Some questions. . .
1. I have a DS14 shelf with 36gb disks, and FC8 shelf's with 18gb disks. Each shelf has a HotSpare. If a 18gb disk fails will it for sure use the 18gb hotspare or is there a chance it will pickup the 36gb hotspare?
2. I'm thinking about migrating all the data on the 36gb drives (including /etc folder) to fewer 72gb drives. What's the best way to do that? Can I add the new 72gb disks to the volume with 36gb drives, then start to remove 36gb drives over time?
3. Right now I have the FC connection going like this: Filer Head --> FC8 shelf --> FC8 shelf --> DS14 shelf.
All of the important data is on the DS14 shelf. The FC8 shelfs aren't on support contracts, and don't have any important data at all. (mostly system images, software install files, etc). Is this a bad idea. How much risk am I putting my important data on the DS14 shelf. We're thinking about getting another FC HBA with two ports because right now we just have a single FC cable. Then I would plan on putting DS14 shelf on the Dual Port HBA (with redundant paths) and putting the FC8 shelfs on our current HBA by itself.
4. With the sysstats tool, what kind of numbers can I expect to max out on, on the Disk KB/s?
Thanks, Steve Evans SDSU Foundation
Some questions. . .
- I have a DS14 shelf with 36gb disks, and FC8 shelf's with 18gb
disks. Each shelf has a HotSpare. If a 18gb disk fails will it for sure use the 18gb hotspare or is there a chance it will pickup the 36gb hotspare?
It will use the 18g hot spare.
- I'm thinking about migrating all the data on the 36gb drives
(including /etc folder) to fewer 72gb drives. What's the best way to do that? Can I add the new 72gb disks to the volume with 36gb drives, then start to remove 36gb drives over time?
No. You need to build a new volume and copy. This need not cause as much downtime as you think because you can copy over a snapshot with ndmpcopy while leaving the filer up. When that is done, make another snapshot and do an incremental copy to pick up the changes that happened during the big copy. Then kick everyone off the filer, take a final snapshot, and do a final incremental copy to pick up the last changes. Your downtime should be pretty minimal.
Steve Losen scl@virginia.edu phone: 434-924-0640
University of Virginia ITC Unix Support