10, 9, 8, 7, 6, 5, 4.5, ...4.5, ...4.5

Well it's been a couple of weeks of using Bacularis on a separate machine, and its going well. I think I finally stabilised on about 4.5TB per backup, for the moment. Yet there are a couple of things I'd still like to figure out.

The backup is currently taking 2 tapes, which is acceptable, but I would prefer to try and get it all onto 1 tape if possible. The reasoning is that the tapes have a listed capacity of 2.5TB raw, and 6.25 compressed. With this in mind I attempted to configure the backup job to compress the data with LZO. Hopefully the next run will only take a single tape, or I'll realise that the 4.5TB I see is already compressed data, i.e. something like 8TB raw and compressed to 4.5TB, and 2 tapes. 

Another concern I have is, that in setting up the volumes for the tapes, I left them under the "default" type, which is configure to be a file. This lead Bacula(or Bacularis) to default them to a storage size of 50GB. While I can manually adjust them to what I think is the correct size, 2500gb, I'm not sure if I should be using that size or the compressed size of 6250gb. I'm also not sure how to tell Bacula to rescan the size of the volume. So I sit either wasting 2/3rd's of the tape, or just a couple hundred MBs, not knowing which is correct. I have attempted to reset the next tape to a size of 0B, which I'm hoping Bacula sees as "unknown" and scans the tape first, but the result of that test wont be until the next backup run on Sunday. 

The last thing that I need to debug is that currently when the job fires, it fires twice. I'm not sure why it fires twice, and I've set it up to cancel duplicate jobs, so the second job just get canceled, but I need to look into why it gets fired in the first place. I suspect that I have a duplicate service running somewhere that picks up the job at the same time as the "true" service, and creates a duplicate job. I'm not 100% sure its a race condition, but thats where I'm leaning for now until I look into it farther.