disk "shrunk" after going into a 2850

[Brett Dikeman]

On Dec 19, 2007 5:45 PM, RB <aoz.syn at gmail.com> wrote:

> This is typical of RAID environments, and expected behavior.  If a
> drive is physically removed from an array and later replaced by
> another without removing, resizing, or otherwise modifying the array
> to make it unaware of that channel/slot, it is expected to treat
> subsequent drives as replacements and automatically perform the
> necessary sizing/copy/parity operations.

Yes, if there is a degraded (rebuildable) array.  If such is the case
and you insert a drive with no RAID metadata, I would expect it to be
claimed and a rebuild to start- although some controllers [optionally]
require you to tell it to use the drive, often by explicitly declaring
the drive a spare.

  In this case, there was no degraded array- there was a failed one.
It was a single JBOD (or Dell/LSI's approximation of it, single-drive
RAID0) with that drive completely gone.  Even if it was a multi-drive
RAID0 array and an incorrect drive was placed in the slot, I'd expect
the controller to do nothing- simply report the array as completely
failed, and not touch it.  I wouldn't expect it to initialize the disk
with controller metadata.

Then again, when I pulled a drive (also in a one-drive array) and then
replaced it a minute later- no amount of cajoling (reboots, power
cycles, trying the "rebuild" command and so on) could get the drive to
come out of "failed" status.  Eventually I just deleted the logical
drive and re-created it.

> If you did not purchase your
> system new and did not receive that documentation, it was rather
> unwise to leap headlong into disaster testing with data you couldn't
> afford to lose.

The system was purchased new.  The documentation was placed somewhere-
by a predecessor.

It was data both verified to be on tape and rsync'd earlier to its
replacement, which had undergone two days of burn-in testing.  We
moved the old drive into another machine to mount read-only to make
another consistent copy for software testing purposes (not "disaster
testing"), and did so only after waiting to see the new drive in use
for several hours.

> The obvious starting place to look for such
> documentation is the big blue question-mark "Support" link on Dell's
> front page - 8 clicks later
> (http://support.dell.com/support/edocs/systems/pe2850/en/index.htm) I
> find a listing of some very pertinent documentation.

Those documents only cover physical installation of the RAID card and
backplane related components.

I did find it just now- the PERC card documentation is elsewhere on
the website.  You have to use "Option 2" to manually navigate through
a tree of devices.  There's no link to it from the search results from
"enter a service tag" or "pick your system from a list" interface.

Brett