2011-01-10 04:49 PM
I have a lot of drives I need to sanitize for four filer decommissions, and was wondering how many disks I can include for the disk sanitize start <disk list> command. Also, is there a way to "queue up" all the disks I want sanitized so the process keeps going until they're all done (assuming only so many disks can be sanitized simultaneously)? I'd really like to automate the process as much as possible as it seems like this is going to be a long process.
Edit: Data ONTAP 188.8.131.52
2011-01-11 10:03 AM
I recall running a disk sanitize a year or more ago and only being able to do about 4 - 7 drives per shelf regardless of how many disk IDs I entered on the command line. It took me about 2 days to do a dozen or more shelves.
2011-01-11 03:24 PM
Disk Sanitization can be done in parallel. According to the documentation, maximum number would be ((# of disks - # of shelves) - # disks in root volume). There are two limitations:
1) Both SES (SCSI Enclosure Services) drives of a shelf cannot be sanitized at the same time. The command will allow one SES drive to be sanitized at a time. You have to make a second pass to get the second SES drive.
2) Only 5 unique commands may be sent at any one time. Each command can specify multiple drives. A unique command has a different set of sanitize patterns defined. That is the trick. Each of the 5 commands must specify a different sanitize pattern. If you try all 5 commands with the same pattern, it won't work. Don't ask me why.
I recall that my customer ran 5 unique commands, and specified 8 or10 disks per command. So they sanitized 40-50 drives per pass. However, they always required a second pass for the other SES drive in each shelf. In addition, they wanted to sanitize the disks used by vol0/aggr0. Therefore, they had to create a new aggr0/vol0 using sanitized disks, and then run sanitize commands on the old aggr0/vol0 disks. They couldn't ever figure out a way to completely automate this task, so it was very manual.
No matter what, this is going to be a long process.
2011-01-12 08:39 AM
Thanks reide. I trimmed the number of drives down per command and will start the marathon next week. Yeah, it will probably take a long time, but not as long if I had to experiment with the number of drives per command, so thanks for your help!!
Now if NetApp would just come up with a way to automate this process..