AWS: create RAID1+0 disk with four EBS volumes

#RAID under linux summary: http://www.ducea.com/2009/03/08/mdadm-cheat-sheet/
#RAID with EC2: http://bruun.co/2012/06/06/software-raid-on-ec2-with-mdadm
#Why RAID10 instead of RAID0: http://blog.9minutesnooze.com/raid-10-ebs-data/
#EBS durability: http://aws.amazon.com/ebs/ (durability section, between 0.1% and 0.5% Anual Failure Rate)

#Initial situation:
#root: dev/sda1
#ebs volumes (1GBx4): /dev/sdb /dev/sdc /dev/sdd /dev/sde

#Script to create and test RAID10:

#Show devices
ls -l /dev/sd* /dev/xv*

#Avoid race lock
sudo udevadm control --stop-exec-queue
#Create RAID
sudo mdadm --create /dev/md0 --run --level=10 --raid-devices=4 --chunk=64 /dev/sdb /dev/sdc /dev/sdd /dev/sde
sudo udevadm control --start-exec-queue

#Now there's an entry under /dev/md/* with the name of the device, for example /dev/md/ip-10-251-94-61:0
#Use it to update the /etc/fstab file adding a line like this one:
#/dev/md/ip-10-251-94-61:0 /media/ebs xfs defaults 0 0

#prewarm with 0s
sudo time shred -vfz -n 0 /dev/md0

sudo yum install -y xfsprogs

#Format and mount
#BTW, in order to freeze the fs for backups see the xfs_freeze command
sudo mkdir -p /media/ebs
sudo mkfs.xfs /dev/md0 -f
sudo mount -t xfs /dev/md0 /media/ebs
sudo chown ec2-user:ec2-user /media/ebs/

#show available space
df -h

#Get 821.7 megabytes of data:
cd /media/ebs
wget http://dumps.wikimedia.org/eswiki/20120906/eswiki-20120906-abstract.xml
head eswiki-20120906-abstract.xml

#status (U means it's OK, F means it's in failed status)
cat /proc/mdstat
#alternative
mdadm --detail /dev/md0

#mark a volume as failed:
sudo mdadm --fail /dev/md0 /dev/sdc
sudo mdadm --remove /dev/md0 /dev/sdc

#Now is safe to detach de volume (using the ec2 console)

#Check it's still working
cp eswiki-20120906-abstract.xml  eswiki-20120906-abstract2.xml

#Create another volume (or reattach the old one) with the name /dev/sdx

#Add the new volume
sudo mdadm --add /dev/md0 /dev/sdx
#A mdadm --detail should indicate all four disk are available and in clean state after a while

#In case you  cannot access de device after a reboot:
sudo mdadm --stop --scan
sudo mdadm --assemble --scan
sudo mount /dev/md0 ebs