can the 1U Xserve SATA drives be replaced with larger aftermarket drives?

[samoir]

Hi there everyone

I have a new 1U Xserve – 8 core / 2008 model with 3x80GB drives onboard [u]Apple Hardware RAID card[/u] running Leopard Server OS X10.5.1
Due to the expense of Apple Hardware, ive tried swapping out the three 80Gb SATA drives (product no: ST380815AS) which shipped with it for after market Seagate ES.2 1000Gb SATA drives (product no: ST31000340NS).

Does anyone know if it’s possible to do this ??

Am trying to setup:
Bay 1: 80GB OS
Bay 2 and 3 as RAID 1 mirror.

The drives are recognised, and i’ve been fully using RAID utility to do this (should i be using Disk Utility instead?).
However – am having issues when using RAID utility. The RAID set and RAID volume can be created, it takes about 6 hours to create, and the volume mounts. However when i do a restart the RAID utility alert comes up, and i get a yellow alert – Viable (Degraded).
On a restart in RAID 1 the RAID Volume appears on the desktop, however with the yellow degraded alert. When i JBOD both and do a restart they simply don’t appear.

Any thoughts, or is it 100% not possible to swap these drives for aftermarket when using the standard Hardware RAID card?

thank in advance

Samoir

[khiltd]

To the best of my knowledge, no Apple mass storage driver performs any sort of vendor string based rejection; a drive is a drive.

Step-by-step instructions are in the docs: http://images.apple.com/server/docs/RAIDAdmin1.2_121406.pdf

[samoir]

khiltd
Thanks for your reply. The RAID pdf you mentioned is the guide for an Xserve RAID (14 drive unit) which i don’t have).
The machine i use is a Xserve 1U with has 3x80Gb ADM’s and an Apple Hardware RAID card. The guide ive been following is this one [url]http://images.apple.com/server/docs/RAID_Utility_User_Guide.pdf[/url] but it gives me a drive error.

There are other posts on Google which say there might be some firmware on the Apple RAID Card itself that needs to match some firmware on the hard drives, there is nothing conclusive to say this is the case though.

I have 10 years experience with Apple desktop tower machines, but this is the first Xserve ive owned, so i don’t know if there is deviant firmware that prevents a customer from upgrading using after market parts, forcing the customer buy genuine Apple 1000GB ADM’s at 3 times the price!!

Would be great to know if anyone has swapped or replaced their hard drives in an Intel 1U Xserve, that has a Apple Hardware RAID card and what proceedure is used to do this.

cheers again

samoir

[khiltd]

I don’t have access to such a machine, but it wouldn’t make much sense for Apple to tether the card to one particular drive since they do not make them and switch models frequently. It’s certainly possible, but not very intelligent on their part. I think they even stopped policing opticals in the DiscRecording framework–just way more trouble than it’s worth.

How long does the array show up as degraded? Are the replacement drives known to be working properly and do they move data fast enough? Seagates are anything but bulletproof, so you might want to try more than one set to narrow it down.

[samoir]

To answer your questions:
– the drives are brand new and in working condition.
– upon restart the RAID set shows up as Degraded and remains this way. I have only left it asay and hour or so, not days or anything, but the status doesn;t seem to change at all within an hour.

My process so far has been remove disks from 80GB SATA HD’s from the sled.
Put in brand new Seagate ES.2 Barracuda 1000GB drives. (server series drives)
Go to Raid Utility
Create new RAID set (Raid 1)
6 hours later the newly created RAID volume shows and is built.
No issues up to this point. (all lights green)

Now i reboot the machine

Then i get a yellow warning RAID set “Viable” (degraded).

From this point m not sure if i have to leave it for some length of time or start the process again ??

My gut feeling is there is some time lag between the server starting and the RAID hardware card detecting the drives, so it thinks there is an error. Maybe?

[khiltd]

“Brand new” isn’t necessarily synonymous with “working.” Have you run any tests on them?

I’d try leaving it alone a little longer, check the connections and also make sure you’re not mixing DMA modes on the bus. Can’t think of anything else off the top of my head. If they really wanted to stop you from using other drives they probably would have just had the driver ignore them altogether.

[samoir]

ive just tested the drives using Disk Utility on a Mac Pro 8 Core and no problems.

Ive dropped them back into the Xserve. Using RAID Utility ive setup some basic JBOD RAID SETS and VOLUMES for each drive, and they mount as volumes on the desktop.

However as soon as the machine is restarted, the issue continues. Either one of the drives doesn’t show on Desktop or within the RAID Utility. I believe its got something to do with the Apple Hardware RAID card and the OS and the hardware RAID card not recognising the drives at the right time, so the OS believes they dont exist. Meaning: does the OS Startup and there is some hardware lag with the Apple RAID card, and so the delay is actually creating havoc with the drives appearing in RAID utility thus not mounting??

Any thoughts at all??

[khiltd]

When the drive refuses to mount does it have an entry in the IORegistry or is it completely AWOL?

[deemery]

For what it’s worth, Apple (at least OS X software) RAID volumes take order -hours- to rebuild RAID sets.

dave

[samoir]

have noted above that it took 6 hours to build this RAID set… but that is not the issue here.
The RAID set builds correctly but get blown out/ corrupted on restarting the machine.

Im begginning to think there is embeded Apple Hardware on the drives that is required.

cheers

samoir

[khiltd]

[QUOTE][u]Quote by: samoir[/u][p]
Im begginning to think there is embeded Apple Hardware on the drives that is required.[/p][/QUOTE]

That’s just ridiculous. Drive manufacturers submit production models to Apple, and Apple engineers then qualify the drives for use with prototype Apple hardware. At the end of this process a list of drives which survived the testing matrices is compiled and passed on to purchasing who then selects the cheapest model which performs as needed. Haggling takes place as the manufacturers strive to win the big contract, and eventually one or two models are selected for a production run. Sometimes, when a minor deviation from a published specification is the only thing keeping an otherwise good (or cheap) drive from qualifying, the manufacturer will alter their firmware in order to correct the problem, but I have never heard of anyone building a custom firmware version specifically for units which are sold to Apple. They simply don’t have the marketshare to make such a demand of any manufacturer. I don’t know of anyone who does.

A very, very long time ago when 30MB hard drives were considered expansive, all drive manufacturers would allow OEM purchasers to poke a custom vendor string into the firmware, so instead of a “SEAGATE” you’d see an “FWB” or “APPLE” if you happened to be a driver engineer, but that’s as far as that ever went. The practice was abandoned completely as specifications stabilized and manufacturers started adhering to them more rigidly. Anyone who puts out an “SATA” drive that does not comply with the SATA spec isn’t going to sell very many drives.

What you’re saying is not impossible, but it is infinitely more likely that the drives you have are simply inadequate or that something is being configured incorrectly. I’m not familiar with Apple’s current RAID driver sources, but mismatched DMA modes is one thing that can definitely cause some problems. I’m sure there are other factors as well.

[samoir]

Yeah i definately agree with you, and think it’s ridiculous as well.
Have replaced 8 Core Mac Pro drives with these exact ones and had no issue (under sotware RAID though). I find it extremely tedious that it takes 6 hours to build a RAID 1 volume, and they don’t mount on restart.

I don;t know anything about DMA, or mismatched DMA modes.
Will look some further looking on that (or if you have a starting point that’d be great 🙂 …).

To answer your previous post, the drives showed up under:
Apple System Profiler [b]>[/b] Hardware RAID [b]>[/b] after restart but not in RAID Utility.
The IORegistry thing is unknown territory for me. However im reading some stuff on the Apple Developer Connnection about IORegistry Explorer.

FYI:
These drives im using are the same ones Apple are shipping as there main 1TB drive, just after market.
The drives generally reappear in RAID Utility later, with the RAID set degraded, but their “State” has changed from “Assigned” to “Roaming”

I will do some looking on what you mentioned.

[samoir]

have totally removed drives 2 & 3 from the Xserve
restarted without them
shut down and started up with new 1TB drives in place.
Am building the RAID volume now. Will let you know how this works out.

Was taking your IO concept and making sure the drive configs have been removed etc.
Thought this might help overcome any confusion to the IO in the start up process.

Will post result later (tommorrow morn when complete)
Sam

[samoir]

All works perfectly until restart. Initialised RAID set last night, 6.5 hours. All green lights. Took IO registry dump (from Apple Sys Profile logs)
and screen shots of it working. When restarted, disk doesn’t show on desktop, doesn’t show in RAID set at all, no Volume information and red lights from RAID utility.

If the firmware is not on the drives themselves, does anyone know if the circuit board in the sleds are programmed with firmware that could be causing this issue.

I think Apple must have done something sneaky to make you buy their 1TB drive and sled!

Gagggh and painful

[khiltd]

I doubt it’s anything more sneaky than requiring ECC DIMMs. Call Apple support and ask what the specific requirements are for that card.

[Author]

Posts