Cannot expand RAID 6 array on P824i-p controller

[dpevunov]

Hello!

We have ProLiant DL580 Gen10 server with P824i-p MR  Gen10 controller (870658-B21).  There are 2 logical drives configured: 

Array 1: 2x480 Gb SATA SSD disks RAID 1

Array 2: 12x1920 Gb SATA SSD disks RAID 6

We want to expand Array 2 by 2x1920 Gb SATA SSD disks. There is no option "Expand" in MR Storage Administrator. Could you please help us to resolve the problem?

[psingh]

hello,

Check the below links this may help you to understand the controller.

https://support.hpe.com/hpesc/public/docDisplay?docId=a00053621en_us&docLocale=en_US

https://support.hpe.com/hpesc/public/docDisplay?docId=a00048927en_us&docLocale=en_US

[dpevunov]

Hello Praveen,

Thanks. But this doesn't help to resolve the issue. I've read this documentation.

From the documentation https://support.hpe.com/hpesc/public/docDisplay?docLocale=en_US&docId=a00053315en_us

Expand array

Increase the capacity of an existing array by adding currently existing unassigned drives to it. Any drive that you want to add must meet the following criteria:

•It must be an unassigned drive.

•It must be of the same type as existing drives in the array (for example, SAS HDD, SAS SSD, SATA HDD, or SATAHDD).

•It must have a capacity no less than the capacity of the smallest drive in the array.

This operation uses the Modify Array option in the HPE MR Storage Administrator user interface. This feature is supported when there is a single logical drive configured in the array.

We have a single logical drive configured per array (2 arrays with 1 logical drive in each).  According to the documentation we should be able to expand it.

Extend logical drive

The logical drive extension feature allows the capacity of a logical drive to be expanded by using unused space on existing disks, without requiring a reboot. Increase the capacity of an existing logical drive by specifying a new size. Once the task is performed, use operating system partitioning software to take advantage of the extended space available. Enable this feature in the HPE MR Storage Administrator user interface using the Expand option within the Logical DriveActions menu.

[Cali]

I'm not sure about the MR Controller and I see no Notice in the Manual, but the SR Controller needs a working Battery Cache Module for Raid transformation.
Can you check this?
Here is a Picture, how it should look like:

https://support.hpe.com/hpesc/public/docDisplay?docLocale=en_US&docId=a00051942en_us

[psingh]

Hello,

 can you please share the screenshot of what exact error you are facing during the procedure?

or Controller is not allowing you to extend it?

[dpevunov]

Hi Cali,

We have HPE Smart Storage Battery

[dpevunov]

Hi psingh, 

there is no error. We have 1 unconfigured good drive and cannot extend the Array 1 with it.

[Torsten.]

You cannot extend a RAID1 by 1 drive, you need 2 drives.

[dpevunov]

Hi Torsten,

I'm trying to expand Array 1 which is RAID6. Please have a look at the screenshot I posted earlier.

[Torsten.]

Ah, OK, you wrote

Array 1: 2x480 Gb SATA SSD disks RAID 1

Array 2: 12x1920 Gb SATA SSD disks RAID 6

but the controller is talking about array_0 and array_1 ...

[duckiu]

hi ，

P824 cannot be expanded in the BIOS，But you can do this under MR Storage Administrator, I've done it successfully with windows.

It's actually migrating first and adding the new hard drive during the migration.

[dpevunov]

Hi dickiu,

I'm trying to do this through MR Storage Administrator. I posted screenshots from it.

[dpevunov]

Hello everybody!

Wanted to share with the others the story end.

We were able to add disks to RAID 6 using "Modify Array" procedure, described in 15.3.1.1 (https://support.hpe.com/hpesc/public/docDisplay?docLocale=en_US&docId=emr_na-a00095199en_us). The transformation took about 3 weeks. During this period we experienced big problems with I/O, descpite the fact that the transformation rate was set to 15% during business hours. You can find more info about setting the transformation priority here https://support.hpe.com/hpesc/public/docDisplay?docLocale=en_US&docId=emr_na-a00053315en_us.

We had to stop ALL I/O and set the transformation rate, BGI rate and rebuild rate to 100% during 2 weekends to speed-up the transformation process.  

We added 2 disks to RAID6 but not simultaneously. We did it 1 by 1 because it was first time and we didn't know what to expect. I think the process could take 2 weeks instead of 3 if we add 2 disks at once. 

And the most interesting what we've got in the end. The transformation finished. We rebooted the server and tryed to resize FS to the new size using resize2fs. The resize was successfull but we got errors during FS mount:

kernel: EXT-fs error (device sdb): ext4_mb_generate_buddy:757: group 266, block bitmap and bg descriptor incossistent: 352 vs 0 free clusters

EXT4-fs (sdb): ext4_check_descriptors: Block bitmap for group 0 overlaps block group descriptors

We were able to repair the superblock of the FS using e2fsck and to mount it but later this day we've got block corruption and the server began to reboot in a cycle. We stopped this only by disabling the resized FS in fstab. Then we switched the production database to the standby server.

We managed to repair the FS only by recreating RAID and loosing ALL PRODUCTION DATA on it. Now it is working OK without any corruptions. 

During all this hell we opened a support case with HPE. But we didn't get much help from them. The most interesing info that we've got from them is that this controller is based on the Broadcom chip and competelly differs from standard НРЕ Smart Array controlers. Any technical nuances of this controller should be found on the Broadcom site https://www.broadcom.com/support/knowledgebase/1211161503168/how-to-expand-an-array-on-megaraid-controllers-adding-drives-to-

So, think twice before bying and using this controller in the production environment!