Hi,
NBU 7.5.0.6
RHEL6.current, x64
My config is as follows:
4-drive LTO4 jukebox (dell/adic) - works fine; not using for this scenario.
standalone SAS LTO6 drive (quantum)
2 4TB sata disks, each configured as a separate BasicDisk storage unit. Each BasicDisk storage unit has a job limit of 1. BasicDisk units configured as staging units, their staging schedules specify that the tape copy will be written to the LTO6 drive. Everything works, except that the job limit isn't being honored by the job that stages to tape.
I have a feeling that this is a basic design flaw - the only check for the job limit is on the target device, not the source device.
For context, here's the application. I have a 100TB video archive that we're storing on rented disk from our university's central IT. We've decided to stage it off to tape to save the non-trivial disk cost.
We bought the standalone LTO6 drive to facilitate this. My scarcest resource is the write bandwidth to that drive (especially considering that tape changes are manual, and ideally we'd change tapes twice during each work day). I thought I'd split up the jobs to write to the two sata disks, so that I could be streaming the tape drive at all times, and avoid reading the data over the network twice (making two copies of everything). When I get this job contention for the staging disk, the tape drive isn't going at full speed.
Anyone have any advice? (other than, "buy an LTO6 drive for your jukebox," which is what I wish I'd done, that would make this simpler).
TIA