cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
2482
Views
0
Helpful
4
Replies

disk policy: WriteBack with "Cisco 12G SAS Modular Raid Controller" and CentOS

support
Level 1
Level 1

Hello Community Members,

 

Server: Cisco UCS 220 M4

SAS Controller: "Cisco 12G SAS Modular Raid Controller"

SAS Controller Firmware: 24.12.1-0456

Virtual Disk: 1.) RAID-0 with 1 no. of SSD-SAS disk of Capacity 1.8 TB (Linux: /dev/sdb)

                   2.) RAID-0 with 5 nos. of SAS Disks of capacity 2.2 TB each (Linux: /dev/sdc)

 

both virtual disks are similarly configured with parameters like:

Access Policy: ReadWrite

Read Policy: No ReadAhead

Cache Policy: DirectIO

Disk Cache Policy: UnChanged

Write Policy: WriteBack with Good BBU

 

Problem:

Linux OSes like CentOS-7 and Ubuntu-20.04 was not able to detect the disk policy of those Virtual Disks as WriteBack, instead detected them as WriteThrough)

 

Output from Linux system:

 

[root@h121 log]# hdparm -W /dev/sdb

/dev/sdb:
SG_IO: bad/missing sense data, sb[]:  70 00 05 00 00 00 00 0d 00 00 00 00 20 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 write-caching = not supported
[root@h121 log]#
[root@h121 log]# hdparm -W /dev/sdc

/dev/sdc:
SG_IO: bad/missing sense data, sb[]:  70 00 05 00 00 00 00 0d 00 00 00 00 20 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 write-caching = not supported
[root@h121 log]#

 

 

 

[root@h121 log]# cat /sys/block/sdb/device/scsi_disk/0\:2\:1\:0/cache_type
write through
[root@h121 log]#
[root@h121 log]# cat /sys/block/sdc/device/scsi_disk/0\:2\:2\:0/cache_type
write through
[root@h121 log]#

 

 

But, at the same time, MS Windows 2019 detects them properly with disk policy: WriteBack - confirmed using LSI MegaRaid Manager

 

Is there any additional configuration required either with Virtual Disk properties or with Linux OS. Need help to resolve this problem.

 

Thanks

SG

4 Replies 4

Kirk J
Cisco Employee
Cisco Employee

I wonder if the linux command is expecting a local drive's internal write caching parameter (which is normally disabled) vs the raid controller's virtual drive write-back caching.  I can understand why the windows megaraid manager is correctly returning the write-back caching enabled as that knows how to check the VD's write-back caching state.

Usually, for the raid volumes, you need to have something like storcli installed, to correctly monitor/query the virtual or physical drives behind a raid controller.  Not all linux disk utilities work correctly against raid controller based drives.  smartctl is a good example of that.

 

Kirk...

Hello Kirk J

 

Thanks for your response.

as suggested, will check the cache status of the physical disks using storcli or mega64 tools

 

This controller has the following hardware specs

Flash Present: true
Flash Size: 32 MB
Memory Size: 512 MB
Cache Memory Size: 370 MB

 

Do we need to upgrade the cache memory, does it play a role in our problem.

 

Thanks / SG

So the policy of 'writeback with good BBU' means writeback caching only is enabled if a BBU/Supercap is present, and has passed all the learning tests.

The 12Gb SAS controllers for the M4s have the following options:

Flash-Backed Write Cache (FBWC) Upgrade Options for Cisco 12G SAS Modular RAID controller
UCSC-MRAID12G-512 512 MB FBWC, which includes a 512 MB memory plus a SuperCap for data cache
power backup for the Cisco 12G SAS Modular RAID controller. Supports JBOD,
RAID 0, 1, 5, and 50.
UCSC-MRAID12G-1GB 1 GB FBWC, which includes a 1 GB MB memory plus a SuperCap for write cache
backup. Supports JBOD, RAID 0, 1, 10, 5, 6, 50, and 60.
UCSC-MRAID12G-2GB 2 GB FBWC, which includes a 2 GB MB memory plus a SuperCap for write cache
backup. Supports JBOD, RAID 0, 1, 10, 5, 6, 50, and 60.
UCSC-MRAID12G-4GB 4 GB FBWC, which includes a 4 GB MB memory plus a SuperCap for write cache
backup. Supports JBOD, RAID 0, 1, 10, 5, 6, 50, and 60.

 

Your BBU/Supercap needs to be in good state in order for write back to be enabled.

The CIMC, storage tab should have a BBU subtab that will indicate it's status.

 

Kirk...

Hello Kirk J

 

Yes, we have good battery and it is green state in CIMC storage page

 

Model we have:

UCSC-MRAID12G-512 512 MB FBWC, which includes a 512 MB memory plus a SuperCap for data cache
power backup for the Cisco 12G SAS Modular RAID controller. Supports JBOD, RAID 0, 1, 5, and 50.

 

And as mentioned earlier, Disk Cache status are enabled and confirmed in the BIOS configuration pages of this controller and under physical drive information page

 

Thanks / SG

 

Review Cisco Networking for a $25 gift card

Review Cisco Networking for a $25 gift card