Bob, Sorry for additional confusion.
> On Aug 11, 2009, at 11:27 AM, Dan Stangel wrote: >> One small clarification on the disk space: You said that we have >> additional space on the repo. We have unused space in the repository, >> but we do not have unallocated space in the SAN storage. In order to >> implement a full repo backup as you described, we would need to either >> shrink the allocated repo size by 1/2, or add twice as many disks as >> we currently have. > > I was talking about using the unused space in the mounted volumes (aka > the repo). Blabby rsync's it's 91 GB onto bashful > (/srv/fossology/repository/bashful/blabby), bashful rsync's it's 93 GB > onto blabby. Happy and sneezy do the same. Shifty and dopey do the > same. You end up with a duplicate repo in the storage we currently > have mounted. I see what you mean - I was thinking we would create separate dedicated virtual disks to serve the backup purpose. But we could certainly use the existing disks instead. > So this is all using the same SAN storage? What is the point of > the repo backup then? I'm guessing that we are using a raid level > to support a single or double drive failure. Is that the case? > What failure mode are you trying to protect from? Each of the agents has its own RAID1+0 (striped, then mirrored) virtual disk. These are housed in MSA1000s. Each virtual disk also has a hot spare disk. So the virtual disks can survive one of the striped halves of the mirror failing (loss of up to 3 disk mechanisms). Likewise the RAID provides redundant controller paths; if one of the MSA controllers fails, the other should take over its duties. Of course this RAID configuration does not protect against user error (e.g. overwriting or deleting a file from repository), filesystem corruption, or damage to the entire MSA, among other failure modes. Dan _______________________________________________ fossology mailing list [email protected] http://fossology.org/mailman/listinfo/fossology
