My attempt to understand dmraid

Sean Bruno sean.bruno at dsl-only.net
Thu Oct 11 17:57:39 UTC 2007


I am trying to delve into dmraid/device-mapper and learn how the magic
happens and I came up with what I think is the "way it works."  Please
critique and correct me where I am wrong or over-simplify the operation
of the current dmraid architecture as it is used under Fedora/redhat
systems.

Assumptions:
  dmraid compatible device(sil3118)
  Hard Drives of equal size(SATA in my case)
  Installation of Fedora Core 7(x86_64)

1.
The initial setup occurs on the RAID controller where I selected RAID1
with my two disks.  This initialized the drives for use with
dmraid/device-mapper by placing the SiL metadata on the drives and set
them to a initialized state.

2.
The system boots off the first hard drive(sda) and loads the kernel from
there.

3.
Booting Fedora Core 7, /etc/rc.sysinit loads the fake RAID1 module
"dm-mirror".  This actually works with the software RAID code in md to
manage the RAID1.  The system then detects that the drives are a RAID1
set by invoking '/sbin/dmraid  -ay -i -p -t' This detects the RAID1 set
and creates /dev/mapper/silXXXXXXX through the device-mapper libraries.


What is unclear to me is where the system mounts the filesystems
on /dev/mapper/silXXXXXX and how the dm-mirror module manages drive
failues.  Can someone fill in the pieces for me?

Sean




More information about the Ataraid-list mailing list