Intel 6300ESB RAID

[tuaris]

I have a Supermicro P4SCi with an embedded Intel 6300ESB RAID controller. I know it's not the greatest option but it's what I have to work with.

In the BIOS the controller is set to RAID mode and I created a mirrored array in the controllers BIOS setup program. The problem is that NAS4Free is seeing two drives instead of one. I was reading I noticed that it's possible that FreeBSD supports this type of embedded control through the "graid" driver (https://forums.freebsd.org/showthread.php?p=195530).

Is it possible to make it work? Or should I just use the FreeBSD software RAID option instead?
Would the FreeBSD software RAID offer better performance and reliability? These are 1TB drives and I would like to store my most precious data on them.

[raulfg3]

Is it possible to make it work? Or should I just use the FreeBSD software RAID option instead?

My suggest is that use soft RAID ALLWAYS, and ZFS for data storage, so in the future if you change your board, your pool still live and do not lose data.

My ZFS RaidZ1 Pool have survive 3 Board Changes and 2 SATa Controolers additions and I don't lose data, and no neeed to save in external disk format and restore.

[tuaris]

Okay I decided to go with FreeBSD's software RAID. Do I leave the BIOS option to RAID mode or change it to SATA mode?

[raulfg3]

Use SATA mode (AHCI if is posible).

PD: Soft Raid can be Raid5 or RaidZ1, I suggest to use raidZ1= ZFS

[tuaris]

I setup the BIOS to use SATA mode, created the RAID mirror using software RAID. Works fine, but now I get the following problem:

Code: Select all

Nov 6 10:04:19 	storage 	kernel: (ada2:ata3:0:0:0): WRITE_DMA48. ACB: 35 00 00 52 e2 40 1b 00 00 00 00 01
Nov 6 10:04:19 	storage 	kernel: (ada2:ata3:0:0:0): CAM status: Command timeout
Nov 6 10:04:19 	storage 	kernel: (ada2:ata3:0:0:0): Retrying command
Nov 6 10:04:19 	storage 	kernel: (ada2:ata3:0:0:0): lost device
Nov 6 10:04:19 	storage 	kernel: GEOM_MIRROR: (pass3:Device Public: provider ada2 disconnected.
Nov 6 10:04:19 	storage 	kernel: ata3:0:0:GEOM_MIRROR0): passdevgonecb: devfs entry is gone
Nov 6 10:04:19 	storage 	kernel: : Device Public: rebuilding provider ada2 stopped.
Nov 6 10:04:20 	storage 	kernel: (ada2:ata3:0:0:0): removing device entry

The drive comes back if I reboot, but then a few hours later it disappears again.
Drives are new, is it a bad controller?

[raulfg3]

NO, is a bad cable or enclosure, use a good SATA cable with metalic clip to retention and do not use external/internal enclosure, connect the disk directly using SATA cable.

http://www.alternate.es/html/product/De ... le/791618/?

[tuaris]

Both Drives are gone now.
The cables are glued on, nothing is loose, this is a 1U rack server with hot swap drive trays.
I am going to switch the controller back to RAID mode, use JBOD, and see if it makes a difference. This server was running Feodora 15 without this problem so I don't think it's hardware related.

Code: Select all

Nov 7 23:47:16 	storage 	kernel: (ada2:ata3:0:0:0): WRITE_DMA. ACB: ca 00 00 a4 70 42 00 00 00 00 00 00
Nov 7 23:47:16 	storage 	kernel: (ada2:ata3:0:0:0): CAM status: Command timeout
Nov 7 23:47:16 	storage 	kernel: (ada2:ata3:0:0:0): Retrying command
Nov 7 23:47:16 	storage 	kernel: (ada2:ata3:0:0:0): lost device
Nov 7 23:47:16 	storage 	kernel: (pass3:ata3:0:0:0): passdevgonecb: devfs entry is gone
Nov 7 23:47:16 	storage 	kernel: GEOM_MIRROR: Synchronization request failed (error=6). ada2[WRITE(offset=20959461376, length=131072)]
Nov 7 23:47:16 	storage 	kernel: GEOM_MIRROR: Device Public: provider ada2 disconnected.
Nov 7 23:47:16 	storage 	kernel: GEOM_MIRROR: Device Public: rebuilding provider ada2 stopped.
Nov 7 23:47:16 	storage 	kernel: (ada2:ata3:0:0:0): removing device entry
Nov 7 23:48:29 	storage 	kernel: (ada1:ata2:0:0:0): READ_DMA48. ACB: 25 00 28 0e a4 40 51 00 00 00 40 00
Nov 7 23:48:29 	storage 	kernel: (ada1:ata2:0:0:0): CAM status: Command timeout
Nov 7 23:48:29 	storage 	kernel: (ada1:ata2:0:0:0): Retrying command
Nov 7 23:48:29 	storage 	kernel: (ada1:ata2:0:0:0): lost device

[raulfg3]

please try to eliminate swap drive trays from equation, do a test conecting drives direct to board without swap drive tray, and chek the log, if still have same errors, you are right, but I have the same errors in the past and the problem was drive trays.

[tuaris]

I will try that.

I noticed that when the drive disappears and the RAID gets degraded Windows fails and I can't access the share via CIFS/SMB anymore. But if I go in through the console or SSH I can access the mount point and the files on the RAID volume. I have to manually restart samba from the command line (/etc/rc.d/samba restart).