Re: Does solaris require disks to be zeroed before using

From: David A.Lethe (davidATsantools.com)
Date: 09/23/04 

Date: Wed, 22 Sep 2004 22:37:42 -0500

On Thu, 23 Sep 2004 03:31:33 GMT, CJT <abujlehc@prodigy.net> wrote:

>David A.Lethe wrote:
>> This is causing some issues at our company, and we can't get any
>> definitive answers. Is there any requirement, or even best-practices
>> policies that SCSI and/or fibre channel disk drives need to have all
>> zeros written to them before partitioning and building a file system
>> on them?
>>
>> The problem is that our disk exercising/test suites finish by leaving
>> the patterns on the disks. The more experienced sysadmins feel it is
>> a good idea to zero all blocks on the disk before using them. Others
>> feel as long as the pattern doesn't look like a partition that could
>> confuse fdisk, then repartitioning and building the filesystem is
>> enough. (Remember, disks just passed extended read/write tests, so no
>> benefit to the disk in terms of defect detection and bad sector
>> remapping. Also the disks could be physical drives or LUNs on a RAID)
>>
>> Does anybody know if there will be issues down the road if we don't
>> write zeros to all blocks as final part of disk test before we start
>> using them?
>>
>> I'm posting in Solaris group, but would be interested in knowing if
>> any O/S's really cared one way or another.
>>
>> David
>>
>FWIW, I've moved used disks from other brands of system (e.g. Dell,
>HP) to Suns without ever worrying about what was on them, and have
>never experienced a problem as a result.
>
>The only problem I can imagine is if a disk were low level formatted
>with an odd sector size (as are some IBM disks), it might need to be
>low level formatted again with a suitable sector size.
The disks would be low-level formatted with correct mode page
settings, just that the data on the disk would be pseudo random
sequences.

We did come up with scenario where a disk that had zeros for unused
data would be "easier" for fsck to recover if/when the filesystem was
shutdown improperly. If the patterns looked like orphaned inodes
then,well, that would be bad.

Quantcast