How to create a worker image for the Cloud
-
-


System used for this tutorial: Ubuntu 10.10 or Debian 6.0
Must be root


  Create the base image
  =====================

- Create an empty disk image (4GB)
  dd if=/dev/zero of=fs-spequloscw.img bs=1M count=4096

- Make the filesystem:
  /sbin/mkfs.ext3 -F -j fs-spequloscw.img

- Mount the image:
  mkdir fs-spequloscw
  mount -o loop fs-spequloscw.img fs-spequloscw/

- Debootstrap a Debian system:
  debootstrap --arch amd64 lenny fs-spequloscw/

- Mount proc
  mount -t proc proc fs-spequloscw/proc

- Chroot into the new filesystem:
  chroot fs-spequloscw/

- Make the /dev devices:
  cd /dev
  /sbin/MAKEDEV generic

- Configure the network by editing /etc/network/interfaces :
  auto lo
  iface lo inet loopback
  
  auto eth0
  iface eth0 inet dhcp

- Edit /etc/apt/sources.list :
   deb     http://ftp2.us.debian.org/debian lenny main contrib non-free

- Update packages:
  apt-get update
  apt-get upgrade

- Install an SSH server:
  apt-get install openssh-server

- Configure SSH connection:
  mkdir /root/.ssh
  cat <path_to_public_key_id_rsa.pub> > /root/.ssh/authorized_keys

- Change hostname : 
 - Edit /etc/hostname :
  spequloscw

 - Edit /etc/hosts
  127.0.0.1 localhost
  127.0.1.1 spequloscw

- Prepare the image for SpeQuloS:
 - Create the spequlos directory:
  mkdir /root/spequlos

 - To install the XWHEP worker: 
  - Install needed packages:
  apt-get install sun-java6-jre zip uuid-runtime

  - Install the XWHEP worker package:
  wget <address_to_xwhep_worker_package>
  cd /opt
  tar xfvz xwhep.worker.tgz

  - Be sure that XW worker autostart is disabled:
  update-rc.d -f xtremweb.worker remove

  - Check the worker config files, eg:
  cat /opt/xwhep-worker-7.1.0/conf/xtremweb.worker.conf

  - Copy the scripts used by SpeQuloS to start the worker:
You can find an exemple of those scripts inside the DG-plugin/XWHEP/startup_script/ subdirectory of the SpeQuloS package. For SpeQuloS to start the appropriate script, it is mandatory that the script is named <DG_name>.sh, where DG_name is the name of the desktop grid registered to SpeQuloS during the installation (see "Registering a desktop grid" in the SpeQuloS installation documentation).	
  cp <address_to_scripts> /root/spequlos/
  chmod +x /root/spequlos/*.sh

 - To install the BOINC worker:
  - Install needed package:
  apt-get install subversion emacs less make m4 pkg-config autoconf libtool libssl-dev g++

  - Download BOINC source code 
  svn co http://boinc.berkeley.edu/svn/trunk/boinc

  - Install libcurl
  wget http://curl.haxx.se/download/curl-7.21.3.tar.gz
  tar xfvz curl-7.21.3.tar.gz
  cd curl-7.21.3
  make
  make install
  cd ..

  - Compile the BOINC client
  cd boinc
  ./_autosetup
  ./configure --disable-server --enable-static
  ln -s /usr/local/lib/libcurl.so.4 /usr/lib/libcurl.so.4
  make

  - Install the BOINC client
  mkdir /opt/BOINC_CLIENT
  cp client/boinc_client client/boinccmd /opt/BOINC_CLIENT
  cd ..

  - Remove unused files
  rm -r boinc curl-7.21.3 curl-7.21.3.tar.gz

  - Create the configuration file
  echo "<cc_config>
    <options>
    <fetch_minimal_work>
      1
    </fetch_minimal_work>
    <report_results_immediately>
      1
    </report_results_immediately>
    </options>
  </cc_config>" > /opt/BOINC_CLIENT/cc_config.xml

  - Copy the scripts used by SpeQuloS to start the worker:
You can find an exemple of those scripts inside the DG-plugin/BOINC/startup_script/ subdirectory of the SpeQuloS package. For SpeQuloS to start the appropriate script, it is mandatory that the script is named <DG_name>.sh, where DG_name is the name of the desktop grid registered to SpeQuloS during the installation (see "Registering a desktop grid" in the SpeQuloS installation documentation). You must also be sure that the value of the variable BOINC_ACCOUNT used in the script is the account number of the BOINC client dedicated to Cloudworkers (its user_id is "1").
  cp <address_to_scripts> /root/spequlos/
  chmod +x /root/spequlos/*.sh


- Cleaning:
  apt-get clean

- Exit chroot:
  exit

- Umount the image:
  umount fs-spequloscw/proc
  umount -l fs-spequloscw



  Creating an image for Eucalyptus
  ================================

- Install Eucalyptus command line tools:
  apt-get install euca2ools

- Create a dedicated disk image:
  cp fs-spequloscw.img emi-spequloscw.img

- Mount the image:
  mkdir emi-spequloscw
  mount -o loop emi-spequloscw.img emi-spequloscw/

- Edit the emi-spequlos/etc/fstab file:
  /dev/sda1 /     ext2    defaults 1 1
  /dev/sda2 /mnt  ext3    defaults 0 0
  /dev/sda3 swap  swap    defaults 0 0
  none      /proc proc    defaults 0 0
  none      /sys  sysfs   defaults 0 0

- Umount the image:
  umount emi-spequloscw/

- Bundle the image file for Eucaplyptus:
  euca-bundle-image -i emi-spequloscw.img -d . -k <euca_private_key> -c <euca_certificate> -u <euca_account_number>

- Upload the image to Eucaplyptus:
  euca-upload-bundle -b <spequlos_bucket> -m emi-spequloscw.img.manifest.xml -a <euca_access_key_ID> -s <euca_access_key_secret>

- Register the image in Eucaplyptus:
  euca-register <spequlos_bucket>/emi-spequloscw.img.manifest.xml -n spequlos-test-1 -K <euca_private_key> -C <euca_certificate>

It returns the EMI identifier.

- Be sure that the security group used with the image (probably the default security group) allows the SSH connexion on port 22

- Test the image:
  euca-run-instances -K <euca_private_key> -C <euca_certificate> <EMI_identifier>

It returns the instance identifier.
  euca-describe-instances -K <euca_private_key> -C <euca_certificate> <instance identifier>

It returns the machine address.
  ssh -i <path_to_private_key_id_rsa> root@<address>
  ...
  exit

To stop the instance: 
  euca-terminate-instances -K <euca_private_key> -C <euca_certificate> <instance identifier>



  Creating an image for Amazon EC2
  ================================

- Install EC2 command line tools (Ubuntu multiverse repository must be enabled):
  apt-get install ec2-ami-tools ec2-api-tools

- Create a dedicated disk image:
  cp fs-spequloscw.img ami-spequloscw.img

- Mount the image:
  mkdir ami-spequloscw
  mount -o loop ami-spequloscw.img ami-spequloscw/

- Edit the ami-spequlos/etc/fstab file:
  /dev/sda1 /     ext2    defaults 1 1
  /dev/sda2 /mnt  ext3    defaults 0 0
  /dev/sda3 swap  swap    defaults 0 0
  none      /proc proc    defaults 0 0
  none      /sys  sysfs   defaults 0 0

- Umount the image:
  umount ami-spequloscw/

- Bundle the image file for EC2:
  ec2-bundle-image -i ami-spequloscw.img -d . -k <ec2_private_key> -c <ec2_certificate> -u <ec2_account_number>

- Upload the image to EC2:
  ec2-upload-bundle -b <spequlos_bucket> -m ami-spequloscw.img.manifest.xml -a <ec2_access_key_ID> -s <ec2_access_key_secret>

- Register the image in EC2:
  ec2-register <spequlos_bucket>/ami-spequloscw.img.manifest.xml -n spequlos-test-1 -K <ec2_private_key> -C <ec2_certificate>

It returns the AMI identifier.

- Be sure that the security group used with the image (probably the default security group) allows the SSH connexion on port 22

- Test the image:
  ec2-run-instances -K <ec2_private_key> -C <ec2_certificate> <AMI_identifier>

It returns the instance identifier.
  ec2-describe-instances -K <ec2_private_key> -C <ec2_certificate> <instance identifier>

It returns the machine address.
  ssh -i <path_to_private_key_id_rsa> root@<address>
  ...
  exit

To stop the instance: 
  ec2-terminate-instances -K <ec2_private_key> -C <ec2_certificate> <instance identifier>



  Creating an image for a KVM hypervisor (e.g. used in OpenNebula)
  ================================================================

- Create a new disk image, including space for MBR and CloudWorker partition.
  dd if=/dev/zero of=mbr bs=8225280 count=1
  cat mbr fs-spequloscw.img mbr | cp --sparse=always /dev/stdin kvm-spequloscw.img 
  echo '1' | sfdisk kvm-spequloscw.img


- Mount the image and chroot to the filesystem
  mkdir kvm-spequloscw
  mount -o loop,offset=8225280 kvm-spequloscw.img kvm-spequloscw
  mount -t proc proc kvm-spequloscw/proc
  chroot kvm-spequloscw


- Install kernel and GRUB into the image
  echo "do_initrd=1" > /etc/kernel-img.conf
  aptitude update
  aptitude install -y linux-image-2.6-amd64 grub


- Configure GRUB:
  mkdir -p /boot/grub
  cp /usr/lib/grub/x86_64-pc/* /boot/grub/
  echo "timeout 1
  title           Debian GNU/Linux, kernel 2.6
  root            (hd0,0)
  kernel          /vmlinuz root=/dev/hda1 
  initrd          /initrd.img" > /boot/grub/menu.lst


- Exit the chroot:
  exit


- Write GRUB booting on image's MBR:
  echo "
  device (hd0) kvm-spequloscw.img 
  root (hd0,0)
  setup (hd0)
  quit" | kvm-spequloscw/usr/sbin/grub --device-map=/dev/null


- Clean:
  umount kvm-spequloscw/proc
  umount -l kvm-spequloscw
  rm mbr


The file kvm-spequloscw.img is ready to be used by any Cloud that uses KVM, such as OpenNebula (using oneimage or econe-upload commands for instance)