raid-5 simulators?

eantoranz · 05-02-2005, 11:03 AM

Are there raid-5 simulators?

We are having problems with a netraid-4m.

All the lights are red, however, when I modprobe the aacraid module, I get thsi messages in dmesg tsaying there are read/write errors.

I'm thinking that it could be possible to get the binary data from the separate disks (in case they are all OK) and use a application that could be simulate the array in order to get the data from it. Is that possible?

Matir · 05-03-2005, 07:58 PM

If, somehow, the software raid algorithms match the hardware raid algorithms, then it should be possible. However, this is unlikely: unfortunately, you'll probably need the real deal.

eantoranz · 05-04-2005, 08:25 AM

I thought the algorithms are standard, aren't they?

eantoranz · 05-04-2005, 08:28 AM

By the way..... once the IT personnel allow me to make the images of the HD that make the array, I'll post you to tell you about my journey. I will try with a software raid.

Matir · 05-04-2005, 08:44 AM

The general algorithm is, but the parity calculations, disk selection, etc. can vary between implementations.

eantoranz · 05-04-2005, 08:47 AM

Oh... I see. I guess I'll have to go to church (last time I did was about 20 years ago

) before trying.

eantoranz · 05-04-2005, 03:44 PM

Told you I was gonna keep you posted.

The guys of IT decided to give some tries yesterday with another guy they looked for. They took out one of the disks from the array. The things they tried, didn't work.

Now, I was able to get the re-initialized disk taken out from the array... and just as one test, I got it on another machine to see if I could get the binary output from the disk. I wasn't.

Here's the output from knoppix once I modprobe the module of the controller:

Code:

root@ttyp1[knoppix]# modprobe megaraid
root@ttyp1[knoppix]# dmesg
megaraid: v1.18k (Release Date: Thu Aug 28 10:05:11 EDT 2003)
PCI: Found IRQ 9 for device 04:03.1
IRQ routing conflict for 04:03.1, have irq 11, want irq 9
megaraid: found 0x8086:0x1960:idx 0:bus 4:slot 3:func 1
scsi0 : Found a MegaRAID controller at 0xf8900000, IRQ: 11
megaraid: [E :B ] detected 2 logical drives
megaraid: channel[1] is raid.
megaraid: channel[2] is raid.
scsi0 : LSI Logic MegaRAID E  254 commands 15 targs 5 chans 7 luns
scsi0: scanning virtual channel 0 for logical drives.
  Vendor: MegaRAID  Model: LD0 RAID1  8677R  Rev:   E
  Type:   Direct-Access                      ANSI SCSI revision: 02
  Vendor: MegaRAID  Model: LD1 RAID0 17365R  Rev:   E
  Type:   Direct-Access                      ANSI SCSI revision: 02
scsi0: scanning physical channel 0 for devices.
  Vendor: HP        Model: SAFTE; U160/M BP  Rev: 1023
  Type:   Processor                          ANSI SCSI revision: 02
scsi0: scanning physical channel 1 for devices.
Attached scsi disk sda at scsi0, channel 0, id 0, lun 0
Attached scsi disk sdb at scsi0, channel 0, id 1, lun 0
SCSI device sda: 17770496 512-byte hdwr sectors (9098 MB)
 sda: sda1
SCSI device sdb: 35563520 512-byte hdwr sectors (18209 MB)
 sdb:SCSI disk error : host 0 channel 0 id 1 lun 0 return code = 25040001
 I/O error: dev 08:10, sector 0
SCSI disk error : host 0 channel 0 id 1 lun 0 return code = 25040001
 I/O error: dev 08:10, sector 2
SCSI disk error : host 0 channel 0 id 1 lun 0 return code = 25040001
 I/O error: dev 08:10, sector 4
SCSI disk error : host 0 channel 0 id 1 lun 0 return code = 25040001
 I/O error: dev 08:10, sector 6
SCSI disk error : host 0 channel 0 id 1 lun 0 return code = 25040001
 I/O error: dev 08:10, sector 0
SCSI disk error : host 0 channel 0 id 1 lun 0 return code = 25040001
 I/O error: dev 08:10, sector 2
SCSI disk error : host 0 channel 0 id 1 lun 0 return code = 25040001
 I/O error: dev 08:10, sector 4
SCSI disk error : host 0 channel 0 id 1 lun 0 return code = 25040001
 I/O error: dev 08:10, sector 6
 unable to read partition table

Now... I see that it recognizes the local array of the computer... no problem. But it has problems with the disk from the array.

I can't cat /dev/sdb (so to make one image). Do you know how I can make it work? I don't care about the "logical" data. I just want to get the raw data from the disk (by the way.... I don't think the disk is dead at all).

Code:

root@ttyp1[knoppix]# cat /dev/sdb
cat: /dev/sdb: Error de entrada/salida

Translation: I/O error

Any ideas?

eantoranz · 05-05-2005, 08:53 PM

OK.... I'll get you updated:

Remember I told you there were 5 disks in the array? Welll.... one of the disks was removed from the array while making tests on it. The tricks didn't work, so I was finally given the chance to get the images of the HD in order to experiment with it... and I soon found myself booting into knoppix.

We turned off the storage and moved the 4 remaining disks into a computer to handle them locally, get the images and take them back so they can keep trying to get the data.

I left the computer at the office copying each the the five disks to another computer so we can keep them untouched.

I took a look at some of the binary information on the disks and there "seems" to be information to rescue.

Now.... let's think about tomorrow:

I'll have 4 files: disk1.dat disk2.dat disk3.dat and disk4.dat.

Let's think about the worst case scenario: I don't know which of the five disks was removed (i'll ask the guys tomorrow anyway).

What will I have to write in raidtab so mkraid notices that one of the images is missing? I can make a number of tries..... no problem with that.

What considerations should I take?

Remember.... tomorrow IS the day!

Matir · 05-05-2005, 09:08 PM

Hrrrm... I think there is a "missing" directive in raidtab... try 'man raidtab' for the format.

eantoranz · 05-05-2005, 09:15 PM

...and then you say you are not a guru! Ha!

Matir · 05-05-2005, 09:26 PM

If I were a guru, I wouldn't be posting questions here, as I have before... though somehow, they never get answered.

eantoranz · 05-06-2005, 02:41 PM

OK... Here we go.

First attempt:

raidtab:

Code:

raiddev /dev/md0
        raid-level              5
        nr-raid-disks           5
        nr-spare-disks          0
        chunk-size              64k

        device                  /mnt/hdb1/discoa.dat
        raid-disk               0
        device                  /mnt/hdb1/discob.dat
        raid-disk               1
        device                  /mnt/hdb1/discoc.dat
        raid-disk               2
        device                  /mnt/hdb1/discod.dat
        raid-disk               3
        device                  /dev/zero
        failed-disk             4

It failded:

Code:

root@ttyp5[knoppix]# raidstart /dev/md0
/dev/md0: Invalid argument
/dev/md0: Invalid argument
/dev/md0: Invalid argument
/dev/md0: Invalid argument
/dev/md0: Invalid argument
/dev/md0: Invalid argument

and the dmesg:

Code:

md: could not lock [dev 00:00], zero-size? Marking faulty.
md: could not import [dev 00:00]!
md: autostart [dev 00:00] failed!
md: could not lock [dev 00:00], zero-size? Marking faulty.
md: could not import [dev 00:00]!
md: autostart [dev 00:00] failed!
md: could not lock [dev 00:00], zero-size? Marking faulty.
md: could not import [dev 00:00]!
md: autostart [dev 00:00] failed!
md: could not lock [dev 00:00], zero-size? Marking faulty.
md: could not import [dev 00:00]!
md: autostart [dev 00:00] failed!
 [events: 00000000]
md: invalid raid superblock magic on [dev 01:05]
md: [dev 01:05] has invalid sb, not importing!
md: could not import [dev 01:05]!
md: autostart [dev 01:05] failed!

This is the ls -l of that directory:

Code:

-rw-r--r--  1 root root 18208522240 2005-05-06 14:09 discoa.dat
-rw-r--r--  1 root root 18208522240 2005-05-06 14:35 discob.dat
-rw-r--r--  1 root root 18208522240 2005-05-06 14:57 discoc.dat
-rw-r--r--  1 root root 18208522240 2005-05-06 15:21 discod.dat

What could I try next?

Could it be that I'm skipping the algorithm parameter?

eantoranz · 05-06-2005, 02:47 PM

I deleted the device /dev/zero from the file. Now I have ONE less invalid argument mesage.

What's next?

Electro · 05-07-2005, 02:02 AM

Use losetup to create four loop devices. You will need to move to a certain point in the file by specifying "-o" because the partition starts somewhere in the file. Partitions do not start at the beginning of any medium. The beginning of the medium contains mbr.

For more information, go to http://www.mega-tokyo.com/osfaq2/ind...0Under%20Linux.

I do not think that RAID 5 algorithms are standard because moving the array from one computer to another computer using a different controller will not work. All controller brands state this in the manuals.

BTW, I have never read using disk images to be used as RAID.

eantoranz · 05-07-2005, 09:27 AM

well.... I just read it. I'll give it a shot and tell you what happens!

Stay tuned!