Server Raid Ultra 160 Won't Recognize New Drives. Help.

Here's the problem, I have an IBM Netfinity 5100 with a ServerRaid 4x
Ultra 160. I originally built the server with three drives in a Raid
5, everything was fine, no problems. Recently I purchased three
additional drives and fully populated the cage, the problem is when I
run either the Mini Config or the Server Raid GUI config it only
recognizes one of the new drives. Drives 1 (Bay #0), 2 (Bay #2) and 3
(Bay #3) are the original and work fine, drive 4 (Bay #4) shows
standby, drive 5 (Bay #8) is fine, and drive 6 (Bay #9) shows up as
disabled. I looked everywhere and can't find a way to put these drives
in "ready" mode. What the heck am I missing? Is there a way to
activate these drives without losing the data I have on my original
array? Help…

Rick

"Rick B." <> wrote in message
news: om...

> Here's the problem, I have an IBM Netfinity 5100 with a ServerRaid 4x
> Ultra 160. I originally built the server with three drives in a Raid
> 5, everything was fine, no problems. Recently I purchased three
> additional drives and fully populated the cage, the problem is when I
> run either the Mini Config or the Server Raid GUI config it only
> recognizes one of the new drives. Drives 1 (Bay #0), 2 (Bay #2) and 3
> (Bay #3) are the original and work fine, drive 4 (Bay #4) shows
> standby, drive 5 (Bay #8) is fine, and drive 6 (Bay #9) shows up as
> disabled. I looked everywhere and can't find a way to put these drives
> in "ready" mode. What the heck am I missing? Is there a way to
> activate these drives without losing the data I have on my original
> array? Help.
>



Rick, I'm not sure what RAID controller is being used in that machine. I
would venture to say that you will have to do a complete backup, including
system state, and reboot to get back into the RAID configuration. Once
there, you will need to clear your old configuration and rebuild the array.
Unfortunately, some of the low-mid class controllers won't let you increase
array size on the fly. I know this isn't what you wanted to her, but this
is about all you can do unless someone else that knows that machine better
than I can help. Good luck and let us know how you make out.



Rita

On Wed, 9 Jun 2004 18:19:18 -0400, "Rita Ä Berkowitz"
<> wrote:

>"Rick B." <> wrote in message
>news:. com...
>
>> Here's the problem, I have an IBM Netfinity 5100 with a ServerRaid 4x
>> Ultra 160. I originally built the server with three drives in a Raid
>> 5, everything was fine, no problems. Recently I purchased three
>> additional drives and fully populated the cage, the problem is when I
>> run either the Mini Config or the Server Raid GUI config it only
>> recognizes one of the new drives. Drives 1 (Bay #0), 2 (Bay #2) and 3
>> (Bay #3) are the original and work fine, drive 4 (Bay #4) shows
>> standby, drive 5 (Bay #8) is fine, and drive 6 (Bay #9) shows up as
>> disabled. I looked everywhere and can't find a way to put these drives
>> in "ready" mode. What the heck am I missing? Is there a way to
>> activate these drives without losing the data I have on my original
>> array? Help.
>>
>
>
>
>Rick, I'm not sure what RAID controller is being used in that machine.

I suspect He means the IBM ServeRAID 4Lx. It's no matter though,
AFAIK all the IBM ServeRAIDs use a common BIOS & configuration tool.

>would venture to say that you will have to do a complete backup, including
>system state, and reboot to get back into the RAID configuration. Once
>there, you will need to clear your old configuration and rebuild the array.
>Unfortunately, some of the low-mid class controllers won't let you increase
>array size on the fly.

Before doing anything so dramatic rechek termination and SCSI ID
conflicts and verify that the drives & cage are good using a different
controller. Also make sure you wipe clean the boot sector of the new
drives. I seem to remember once a loose molex conector and some MBR
info giving me similar confusion a long time ago.

yes you can add drives & volumes to a serveraid without having to
reinitialize pre-existing volumes, and you can perform hotswap,
failover, etc. It's not really such a low-end controller, just an
older one (whose software is still being developed). Software feature
wise the older Serveraids are barely distinguishable from the newest &
most expensive ones.

When you boot off the Serveraid support CD the new drives should be
recognized and you should be prompted to create a new volume. there
is an option to force discovery of new drives, but you shouldn't need
to do that. Remember the configuration software you use must match
the BIOS revision

Of what I remember of ServeRAID's, once they designate a drive as
defunct, it's not always so easy to bring it back online unless you
move it to a different SCSI ID (I think it remember's the drive's
serial number). If you know the setup to be otherwise good, and you
have a complete backup on hand (trueimage & ghost work well with this
controller) you could clear the controller's configuration, then on
reboot use the mini-config to write the raid config from the drives to
the controller, then boot into the serveraid CD and the insert the new
hotswap drives and force discovery of them. They should be RDY at
that point to configure volumes

I haven't had to mess with the cofig of our ServeRAID based machines
in some time, so sorry if my memory is a little rusty.

Hope this helps.

"newB" <> wrote in message
news:...
> On Wed, 9 Jun 2004 18:19:18 -0400, "Rita Ä Berkowitz"
> Before doing anything so dramatic rechek termination and SCSI ID
> conflicts and verify that the drives & cage are good using a different
> controller. Also make sure you wipe clean the boot sector of the new
> drives. I seem to remember once a loose molex conector and some MBR
> info giving me similar confusion a long time ago.
>



Thanks for the wonderful information. I'm curious, aren't these SCA drives?
If so, these plug directly into a backplane and there shouldn't be a need to
fiddle with SCSI ID and termination.


> yes you can add drives & volumes to a serveraid without having to
> reinitialize pre-existing volumes, and you can perform hotswap,
> failover, etc. It's not really such a low-end controller, just an
> older one (whose software is still being developed). Software feature
> wise the older Serveraids are barely distinguishable from the newest &
> most expensive ones.
>



Great news.


> When you boot off the Serveraid support CD the new drives should be
> recognized and you should be prompted to create a new volume. there
> is an option to force discovery of new drives, but you shouldn't need
> to do that. Remember the configuration software you use must match
> the BIOS revision
>

I would think that it would be straightforward for him. The possibility
does exist that he might have drives of questionable functionality.



Rita

Just to clarify, there shouldn't be any reason for you to actually
delete and reinstall your old Raid 5 volume & data. The process is
simple if the new drives will make a new volume. Otherwise use the
Logical Drive Micgration tool:

http://www-307.ibm.com/pc/support/si...id=MIGR-4JHS2W

On Thu, 10 Jun 2004 18:08:40 -0400, "Rita Ä Berkowitz"
<> wrote:

>
>"newB" <> wrote in message
>news:.. .
>> On Wed, 9 Jun 2004 18:19:18 -0400, "Rita Ä Berkowitz"
>> Before doing anything so dramatic rechek termination and SCSI ID
>> conflicts and verify that the drives & cage are good using a different
>> controller. Also make sure you wipe clean the boot sector of the new
>> drives. I seem to remember once a loose molex conector and some MBR
>> info giving me similar confusion a long time ago.
>>
>
>
>
>Thanks for the wonderful information. I'm curious, aren't these SCA drives?
>If so, these plug directly into a backplane and there shouldn't be a need to
>fiddle with SCSI ID and termination.

It was the molex power connector to the backplane which was the
problem. Ended up making everything flaky.

When you have a situation where the backplane allows you to manually
configure SCSI ID and/or there are multiple backplanes on the same
channel, ID can be an issue. I guess this isn't really relevant for
his netfinity though. With the Netfinity EXP racks, you do sometimes
have to give a little thought to their "auto" configuration.

>
>
>> yes you can add drives & volumes to a serveraid without having to
>> reinitialize pre-existing volumes, and you can perform hotswap,
>> failover, etc. It's not really such a low-end controller, just an
>> older one (whose software is still being developed). Software feature
>> wise the older Serveraids are barely distinguishable from the newest &
>> most expensive ones.
>>
>
>
>
>Great news.
>
>
>> When you boot off the Serveraid support CD the new drives should be
>> recognized and you should be prompted to create a new volume. there
>> is an option to force discovery of new drives, but you shouldn't need
>> to do that. Remember the configuration software you use must match
>> the BIOS revision
>>
>
>I would think that it would be straightforward for him. The possibility
>does exist that he might have drives of questionable functionality.
>
Agreed. I just think he should try a few more things before
RMAing/scrapping them. Sometimes it's the stupidest things that cause
the most downtime/frustration.

On rereading the original post it looks like he needs to remove the
hot spare from the config to chage drive status in Bay #4 and test the
functionality of the drive in Bay #9 by moving to a different bay or
computer. Again if all the drives and bays work correctly my
suggestion re clearing and restoring the raid config should force
things to work without loosing any data on the original array.

"newB" <> wrote in message
news:...

> It was the molex power connector to the backplane which was the
> problem. Ended up making everything flaky.
>



OK, I understand now. I wasn't sure if you were talking about a Molex on
each drive. Yes, this can be a problem if power is lost to the backplane.


> When you have a situation where the backplane allows you to manually
> configure SCSI ID and/or there are multiple backplanes on the same
> channel, ID can be an issue. I guess this isn't really relevant for
> his netfinity though. With the Netfinity EXP racks, you do sometimes
> have to give a little thought to their "auto" configuration.
>



The only experience I had with Netfinity was with their older EXP15 racks.
What I remember from years back they didn't have any options for selecting
individual drive IDs, but had a micro dial switch on the rear to select rack
ID and a DIP switch to select a few options as well as separating the
backplane into two electrically isolated units. Again, I'm going by memory
on this.



> Agreed. I just think he should try a few more things before
> RMAing/scrapping them. Sometimes it's the stupidest things that cause
> the most downtime/frustration.
>

Absolutely, I agree since these units are so straightforward that they leave
little room for user error.


> On rereading the original post it looks like he needs to remove the
> hot spare from the config to chage drive status in Bay #4 and test the
> functionality of the drive in Bay #9 by moving to a different bay or
> computer. Again if all the drives and bays work correctly my
> suggestion re clearing and restoring the raid config should force
> things to work without loosing any data on the original array.



It's a long shot that might pay off for him. Generally, from a time,
manpower, and economics standpoint I would weigh the benefits of just
backing up and blowing away the old configuration instead of monkeying with
an option that might involve too much resources for his learning curve.



Rita

Thanks for all the followup, I did try using different drives in
different bays before I posted this message. It made no difference
whatsoever which bay the drives were physically located, the
configuration utility always reported the exact same status for the
drives in question. I'll check the backplane and all the cable
connections, if that fails I guess I'll ghost it and blow it away.
Fortunately the server is not yet in full production, thus giving me a
little wiggle room to do this.

Thanks Again,

Rick

"Rick B." <> wrote in message
news: om...

> Thanks for all the followup, I did try using different drives in
> different bays before I posted this message. It made no difference
> whatsoever which bay the drives were physically located, the
> configuration utility always reported the exact same status for the
> drives in question. I'll check the backplane and all the cable
> connections, if that fails I guess I'll ghost it and blow it away.
> Fortunately the server is not yet in full production, thus giving me a
> little wiggle room to do this.
>



Good luck. One suggestion is that you might also want to do a full backup
using your backup/tape utility to be 100% sure that you have some redundancy
before you proceed. I was never fully confident with Ghost.



Rita

On Fri, 11 Jun 2004 06:57:57 -0400, "Rita Ä Berkowitz"
<> wrote:

<snip>
>
>> On rereading the original post it looks like he needs to remove the
>> hot spare from the config to chage drive status in Bay #4 and test the
>> functionality of the drive in Bay #9 by moving to a different bay or
>> computer. Again if all the drives and bays work correctly my
>> suggestion re clearing and restoring the raid config should force
>> things to work without loosing any data on the original array.
>
>
>
>It's a long shot that might pay off for him. Generally, from a time,
>manpower, and economics standpoint I would weigh the benefits of just
>backing up and blowing away the old configuration instead of monkeying with
>an option that might involve too much resources for his learning curve.
>

I sort of agree with your basic philosophy but clearing and restoring
the controller configuration would have the same effect without the
extra step & time & inviting potential for error/loss in completely
destroying and restoring all data. The solution could come even
faster if he could just determine a drive to be bad &/or simply
correct a controller configuration error. I find nondestructive
solutions often preferable and less time and resource intensive.
Regardless we both agree he needs comprehensive, up-to-date backups to
fall back on. Sidestepping the 'learning curve' wont help the next
time it or something similar needs to be serviced or upgraded when it
is in full production.

Rick, since you have some time and are interested in taking time to
blow away & restore the array, why not also get the latest support CD,
upgrade the BIOS and use the most recent tools? I'm happy enough with
6.11 but 7.0 came out 2004/05/05.

http://www-306.ibm.com/pc/support/si...GR-495PES.html