I have a perc5i (and perc5e) raid controller in my poweredge 2970
running centos 5.5. The RAID controller config is set for write-back
policy, which I can confirm by checking in OpenManage. I also called
Dell and ran their reporting application and had them double check to
make sure everything looks OK hardware wise and they confirm that it does

Yet when CentOS boots is appears to use the write through policy as can
be seen below from dmesg output. Am I wrong that CentOS should query the
RAID controller and go into the correct write mode: write-back?

hub 1-5:1.0: 4 ports detected
Vendor: DP Model: BACKPLANE Rev: 1.00
Type: Enclosure ANSI SCSI revision: 05
Vendor: DELL Model: PERC 5/i Rev: 1.03
Type: Direct-Access ANSI SCSI revision: 05
SCSI device sda: 2435317760 512-byte hdwr sectors (1246883 MB)
sda: Write Protect is off
sda: Mode Sense: 1f 00 00 08
SCSI device sda: drive cache: write through
SCSI device sda: 2435317760 512-byte hdwr sectors (1246883 MB)
sda: Write Protect is off
sda: Mode Sense: 1f 00 00 08
SCSI device sda: drive cache: write through
sda: sda1 sda2 sda3
sd 0:2:0:0: Attached scsi disk sda
megasas: 0x1028:0x0015:0x1028:0x1f01: bus 16:slot 14:func 0
GSI 21 sharing vector 0x62 and IRQ 21
ACPI: PCI Interrupt 0000:10:0e.0[A] -> GSI 18 (level, low) -> IRQ 98
megasas: FW now in Ready state

_______________________________________________
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos

Can anyone with write-back cache working just confirm that they see some
sort of indication of write-back in dmesg?

As I mentioned I talked with Dell and they claim it is working, however
I spoke with VMWare and they said if it isn't indicated in dmesg, then
it isn't working. It would just be good to know it is indicated as
functional in this way for others so I can have something to go on with
Dell.

On 2/11/2011 10:00 AM, James Chase wrote:
> I have a perc5i (and perc5e) raid controller in my poweredge 2970
> running centos 5.5. The RAID controller config is set for write-back
> policy, which I can confirm by checking in OpenManage. I also called
> Dell and ran their reporting application and had them double check to
> make sure everything looks OK hardware wise and they confirm that it does
>
> Yet when CentOS boots is appears to use the write through policy as can
> be seen below from dmesg output. Am I wrong that CentOS should query the
> RAID controller and go into the correct write mode: write-back?
>
> hub 1-5:1.0: 4 ports detected
> Vendor: DP Model: BACKPLANE Rev: 1.00
> Type: Enclosure ANSI SCSI revision: 05
> Vendor: DELL Model: PERC 5/i Rev: 1.03
> Type: Direct-Access ANSI SCSI revision: 05
> SCSI device sda: 2435317760 512-byte hdwr sectors (1246883 MB)
> sda: Write Protect is off
> sda: Mode Sense: 1f 00 00 08
> SCSI device sda: drive cache: write through
> SCSI device sda: 2435317760 512-byte hdwr sectors (1246883 MB)
> sda: Write Protect is off
> sda: Mode Sense: 1f 00 00 08
> SCSI device sda: drive cache: write through
> sda: sda1 sda2 sda3
> sd 0:2:0:0: Attached scsi disk sda
> megasas: 0x1028:0x0015:0x1028:0x1f01: bus 16:slot 14:func 0
> GSI 21 sharing vector 0x62 and IRQ 21
> ACPI: PCI Interrupt 0000:10:0e.0[A] -> GSI 18 (level, low) -> IRQ 98
> megasas: FW now in Ready state
>
> _______________________________________________
> CentOS mailing list
> CentOS@centos.org
> http://lists.centos.org/mailman/listinfo/centos

_______________________________________________
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos

on 2/16/2011 5:31 AM James Chase spake the following:
> Can anyone with write-back cache working just confirm that they see some
> sort of indication of write-back in dmesg?
>
SCSI device sda: drive cache: write back, no read (daft)
SCSI device sda: drive cache: write back, no read (daft)
SCSI device sdb: drive cache: write through
SCSI device sdb: drive cache: write through

But not a PERC... Maybe your driver is not reporting back properly?


_______________________________________________
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos

On Wed, Feb 16, 2011 at 08:31:05AM -0500, James Chase wrote:
> Can anyone with write-back cache working just confirm that they see some
> sort of indication of write-back in dmesg?
>
> As I mentioned I talked with Dell and they claim it is working, however
> I spoke with VMWare and they said if it isn't indicated in dmesg, then
> it isn't working. It would just be good to know it is indicated as
> functional in this way for others so I can have something to go on with
> Dell.

You can verify your cache policy with MegaCLI tool. Check your output
from `MegaCli -CfgDsply -a0' command.

--
Dominik Zyla

_______________________________________________
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos

On Wed, Feb 16, 2011 at 9:31 PM, James Chase <james@wintercastle.net> wrote:
> Can anyone with write-back cache working just confirm that they see some
> sort of indication of write-back in dmesg?

I thought this thing is really masked from the OS and handled by the
raid controller BIOS completely??
_______________________________________________
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos

> You can verify your cache policy with MegaCLI tool. Check your output
> from `MegaCli -CfgDsply -a0' command.
>
What a cool tool. It looks like it should be active according to
MegaCli, though strange that Scott sees his drives in write-back mode
via CentOS and I do not and I also have these horrific NFS speeds via
VMWare that so many people have found to be caused by write-back not
being active...

I don't know what the answer is to this question. I went with iSCSI
instead of NFS, the benefits being that it trounces my NFS setup
performance wise, and seems stable despite warnings of high I/O causing
VMWare to drop the LUN (I tested with bonnie++ as it seems this is an
effective tool in getting the LUN to drop in VMWare).

Product Name: PERC 5/i Integrated

Memory: 256MB

BBU: Present
...

Default Cache Policy: WriteBack, ReadAheadNone, Direct, No Write Cache if Bad BBU

Current Cache Policy: WriteBack, ReadAheadNone, Direct, No Write Cache if Bad BBU

Access Policy: Read/Write

Disk Cache Policy: Disk's Default


_______________________________________________
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos