sofware raid issue

maestro2912 · 12-04-2009, 03:41 AM

Hi,
I have problem with software RAID on Linux file server. Disks are still 24/7 working, even nobody access to data. I'm afraid that disks won't live a long time if it will continue.

Background:

I want to set up file-server with with 2x HDD 1 TB disks. I decided to use Linux software RAID, as "HW fake-RAID" on motherboard is supported only in Windows (I guess I can't recover data, if one my HDD fails where there is only Linux present. All I have is an BIOS utility, that does not provide recovery functionality).

So I
1) on both disk, removed all partitions and created new one 1TB partition with fd-type
2) erased superblocks
2) add disks together in RAID 1 with mdadm
3) formatted them to ext3 partition (because I need Debian Lenny as OS)
4) waited while they synchronized
5) move data on them

Problem1)
Disks in RAID are all time working, even no one use them. I use no special services, clean Debian Lenny install, with just Desktop & File-server package suites. I also install FTP. Only one process seems to use disks, kjournald, that has three instances.

Is this correct? Will disks live more then just one year with such disk activity?

I have bought disks WD Caviar Green, SATA Hard Drives, 1 TB, 32MB cache
http://www.wdc.com/en/products/products.asp?DriveID=336

Trouble is that they try to go to standby mode (slow rotations, park reading heads) after some short time. But parking heads produce different noise, so this is not an issue.

They are not 100% RAID disks, but they have small power consumption. This was wrong, and if is only this problem, I will buy RAID disks.

Any tips, what could be problem?

Even if I go to init 1, unmount /dev/md0 and stop RAID with mdadm -S /dev/md0, disks are still working.

Some information about system:

Code:

uname -a
Linux zazriva 2.6.26-2-686 #1 SMP Wed Nov 4 20:45:37 UTC 2009 i686 GNU/Linux

# dmidecode -t baseboard
# dmidecode 2.9
SMBIOS 2.4 present.

Handle 0x0003, DMI type 2, 8 bytes
Base Board Information
        Manufacturer: Hewlett-Packard
        Product Name: 0A50h
        Version: Not Specified
        Serial Number: HUB7020B6Q

Problem 2)
Small problem, fdisk is complaining about wrong partition table ?!
Disks are absolutely new, I guess I didn't do nothing wrong and I was trying a lot of things to work it out, and nothing solved it, so I ignore this. I can access data, RAID is writable.

Code:

#fdisk -l

Disk /dev/sdb: 1000.2 GB, 1000204886016 bytes
255 heads, 63 sectors/track, 121601 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Disk identifier: 0x000338db

   Device Boot      Start         End      Blocks   Id  System
/dev/sdb1               1      121601   976760001   fd  Linux raid autodetect

Disk /dev/sdc: 1000.2 GB, 1000204886016 bytes
255 heads, 63 sectors/track, 121601 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Disk identifier: 0x000338db

   Device Boot      Start         End      Blocks   Id  System
/dev/sdc1               1      121601   976760001   fd  Linux raid autodetect

Disk /dev/md0: 1000.2 GB, 1000202174464 bytes
2 heads, 4 sectors/track, 244189984 cylinders
Units = cylinders of 8 * 512 = 4096 bytes
Disk identifier: 0x00000000

Disk /dev/md0 doesn't contain a valid partition table

# fdisk -v
fdisk (util-linux-ng 2.13.1.1)

Thank everybody who respond to my questions.

jhwilliams · 12-05-2009, 12:06 PM

Quote:

Originally Posted by maestro2912

Disks in RAID are all time working, even no one use them. [...] Only one process seems to use disks, kjournald, that has three instances.

That doesn't seem out of line to me, provided it isn't heavy activity. Note that kjournald mostly works with /proc, which isn't actually on disk. But, you may like to add a 'noatime' flag to your fstab: http://ubuntuforums.org/archive/index.php/t-369759.html

Quote:

Is this correct? Will disks live more then just one year with such disk activity?

Almost certainly yes, but you'll have to define "such disk activity." If they're just mulling a few bits here and there, who cares? If you're seeing something actually particularly abnormal, then there might be cause for concern. Regardless, disks rarely actually fail. ("Rare" is subjective.)

Quote:

I have bought disks WD Caviar Green, SATA Hard Drives, 1 TB, 32MB cache
http://www.wdc.com/en/products/products.asp?DriveID=336

Trouble is that they try to go to standby mode (slow rotations, park reading heads) after some short time. But parking heads produce different noise, so this is not an issue.

They are not 100% RAID disks, but they have small power consumption. This was wrong, and if is only this problem, I will buy RAID disks.

None of this seems relevant to me -- there may be disks marketed "for RAID," but really, it doesn't matter.

To summarize, I doubt that you /actually/ have a problem. Can you substantiate the amount of I/O you are seeing with some figures? Regardless, check out that link.