RAID drive replacement help!!

Hi,

We have a ML370 that has a 5 disk RAID 5 array.. One of the disks has
the red light flashing and the event log says the status is "predictive
failure".. I have just ordered a new drive which should be here in the
next day or so..

Now for the question..

What is the procedure for replacing the drive?

The new one is exactly the same as the old one..

Can I simply pull the old one and insert the new one or should I
shutdown the server and swap the drives when the server is off?
Will it automatically rebuild or do I have to tell it there is a new drive?

Really can't afford for this server to go down so would rather not "try"
my luck..

Any help greatfully appreciated..

"WipeOut" wrote in message
...
Hi,

We have a ML370 that has a 5 disk RAID 5 array.. One of the disks has the
red light flashing and the event log says the status is "predictive
failure".. I have just ordered a new drive which should be here in the
next day or so..

Now for the question..

What is the procedure for replacing the drive?

The new one is exactly the same as the old one..

Can I simply pull the old one and insert the new one or should I shutdown
the server and swap the drives when the server is off?
Will it automatically rebuild or do I have to tell it there is a new
drive?

Really can't afford for this server to go down so would rather not "try"
my luck..

Any help greatfully appreciated

Pop the failing drive, and replace. All three of the LED's on the disk
should start flashing, indicating that its rebuilding into the RAID set. No
reboot should be required, no downtime. Its the definition of
'hot-swappable'.

The HP arraycontrollers are legendary in thier ability to do this stuff. NO
SWEAT!

- LC

On 2007-03-13, Nut Cracker wrote:

Pop the failing drive, and replace. All three of the LED's on the disk
should start flashing, indicating that its rebuilding into the RAID set. No
reboot should be required, no downtime. Its the definition of
'hot-swappable'.

The HP arraycontrollers are legendary in thier ability to do this stuff. NO
SWEAT!

There was a post on a blog (Jeremy Zawodny's?? I think?) that was talking
about a home-brew linux based RAID server that died recently when a
disk died and the controller freaked out and nuked the rest of the RAID
data on the other disks making it unrecoverable.. In his case he had a
backup plan in place and nothing much was lost.. However, I'm in your camp
and believe that no home-use card that you get a your local CompUSA or Fry's
is going to behave nearly as well as the Compaq/HP systems are in the same
scenario.. I've not heard of something like his situation happening
on HP/CPQ hardware.. Of course, YMMV and I may be talking out of my rear!
(8-

"Rick F." wrote in message
...
On 2007-03-13, Nut Cracker wrote:

Pop the failing drive, and replace. All three of the LED's on the disk
should start flashing, indicating that its rebuilding into the RAID set.
No
reboot should be required, no downtime. Its the definition of
'hot-swappable'.

The HP arraycontrollers are legendary in thier ability to do this stuff.
NO
SWEAT!

There was a post on a blog (Jeremy Zawodny's?? I think?) that was talking
about a home-brew linux based RAID server that died recently when a
disk died and the controller freaked out and nuked the rest of the RAID
data on the other disks making it unrecoverable.. In his case he had a
backup plan in place and nothing much was lost.. However, I'm in your camp
and believe that no home-use card that you get a your local CompUSA or
Fry's
is going to behave nearly as well as the Compaq/HP systems are in the same
scenario.. I've not heard of something like his situation happening
on HP/CPQ hardware.. Of course, YMMV and I may be talking out of my rear!
(8-


I have seen that happen with Dell PERC controllers ... A LOT. Dell even has
a name for the event, which they call 'sympathy failures'.

I cringe whenever I have to work on those. Although I will say, the veritas
based ArrayManager application (which replaces Windows Disk Mangement) isnt
that bad. Not as good as Veritas Volume Manager, or Veritas Storage
Foundation for Windows ... but not a bad product. Its actually more
functional that the ACU, but obviously at the cost of reliability.

- LC

Nut Cracker wrote:
"WipeOut" wrote in message
...
Hi,

We have a ML370 that has a 5 disk RAID 5 array.. One of the disks has the
red light flashing and the event log says the status is "predictive
failure".. I have just ordered a new drive which should be here in the
next day or so..

Now for the question..

What is the procedure for replacing the drive?

The new one is exactly the same as the old one..

Can I simply pull the old one and insert the new one or should I shutdown
the server and swap the drives when the server is off?
Will it automatically rebuild or do I have to tell it there is a new
drive?

Really can't afford for this server to go down so would rather not "try"
my luck..

Any help greatfully appreciated

Pop the failing drive, and replace. All three of the LED's on the disk
should start flashing, indicating that its rebuilding into the RAID set. No
reboot should be required, no downtime. Its the definition of
'hot-swappable'.

The HP arraycontrollers are legendary in thier ability to do this stuff. NO
SWEAT!

- LC



OK that makes me feel better.. Lets hope all goes well..

Thanks..

On 2007-03-13, Nut Cracker wrote:

I have seen that happen with Dell PERC controllers ... A LOT. Dell even has
a name for the event, which they call 'sympathy failures'.

I cringe whenever I have to work on those. Although I will say, the veritas
based ArrayManager application (which replaces Windows Disk Mangement) isnt
that bad. Not as good as Veritas Volume Manager, or Veritas Storage
Foundation for Windows ... but not a bad product. Its actually more
functional that the ACU, but obviously at the cost of reliability.

Interesting.. In this case, Jeremy (see link below if you're interested)
is suggesting that had he just used Linux' software Raid instead of HW,
things would have been better.. Anyway, in case you've got nothing better
to do :

http://jeremy.zawodny.com/blog/archives/008696.html

"Nut Cracker" wrote
"WipeOut" wrote in message
What is the procedure for replacing the drive?

Can I simply pull the old one and insert the new one or should I shutdown
the server and swap the drives when the server is off?
Will it automatically rebuild or do I have to tell it there is a new
drive?

Pop the failing drive, and replace. All three of the LED's on the disk
should start flashing, indicating that its rebuilding into the RAID set.
No reboot should be required, no downtime. Its the definition of
'hot-swappable'.

The HP arraycontrollers are legendary in thier ability to do this stuff.
NO SWEAT!

There's no legend about it. It's more like hard fact.

LC's procedure will work in about 98% of all disk replacements. The
remaining 2% is a faulty (new) disk or a faulty backplane, but the chances
of data loss are about 1 in 1000.

"Jez T" wrote in message
...

"Nut Cracker" wrote
"WipeOut" wrote in message
What is the procedure for replacing the drive?

Can I simply pull the old one and insert the new one or should I
shutdown the server and swap the drives when the server is off?
Will it automatically rebuild or do I have to tell it there is a new
drive?

Pop the failing drive, and replace. All three of the LED's on the disk
should start flashing, indicating that its rebuilding into the RAID set.
No reboot should be required, no downtime. Its the definition of
'hot-swappable'.

The HP arraycontrollers are legendary in thier ability to do this stuff.
NO SWEAT!

There's no legend about it. It's more like hard fact.

LC's procedure will work in about 98% of all disk replacements. The
remaining 2% is a faulty (new) disk or a faulty backplane, but the chances
of data loss are about 1 in 1000.

1 in a 1000 .... wouldnt the percentage then be, " ... about 99.90% of all
disk replacements" ?

just having some fun, Jez

=)

- LC