight because that's the design (which I don't like much TBH) so it almost sounds like I want/need the schedule to be "The next saturday after the last Friday of the month" if I wanted to guarantee to ONLY capture the results of the Friday job?

The principle of GFS tape jobs is great, they just seem such a major headache in practice & Iv'e been banging my head over it for months now trying to understand the logic (even having read the Veeam documentation countless times for whats *supposed* to happen!).

Is it just me?

[sg_sc]

You could always disable the schedule for the tape job and trigger the job manually with post-backup script on the job that makes the new full.

[pkelly_sts]

I'd forgotten about this post. It actually turned out in my case to be an issue which has since been resolved with support (though i can't remember the specifics now as this was a while ago). I'm not convinced that disabling a GFS tape job is a good idea though, I can't remember the details but I do remember something about a new session being started (or the existing session being closed) whenever you "disable" a tape job so it's something I avoid as much as possible.

To be fair things seem to have settled down a bit in my GFS scenario from this perspective, it's frustrating to troubleshoot due to only happening once-monthly!