RAID 1 Protected Mode - Error configuring local disk controller

[Danny Sandner]

Hey guys,

I stuck on a little problem:

I have a Service Profile "ServerWin" with local disk policy "RAID1 - Protected" on Server1 (B200M3). Servers was installed with Win 2k8R2.

For a test, i disassociated the Service Profile "ServerWin" from Server1. I removed local disk 1 (local disk 0 is still in the server) as backup disk for the Windows Server "ServerWin".

I associated Service Profile "ServerESX" with local disk policy "no raid" on Server1 and installed ESXi on disk 0. After finishing the test, I disassociated "ServerESX" from Server1.

Now I want to go back to my Windows Server.

I remount in the second disk (disk 1) and associated the service profile "ServerWin" on Server1. Now I get

Drive state local disk 0: Unconfigured Good

Drive state local disk 1: Online

Why UCS is not taking local disk 1, which is RAID 1 protected already, rebuild disk 0 and go on with association of the service profile?

regards

/Hugo

[Keny Perez]

Hugo,

One question, have you had a chance  to get to the RAID controller and check that both disks are exactly of the same size in Megabytes? Not GB....

-Kenny

[Danny Sandner]

Hi, Both disks are the same, exactly the same disks. But I have a solutions:

I went to the MegaRAID Webbios and add the "unconfigured Good" Disk 1 as "HSP" (Hot Spare). After, the rebuilding started and the service profile association went on. That was all.

I dont know, if there is a better solution or this is the only one. It is a little bit tricky, because you have to associate the service profile and go directly to the KVM console and MegaRAID Webbios to make changes.

/Hugo

[Keny Perez · Accepted solution]

I have done the same as a workaround before, not sure if the system recognizes one of the disks as part of a array already or something...  not sure if there is a bug about this or an expected behavior.

Would be interesting to see if you can replicate this at will.

-Kenny

[asafayan]

Hi,

I have a C Series server.  I have 2 hot spares in slots 11 and 12.  They were both 146G drives.  I had to replace the drive in slot 12 with a 300G drive and it came up as "unconfigured good". 

Is the normal/expected behavior of the controller to create this new drive as a Hot Spare because the drive, that was previously in this slot, was a Hot Spare? 

In others - how does the controller know that I don't want to add this drive to the RAID array and not as a Hot Spare? 

Is the process of adding a drive manually in the WebBIOS as a Hot Spare intuitive? If not, do you have any bullet point steps to do so?

TIA,

Amir

[Keny Perez]

It is expected to come up as unconfig good cause the disks being replaced is a hot spare, despite the fact they can configured as hot spare for a dedicated array, it does not mean they part of the array. They will part of it til there is a disk failure.

If your server is a rack server with fw 1.5/2.0, the process of configuring the new disk is easy. 

Do you have a blade server or rack server (integrated in UCSM or standalone)?

-Kenny

[asafayan]

Hi Kenny,

It is a UCS C210 M2  standalone running firmware version 1.4.(3u).

Here is the output of the show physical-drive command:

ucs-c2xx-m2 /chassis/storageadapter # show physical-drive

Physical Drive Number Controller Info Valid Status                 Manufacturer   Model          Predictive Failure Count Drive Firmware Coerced Size   Type
--------------------- ---------- ---------- ---------------------- -------------- -------------- ------------------------ -------------- -------------- -----
1                     SLOT-5     Yes        online                 SEAGATE        ST9300653SS    0                        0002           285148 MB      HDD
2                     SLOT-5     Yes        online                 TOSHIBA        MK3001GRRB     0                        5702           285148 MB      HDD
3                     SLOT-5     Yes        online                 TOSHIBA        MK3001GRRB     0                        5702           285148 MB      HDD
4                     SLOT-5     Yes        online                 TOSHIBA        MK3001GRRB     0                        5702           285148 MB      HDD
5                     SLOT-5     Yes        online                 SEAGATE        ST9300653SS    0                        0002           285148 MB      HDD
6                     SLOT-5     Yes        online                 SEAGATE        ST9300653SS    0                        0002           285148 MB      HDD
7                     SLOT-5     Yes        online                 SEAGATE        ST9300653SS    0                        0002           285148 MB      HDD
8                     SLOT-5     Yes        online                 SEAGATE        ST9300653SS    0                        0002           285148 MB      HDD
9                     SLOT-5     Yes        online                 SEAGATE        ST9300653SS    0                        0002           285148 MB      HDD
10                    SLOT-5     Yes        online                 SEAGATE        ST9300653SS    0                        0002           285148 MB      HDD
11                    SLOT-5     Yes        hot spare              TOSHIBA        MK1401GRRB     0                        5702           139236 MB      HDD
12                    SLOT-5     Yes        unconfigured good      TOSHIBA        MK3001GRRB     0                        5702           285148 MB      HDD
ucs-c2xx-m2 /chassis/storageadapter #

Are you stating that it is expected to be in an "unconfigured good" state?  I thought that the controller should recognize the new drive being inserted and configure it as a "hot spare" - given that the previous drive in slot 12 was also a hot spare.

It is RAID 5 and my controller is a LSI MegaRAID SAS 9261-8i with firmware 12.12.0-0149.

TIA,

Amir

[Keny Perez]

Amir,

I have never tried it myself but look at it this way... when the disk is configured in an array the RAID controller will have to rebuild the array and therefore a task to synch up the guy will take place but with a hot spare I dont see that...  you might want to look at the user guide of the controller or even simulate that again by removing the new disk and installing the old one and see what happens...

If you may do it, let us know...

-Kenny

[Danny Sandner]

Hi again,

Is there a online management tool for the RAID controller, where I can manage the Disks and the RAID from Windows like HP Smart Array Configuration Utility for HP Servers?

I see you can have look to the storage adapter via CLI. Can you configure those via CLI,too? The only way to manage the disks and raid I know at the moment is the WebBIOS of the storage controller.

/Danny

[Keny Perez]

Danny,

You can use MegaCLI to manage the RAID controller in a command line interface or the newer StorCLI (also a command line tool) or the GUI version "MegaRAID Storage Manager".

Again, with the CIMC 1.5 and later, managing the array is a lot easier but that is for rack servers only as UCSM manager users normally define the disks configuration through the service profile...

HTH,

-Kenny

[asafayan]

Summary of my situation:

The original 146G drive in slot 12 was set to ‘hot spare’.  The client then replaced this 146G drive with a 300G drive.  The drive state was "unconfigured good" and never transitioned, automatically, to hot spare.

I followed these steps, without service impact, to designate the drive as a hot spare:

STEP 1

1. Access ESXi1 via vSphere

2. Select Host >> Configuration >> Storage >> Browse datastore2 >> Upload MegaCLI and libstorlib.so to the Datastore

STEP 2

Make MegaCli Executable

1. SSH into ESXi IP address.
2. Browse to the data store in the shell.
   1. cd /vmfs/volumes/datastore2
   2. type ls  to check if the two files are present.
   3. now type-  chmod 777 MegaCli (to grant executable permission to MegaCLI)

STEP 3:

Set Global Hot Spare

1. From ESXi shell, issue the following command:

./vmfs/volumes/4fb355df-040855a4-00ca-001018d7454a # ./MegaCli -PDHSP -Set -PhysDrv [18:12] –a0

Here is where it gets weird.  I informed the client that the RAID controller does not automatically set the drive to ‘hot spare’ and that requires intervention, on her part, to do that.  She informed me that she had done this very same task (replacing hot spare 146G drives with 300G drives) on their other C Series – AND the drives were automatically set to ‘hot spare’.

What I'm struggling to understand now is - what is the expected behavior when a replacement hot spare is inserted ?  Does the drive automatically transition to the hot spare state OR do you have to manually configure it as a hot spare?

Does it depend on the firmware version?

ESXi 1 info  / server required us to manually set the drives to hot spare:

BIOS Version: C200.1.4.3k.0

Firmware Version: 1.4(3u)

LSI MegaRAID SAS 9261-8i – Firmware Package Build: 12.12.0-0149

ESXi 2 info  / drive automatically set to hot spare:

BIOS Version: C200.1.4.3h.0

Firmware Version: 1.4(3p)

LSI MegaRAID SAS 9261-8i – Firmware Package Build: 12.13.0-0176

I reviewed the tech support from the C Series and, in fact, I see where she is 100% accurate:

6:2014 Oct  8 11:27:32:BMC:storage:-: SLOT-5: Global Hot Spare created on PD 27(e0x12/s11) (global,rev)

6:2014 Oct  8 11:27:32:BMC:storage:-: SLOT-5: State change on PD 27(e0x12/s11) from UNCONFIGURED_GOOD(0) to HOT SPARE(2)

Thanks,

Amir

[Keny Perez]

Amir,

That's very interesting... it would be worth to check the LSI user's guide to see what is expected to see.

-Kenny