Building Software Raid.

Home » CentOS » Building Software Raid.
CentOS 9 Comments

dear All,

I’m spending the afternoon trying to build software raid using this (
excellent ) guide :

I’m doing this not only because this machine has no raid controller – I
could go out and buy one – but also because I hope software raid will provide easier monitoring.

This is a newly installed, updated CentOS6.4 standard machine.

So far I arrived at part1 step 4:

mdadm –add /dev/md_d0 /dev/sdb

and this is on 16.6 %

Looking at part 2 , step 2 :

————> Edit /etc/fstab, you must change all mounts from using LABEL= to explicit device names, like /dev/md_d0p1, /dev/md_d0p2, …

I’m wondering where I can find the names md_d0p1, p2 and so on.

Can anyone help me on this?

Thanks very much.

Greetings, J.

9 thoughts on - Building Software Raid.

  • Make sure to set MAILADDR in /etc/mdadm.conf I generally use a system account (generally root which is aliased to a proper address in /etc/aliases – don’t forget to run newaliases if you change /etc/aliases).

    * I should probably edit that wiki page and these few notes. ;)

    When you set up the software raid arrays, you’d be setting the device names.

  • Op 30-10-13 17:01, SilverTip257 schreef:

    Hello All,

    maybe something went wrong.

    The optional step 4 is finished. I now have

    /cat /proc/mdstat//
    //Personalities : (raid0) ( raid1 ) enz….//
    //md_d0 : active raid1 sdb (1) sda (0)//
    // 488386496 blocks (2/2) ( uu )//
    //unused devices : none/

    that’s it. no p1, p2 names. Is this because I used lvm?

    Greetings and thanks, J.

  • Op 30-10-13 17:11, Johan Vermeulen schreef:

    some additional info:

    I used the standard ( lvm ) partitioning when installing, than, after reboot, used system-config-lvm to shrink the last lvm.

    no I see I have /dev/md/md_d0 /dev/md_d0p1 /dev/md_d0p2
    I don’t know what is what….

    greetings, J.

  • From: Johan Vermeulen

    Not sure what you did with lvm but I would say that md_d0 is the (raid) device and md_d0p* are its partitions… Just do fdisk -l /dev/md/md_d0 (or gdisk).


  • Op 30-10-13 17:38, John Doe schreef:

    Hello All,

    fdisk -l was what I needed to bring clarity in the name-issue.

    By now I tried this on a just-installed CentOS6.4, updated machine, this time with partitions instead of lvm.

    I was able to follow the wiki all the way without EM
    /etc/fstab now reads

    //dev/md_d0p3 root
    /dev/md_d0p1 home /

    and so on.

    I adapted /etc/grub.conf to /root=/dev/md_d0p3/ and deleted the rd_NO_MD
    and ran the dracut commands.

    BUT when rebooting I get :
    /dracut warning : No root device “block:/dev/md_d0p” found./

    searching through all the files, I see that /dev/mapper only contains control in stead of the usual links.

    Could this be the issue?
    And if it is, how can I generate or manualy add the usual links in this file?

    Greetings, J.

  • Op 01-11-13 17:41, Johan Vermeulen schreef:

    /Dear All,

    I would like to renew my request for help with this wiki:

    Still stuck on EM : dracut warning: No root device “block:/dev/md_dop3”
    found. I think this is a bit of a pain to troubleshoot, because

    * to change anything, I have to chroot to /mnt/sysimage every time.
    * googling this, I see a a threat marked solved here :

    but I can’t access it.

    So far I tried this :


    so i ran dracut –mdadmconf initramfs…….

    but that did change anything

    also mentions an almost empty /dev/mapper, so I am now trying to make a correct symlink in /dev/mapper in order to be able to boot the system.

    So far I haven’t found the right one.

    Can anyone shed some light on this?

    Many thanks.


    < DYQFjAB&

  • There used to be another RAID HOWTO on the CentOS Wiki, I have a printout of it here, somewhere,… I used it when I built a new box a few years ago on CentOS-5 and it worked fine. where “fine” means when one of the drives gave some sort of problem it got ejected from the RAID array and the system continued to work properly. And replacing it allowd me to rebuild the array and proceed.

    When I did a bare-metal installation of CentOS-6 at the beginning of this year I went looking for the newest version of it, and it no longer seemed to be available at the URL it says it came from. But I used it anyway even though it wasn’t exactly perfect for C6, it was close enough that it wasn’t difficult to figure out what should be done, and I’ve got a nice Raid-1
    system running here now.

    I’d give you the URL it claimed to be from if I could find the printout, and if it turns up soon I’ll come reply to this mail and give it if the URL is in fact working.