Best practice for upgrading drives on 2810SA?

Currently have a 2810SA with 4 250G hard drives in a Raid 5
configuration. I've realized that if I continue to put 250G drives up
to the maximum of 8, I will not be able to take advantage of full 2T
drive limit of XP. I would like to upgrade the drives to 300G
(possibly 400G).

What is the best way to accomplish this? Replace 250G drives first?
Add 400G drives then replace 250G drives? By reading other posts, if I
mix drive sizes, it will only see drives sizes as big as the smallest
in the array. Once I have migrated all of them to 300G drives, will it
start seeing all drives as 300G drives?

I already have data on the current array so rebuilding from scratch is
last option.

Thanks.

wrote in message
oups.com...
Currently have a 2810SA with 4 250G hard drives in a Raid 5
configuration. I've realized that if I continue to put 250G drives up
to the maximum of 8, I will not be able to take advantage of full 2T
drive limit of XP. I would like to upgrade the drives to 300G
(possibly 400G).

What is the best way to accomplish this? Replace 250G drives first?
Add 400G drives then replace 250G drives? By reading other posts, if I
mix drive sizes, it will only see drives sizes as big as the smallest
in the array. Once I have migrated all of them to 300G drives, will it
start seeing all drives as 300G drives?

I already have data on the current array so rebuilding from scratch is
last option.

Thanks.

Just as an aside, would it not be "better" to leave one as the hot spare?
ted

Just as an aside, would it not be "better" to leave one as the hot
spare?


Let me see if I understand raid 5. The raid 5 array uses the
equivalent of one drive's disk space in order to store parity
information. This parity information is used so that in the event one
drive dies, no information from that drive is lost. Although I don't
have a hot spare, I can just replace that drive and it would restore
the "missing" drive to its original order in the array?

If the above is true, I'm not in a critical enough environment that I
would need a hot spare. If the above is not true, please let me know.

On 25 Jan 2005 12:38:04 -0800, wrote:

Just as an aside, would it not be "better" to leave one as the hot
spare?

Let me see if I understand raid 5. The raid 5 array uses the
equivalent of one drive's disk space in order to store parity
information. This parity information is used so that in the event one
drive dies, no information from that drive is lost. Although I don't
have a hot spare, I can just replace that drive and it would restore
the "missing" drive to its original order in the array?

If the above is true, I'm not in a critical enough environment that I
would need a hot spare. If the above is not true, please let me know.

You are correct.

Hot spare and RAID are often rolled together, but in actuality hot
sparing is a service feature, while RAID is an availability feature.
Of course, the one plays into the other, in that following a failure
of a RAID member, the rebuild onto the spare means that the window
during which a second failure might result in data loss is much
reduced.

Against which, I have seen a scenario in which the hot-spare,
auto-rebuild feature *resulted* in data unavailability:

Following a RAID member failure, the rebuild to the hot spare started,
and during the reconstruction, obviously, all sectors on the remaining
RAID disks were read. Unfortunately, there was a bunch of bad sectors
on one of the RAID disks, which caused the control to fail the that
disk, too, and thus the array. No data was at compromised, since the
bad sectors were not in any files, luckily (they were not even in
parts of the disks that were being used -- I suspect because the
previous administrator had seen something like that problem, and move
the server using that section elsewhere).

Granted, that demonstrates a questionable RAID controller design (as a
very general rule, you should not drop a disk out of a RAID set for
READ problems, only for writes.

But it happened...

Malc.

On 24 Jan 2005 09:04:40 -0800, wrote:

Currently have a 2810SA with 4 250G hard drives in a Raid 5
configuration. I've realized that if I continue to put 250G drives up
to the maximum of 8, I will not be able to take advantage of full 2T
drive limit of XP. I would like to upgrade the drives to 300G
(possibly 400G).

What is the best way to accomplish this? Replace 250G drives first?
Add 400G drives then replace 250G drives? By reading other posts, if I
mix drive sizes, it will only see drives sizes as big as the smallest
in the array. Once I have migrated all of them to 300G drives, will it
start seeing all drives as 300G drives?

I already have data on the current array so rebuilding from scratch is
last option.

Thanks.

I have been staying away from Adaptec raid products for some time so
sorry if this is just a generic description:

You can replace the 250G drives one at a time with 400G ones. You
have to wait for the entire rebuild process on each to complete before
moving on to the next drive. When you are done you will have an array
of 400G drives that will be the same size as the 250G drive array. At
that point you should be able to use the "Online Capacity Expansion"
feature of the management software to expand the array so it uses all
the space. When that is complete you will probably have to enlarge
the partition(s) on the array to fill up the entire space using tools
specific to your choice of OS & file system.

Just make sure you stress test the new drives before adding them to
the array. Double check what I'm telling you with the manual to fill
in the specifics.