Microserver Gen8 Failure - Move Disks

Additional information ....

OK, so I booted the server and I could see a 1785-Drive Array Not Configured message displayed once the Bl20i controller started.

I hit F5 to go into the  HP Smart Storage Administrator and selected HPE Smart Storage Administrator. Once in there I selected the Bl20i controller and it showed on the right-hand-side of the screen there were 5 UNASSIGNED DRIVES. One of these will be the SSD drive (with my Windows OS on) and the other 4 the SATA drives (where my data is). When I select CONFIGURE under the ACTIONS section, I can see the 5 UNASSIGNED DRIVES again.

So from memory this looks like it was when I originally installed the system, which makes sense but I'm not sure what to do now as I didn't think I'd be creating a new array as it would destroy the data on the four SATA disks?

Do I need to go into the BIOS and change the boot to the Bl20i controller? I had a look at this earlier and there was a big red warning saying that if you selected that option it would destroy the data on the disks?

If it is showing the drives as unassigned in B120i mode then that implies there is no Smart Array metadata on them which would in turn imply that the previous server was not in B110 RAID mode, The "Dynamic" Smart Array controllers are not real RAID controllers but HPE's software running on the Intel chipset, You just have to change it to AHCI mode as detailed in https://support.hpe.com/hpsc/doc/public/display?docId=emr_na-c03871499. Since you were using Storage Spaces and didn't setup RAID on the B120i before you must have turned the B120i function off, a good idea as it's only a fakeRAID controller.

If you do accidentally create an array using the B120i it'll overwrite the first track with HPE's metadata which will mess up the beginning of the disks.

Thanks for the feedback.

That makes sense as I didn’t think I setup any raid because I knew I planned to use storage spaces.

I had a look at the link you provided and I had selected the AHCI SATA option and that allowed me to boot from the SSD drive. I tested that it booted the OS and that was OK. I only tested the initial boot where the Windows logo is displayed and then powered off. But this was without the 4 SATA disks in and I was informed elsewhere that if you add in the SATA drives the micro server won’t boot because when the SATA drives are in it doesn’t boot off SATA 5 connection? That didn’t make much sense to me because the SSD has the OS so how else would it have booted on my old server?

The question now is if it’s booting off the SSD and the 4 SATA disks are not in an array, if I let the OS just boot and see if it identifies the storage pool? Will this do any halm?

What version of Windows is running on the replacement Microserver?
What we want to do is to boot into Safe Mode and run DISKMGMT.MSC after skipping all messages about checking the volumes.  If your folders show up in explorer, you can then copy them to another storage device to create a backup.  If they don't show up, DISKMGMT will at least tell us what their configuration might be.
If it is running Windows 7, hit the F8 key several times after the HP logo appears to choose Safe Mode.  If it is running 8 or 10, it's a little trickier.

It’s Windows 10.

Thanks again for the feedback.

Can you just confirm something for me. You mention that we wanted to go into safe mode, which I thought was to avoid fully booting the OS. If I'm going into Windows to set options so that it will restart in safe mode, then at that point I've already fully loaded the OS which is what I thought you were trying to avoid?

Or are you saying to boot the OS without the 4 SATA drives in and just the SSD with the OS, put it into safe mode using the procedure. Shutdown and put the SATA drives in, then power on?

I assume I can put a USB stick in with the Windows 10 install and get to safe mode that way?

If you use the power button on the lock/login screen from a cold power on boot, it doesn't fully load all of the Windoze startup items and is about the best we can do.  The biggest item is that you want to press any key to abort having CHKDSK (autocheck) so it doesn't change anything on the drives.
And, I was saying to do it with NONE of the old PC's drives attached.  Use the one that came in the replacement unless we are already past that and you have seen the old drive boot successfully.

OK, so to summarise, I will do the following:

• Take out the 4 SATA disks, leaving just the SSD that has the OS installed
• Boot that up to the Windows lock screen and put it into safe mode as you've described
• Put the 4 SATA disks back in and let it boot into safe mode (cancelling any disk checks whilst it boots)
• Run DISKMGMT.MSC to check if it sees the drives and if not, collect any details that are there and report back here :)

That's it!  You don't perchance have something you could copy the files to, do you?  Once you have a backup, we're in way better shape.

Thanks.

I have lots of storage options like USB drives and other SATA drives, but the problem is how would I copy the data as I would need to go through the above to access it and if I could access it, then the problem would be solved anyway?

Until you have THREE copies of everything you care about, you can lose everything quicker than you can blink your eyes and what I was saying is that, if disk management shows the drives and you can explore them, now is the perfect time to get a second copy.
And, as a note, if the original boot drive does not work in the new PC, you will have serious permission issues with its drives and they can easily be time bombs that don't go off for several months.

... using four SATA disks in Storage Spaces Parity ...

Do _not_ touch those four disks with the HP RAID setup. That will crater the Storage Spaces (SS) setup.

Once the standalone SSD OS has finished booting it should see the SS Pool and any Virtual Disks.

Elevate PowerShell and:

Get-StoragePool
Get-VirtualDisk

Select allOpen in new window

If any of the above shows "DETACHED" then re-attach the pool and virtual disks. That should be it.

@Philip Elder - thanks for the feedback.

Can you confirm that you are saying I would need to boot using AHCI SATA so that the Microserver is booting without the RAID controller card in the equation?

I can confirm that if I set the BIOS options to AHCI SATA, all of the five disks are shown - the SSD drive and the 4 SATA drives. If that's the case, then it that doesn't work as the OS will not boot on SATA Port 5 if the SATA drive bays are in use. I was informed in a separate forum that the RAID controller card had to control the drives and as I moved the disks from the old server to the new server, I have to create a RAID 0 Array for each disk individually, e.g. one for the SSD and four more for the 4 SATA drives. They have explained that RAID 0 in this configuration is just like having a JBOD configuration. They have then advised that the controller handles access to the drives, including the SSD one and that you then set the boot drive to the SSD in the same HP SSA/BIOS utility. This made sense to me. I just can't remember originally setting up each drive with a RAID 0 drive (but I may have).

So, are you saying there is another way to boot with the SATA drives in the system? I know the OS boots, because if I remove the four SATA drives, leaving just the SSD, WIndows will boot. At the moment, I've just let it get to the Windows logo and then turned it off so it doesn't re-configure itself without the storage space / pool missing (although I understand you can move the storage pools from one server to another as long as they are on the same OS/update revision).

Does this make sense?

@Davis McCarn - unfortunately as I've mentioned in the post above, it's not possible to boot the server if the SATA drives are in the bay. If I remove them and change the SATA options in the BIOS to AHCI, then I can boot. With the drives in the bay, it simply states non-system disk as if there's no OS installed.

Apparently, I have to create a RAID 0 array for each drive in the HP SSA / BIOS. Then the RAID controller has control and you can nominate a logical drive to boot from, in this instance it would be the SSD drive.

If you had created RAID0 on each disk the controller would should the metadata on them and they would show as assigned.

What you may have done is followed the steps in http://www.admfactory.com/hp-microserver-gen8-boot-from-ssd-install-on-odd-bay/ and then turned RAID mode off again. Not entirely sure what that will do as don't have one to play with but that page may jog your memory.

Simple thing to do is to plug the four disks into a desktop machine with free SATA ports and AHCI set, boot, and check for the Storage Spaces setup.

I do hope there is a backup of everything on those drives just in case something hiccups or has hiccupped?

@andyalder - thanks for the link. I've read it and it makes sense what's in there, but I really can't remember how I set it up. I can't imagine that I would turn the RAID on and then off (but doesn't mean I didn't). The information in that article aligns with what has been proposed in another forum.

I raised the same question about the RAID metadata as I read in an HP technical paper that the metadata travels with the disks (and is on the controller), so my point was why didn't it just work when I moved all the same disks to exactly the same server hardware. He mentioned that if I had set the disks up as RAID0, there is no metadata. For other RAID levels, e.g. 1, 5, 6 etc. there is. That also made sense to me.



@Philip Elder - It's strange you should mention that because I thought of that myself, so I'd purchased a 4 port SATA PCI-E card which I could install in another server. It arrived today. I've just not tried it yet because I wasn't sure whether connecting them would cause any issues to the data on the disks, but I have it as a contingency. If it won't disrupt the data, then I would have nothing to lose. Any ideas? I've already been informed that storage spaces will roam but to ensure the Windows 10 versions are the same (I'm 90% sure they were both on the recent major update).

I don't have a backup because I can download the data again, but there's a lot of it so wanted to avoid it. Hence the reason why I put redundancy in the disks - only to have this issue!

Go for it. Plug them in and run the PowerShell commands in an elevated console.

I do not know that much about the specifics of Microsoft's Storage Spaces and Pools; but, with most traditional RAID controllers, it is critical that the drives be physically mounted in the exact same slots as they were in the original setup.  If you swap positions, you can be in very deep kimchi!
I did just read that Storage Spaces keeps the arrays configuration information on the media surface of the drives which is very good and makes it much easier than having it stored in ROM on the controller.
But; are you also saying that having the pool (SATA drives) connected results in the PC failing to boot?
If so, is there any chance you "moved" one of your folders (Pictures, Documents, Videos, etc.) to the storage pool?  If you did, that would explain Windows failing to boot if the pool is offline.

There is metadata on RAID0 individual disks with genuine Smart Array controllers but maybe not with these fakeRAID ones. Also genuine Smart Array controllers don't store the whole settings on the controller but these ones may do.

Do you have four spare drives, any size and condition will do. If so fit them and then follow that page above using the "create arrays with RAID0 option" then take them out and put your data drives in. If config is stored on the controller that may set it to the same state as the one stored on the failed server. You can also experiment with the spare test drives to see if the controller sees them after you use the clear config option with them removed and then put them back in again.

@Davis McCarn - If the SATA drives (where I happen to have the media pool located) are connected in the Microserver, the SSD that has the OS fails to boot, even when the BIOS settings are set to disable the RAID controller and everything is by AHCI SATA control. This is apparently in the design of the Microserver and is why there are workarounds out there to stop it working that way.

It definitely doesn't have anything to do with the Windows OS/data location. I've not moved anything and the issue with non-booting is that the system doesn't even get to read the SSD drive in the boot process because of the issue above. That's why you get the standard BIOS message about Non-System Disk which is what you get when the computer can't find a bootable OS. This is why someone on another forum highlighted you have to create a RAID0 array with every disk, starting with the SSD and making that bootable and then doing the same for each of the SATA hard drives.

@andyalder - Yes, I believe that's correct - the RAID controller in the HP Microservers are apparently fake RAID, i.e. software/firmware RAID running on a chip on the motherboard.

I like the idea of the test drives, so I may well give that a shot - I just need to find four spare drives.


I'm wondering whether the suggestion from Philip Elder is viable. I have purchased the 4-PORT SATA card, so if that won't destroy any data, then I could test that in another server. What does everyone think?

I would just connect one drive assuming Storage Spaces was setup redundantly, then destroying it won't matter too much. Then boot an imaging CD such as a winpe one or a linux based imaging one and see if it sees the partition data, if it is there but at an offset then there is metadata on track0. Don't let the CD write to the disk of course.

@andyalder - another good suggestion worth considering. Thanks.

@Phllip Elder - so the plan is for me to put the 4 port PCI-e controller in my other PC and attach these four drives. As you mention, as long as the HP server RAID card didn't do anything with them, then I just run those two commands you specified earlier? From my perspective I don't think it did anything special but in order to get the OS to boot when the SATA drives were in there I'm assuming you create the logical drives and set the one you want to boot and that configuration is all stored in the RAID controller even though in this instance we are not really using any RAID  because it's just one disk in each array.

My only question here is I thought storage spaces were portable and that simply be moved to another machine running the same OS revision?  If that's the case why wouldn't the storage pool simply show up without having to run those commands or are they always necessary anyway when moving locations?

Boot order is usually a setting in the BIOS itself. One can choose whatever drive or logical disk on a RAID array as boot drive there.

So no, I don't believe the software RAID setup would have any impact on that.

As far as SS portability, because, the original owner owns them. ;)

@Philip Elder - well, here's the update...

As you know I purchased a PCI-E 4 Port SATA card so I connected this to another server an hour ago. I then attached the SATA disks to each of the four ports and powered on. On the POST screen the SATA card was displayed and showed only three disks. Through process of elimination, I confirmed that port 4 on the card wasn't working (typical!). Anyway, as I'd setup the Storage Space with a Pool with parity, I speculated this was the same as having RAID5 and losing one disk.

I let Windows completely boot and the storage pool automatically mounted and the data was available in Windows Explorer. I'm now in the process of copying it across the network to a Synology NAS. Strange thing is even with just the three drives it's copy very quickly (120MB/sec), so that's great. Thanks again for your help, much appreciated.

@Everyone else - thanks for all of your input. There was some great ideas.

Great assistance from all!

Obviously no HPE metadata on the disks then or Windows wouldn't have found the data. There's definitely metadata on a single disk RAID0 on a real Shart Array controller as if you take one out and put in another server it lists all the disks that were in the old server as missing. Guess these "dynamic" ones may behave differently.

Yes, that makes sense and is why I went for the just moving it approach to see if it would find the data. I was advised to re-create the RAID0 disks individually and if that did write any metadata, then it would have potentially destroyed the data.