hello,

I'm currently experiencing a RAID 1 installation problem under linux. Distribution is centOS 4.2.
Hardware is 2 same seagate harddrives of 160GB and a Serial ATA FastTrak S150 TX4 (IDE) card of Promise Technology brand. The card h
as 4 S-ATA internals ports (2 are used with the 2 drives).

The install stage is the following :
- first, I connect only ONE drive on FastTrak card and install centOS distrib.
- (the boot on this disk goes correctly)
- secondly, I connect the second drive on board.
- I configure the RAID1 (mirrored) in BIOS of FastTrak card. -> The 1st drive is byte per byte copied on 2nd drive (take a lot of t
ime). If I deconnect the 1st drive, the boot on 2nd drive goes correctly.
- If I connect 2 drives, the RAID is functional :

FastTrak S150 TX4 (tm) BIOS Version ...

ID MODE SIZE TRACK-MAPPING STATUS
=======================================================
1 1x2 Mirror 159932M 19444/255/63 Functional

Press <Ctrl-F> to enter FastBuild (tm) Utility...


Grub splash screen appear and the boot on CentOS goes correctly at the beginning.


ata3: SATA max UDMA/133 cmd 0xF8802300 ctl 0xF8802338 bmdma 0x0 irq 209
ata4: SATA max UDMA/133 cmd 0xF8802380 ctl 0xF88023B8 bmdma 0x0 irq 209
ata1: dev 0 ATA, max UDMA/133, 312581808 sectors: lba48
ata1: dev 0 configured for UDMA/133
scsi0 : sata_promise
ata2: dev 0 ATA, max UDMA/133, 312581808 sectors: lba48
ata2: dev 0 configured for UDMA/133
scsi1 : sata_promise
ata3: no device found (phy stat 00000000)
scsi2 : sata_promise
ata4: no device found (phy stat 00000000)
scsi3 : sata_promise
Vendor: ATA Model: ST3160023AS Rev: 3.00
Type: Direct-Access ANSI SCSI revision: 05
SCSI device sda: 312581808 512-byte hdwr sectors (160042 MB)
SCSI device sda: drive cache: write back
sda: sda1 sda2
Attached scsi disk sda at scsi0, channel 0, id 0, lun 0
Vendor: ATA Model: ST3160023AS Rev: 3.00
Type: Direct-Access ANSI SCSI revision: 05
SCSI device sda: 312581808 512-byte hdwr sectors (160042 MB)
SCSI device sda: drive cache: write back
sdb: sdb1 sdb2
Attached scsi disk sdb at scsi1, channel 0, id 0, lun 0

[...]
EXT3-fs: mounted fs with ordered data mode
[...]
SELinux: Completing initialization.
SELinux: Setting up existing superblocks.
SELinux: initialized (dev sda1, type ext3), uses xattr
SELinux: initialized (dev tmpfs, type tmpfs), uses transition SIDs
[...]


But after put error messages that fs is readonly :

Verification du système de fichiers racine
/: clean, 29956/19283968 files, 805308/38564024 blocks [OK]
Remontage du système de fichiers racine en mode lecture-écriture : mount: LABEL=/ en double - n'a pas été monté
[ECHOUE]

And the startup is quickly stopped because fs is read-only. (nothing can be writed).

- With boot from linux rescue, I can see the card is goodly reconizedi. The 2 SCSI 'sata_promise' and 'ata_piix' drivers are loaded.
- Partition organisation is as following :

/dev/sda1 Id=83 Linux
/dev/sda2 Id=82 Linux swap

/dev/sdb1 Id=83 Linux
/dev/sdb2 Id=82 Linux swap

- Grub configuration from /dev/sda is as following :

/boot/grub/grub.conf file :

default=0
title CentOS-4 i386 (2.6.9-22.ELsmp)
root (hd0,0)
kernel /boot/vmlinuz-2.6.9-22.ELsmp ro root=LABEL=/
initrd /boot/initrd-2.6.9-22.ELsmp.img
title CentOS-4 i386-up (2.6.9-22.EL)
root (hd0,0)
kernel /boot/vmlinuz-2.6.9-22.EL ro root=LABEL=/
initrd /boot/initrd-2.6.9-22.EL.img


'menu.lst' is a link on './grub.conf'.

/boot/grub/device.map file :
(fd0) /dev/fd0
(hd0) /dev/sda

Grub config from /dev/sdb is exactly the same. (I volontary not modified sda in sdb for obvious reasons -> RAID 1 mode).


hdparm /dev/sda or sdb give following error message :
HDIO_GET_MULTCOUNT failed: Inappropriate ioctl for device
[...]

I see in online ressource that FastTrak is no fully hardware RAID 1 (there a software part done by driver). Is it a problem ?

In boot linux rescue mode, I can see following logs :
md: Autodetecting RAID arrays.
md: autorun ...
md: ... autorun DONE.


Thanks in advance for your help.


Before send this message, I test another install stage with disabling SELinux. And it seems to work (the startup). But I haven't tested RAID. In FastBuild utility (BIOS of RAID card), I'm currently rebuilding the RAID array (it takes a lot of time). I'll give feed back any result in this thread. But, in waiting, I'm asking the best way to deals with linux distrib on a hardware RAID 1 . Am I forced to use software RAID 1 of linux with 'mdadm' utility ? Or is it possible to have a only hardware RAID-1 like 2 drives seen inone by the OS ?

hello,

As writed in my previous message, I can give feed back about this new install stage :

If I boot the server with the 2 drives donnected, the following message appears :

Checking root filesystem
/: clean, 27191/19283968 files, 792349/38564024 blocks [ OK ]
Remouting root filesystem in read-write mode: mount: LABEL=/ duplicate - not mounted
[ FAILED ]
ERROR: dos: reading /dev/mapper/pdc_jihabdffb[No such file or directory]
Setting up Logical Volume Management: [ OK ]
mount: LABEL=/ duplicate - not mounted
Checking filesystems
Mounting local filesystems: mount: none already mounted or /dev/pts busy
[...]
-> and a lot of error messages with read-only fs problem


So I'm unabled to do anything. This read-only error is still remaining.

If I boot with one drive, I can boot without errors.
cat /proc/mdstat give :
Personalities :
unused devices: <none>

I read in http://www.tldp.org/FAQ/pdf/Linux-RAID-FAQ.pdf that previous lines seems to determine that current kernel have raid directly compiled in kernel.

I'm asking about this error of read-only booted filesystem when I boot with 2 drives connected. Is it possible that loader 'grub' is bad configured ?

Is it possible to do only hardware RAID1 in linux ? And with this FastTrak S150 TX4 card ? (I'll ask Promise Inc about this, but if someone has feed back about this, I'm very interested)

Any help would be appreciate.

I haven't tried hardware raid yet.

The error message means the "mount -a" command by rc.sysinit does find both partitions as single separate filesystems that have the same e2label="/", while they where supposed to appear (via the driver?) as one single raid1 partition.

The "Raid controller" is not raiding the drives!

It is clear that the disks appear to the kernel as sda and sdb. Therefore to experience raid, you should configure software raid.

Although the mdadm tool makes it nowadays easy to configure raid from the command line, I would suggest that if you don't have valuable data on those partitions yet, you blow them over and use the anaconda redhat installer to create software raid on those partitions.