How to compile and install Eucalyptus 3.3.0 on Ubuntu 13.04 from Github sources (cloud-in-a-box)

 

This is an updated guide on how to build and install Eucalyptus 3.3.0 on Ubuntu 13.04. I have tested these steps on a freshly installed & updated Ubuntu 13.04 machine on Sunday, July 21st, 2013. There is another guide here for Eucalyptus 3.2 and Ubuntu 12.04. :)

Prerequisites

  • A Linux-capable machine with virtualization extensions
  • Space on your disk to install Ubuntu 13.04
  • More than 30GB of free space in /var/lib/eucalyptus. I had one 67GB root partition with 45GB free. Please note that not having sufficient free space on the Node Controller (NC) machine will prevent you from launching instances with a "Not enough resources (0 in default 1): vm instances" message.

WARNING: Please, as you follow this guide, copy and paste one command at a time, or type them by hand. Copying multiple commands at a time might lead to some of the commands not being executed.

Step 0: Download & install Ubuntu 13.04 on your system

You can find an x86_64 version of Ubuntu 13.04 here: http://www.ubuntu.com/download/desktop Install, then boot into your fresh installation!

Step 1: Installation prerequisites

Here, we install required libraries and binaries for Eucalyptus to build and run. We also update your system and set you as a password-less sudoer. Add yourself as a password-less ‘sudoer’ since a lot of the commands will require super user access and it’ll be much easier not to type your password 72323854542921 times!

# add yourself as a sudoer
me=`whoami`
file=/etc/sudoers.d/$me
sudo touch $file
sudo chmod 0440 $file
sudo sh -c "printf \"%s\\tALL=(ALL)\\tNOPASSWD: ALL\\n\" $me >$file"
sudo chmod 0440 $file

Upgrade your Ubuntu box with the latest packages & kernel. You will need to restart if this step ends up installing a new kernel version.

# update your system
sudo apt-get update
sudo apt-get upgrade
sudo apt-get dist-upgrade

Install some (some not) useful applications, including the build & runtime dependencies for Eucalyptus.

# install some useful applications
sudo apt-get -y install ssh subversion git vim rsync wget

# install eucalyptus build dependencies
# WARNING: install these now, not later
sudo apt-get -y install cdbs debhelper libaxis2c-dev librampart-dev \
  libvirt-dev libfuse-dev libfuse2 libcurl4-openssl-dev \
  libssl-dev ant-optional zlib1g-dev pkg-config swig python \
  python-setuptools open-iscsi libxslt1-dev gengetopt ant \
  postgresql-server-dev-9.1 openjdk-7-jdk groovy libcap-dev

# install eucalyptus runtime dependencies
# WARNING: install these now, not later
sudo apt-get install -y bridge-utils iputils-arping libapache2-mod-axis2c adduser \
  apache2 apache2-mpm-worker bridge-utils dhcp3-server euca2ools file \
  iptables iputils-arping libapache2-mod-axis2c libaxis2c0 libc6 \
  libcrypt-openssl-random-perl libcrypt-openssl-rsa-perl libcrypt-x509-perl \
  libcurl3 libdevmapper1.02.1 libpam-modules librampart0 libssl1.0.0 libvirt0 \
  libvirt-bin libxml2 libxslt1.1 lvm2 open-iscsi openssh-client openssh-server \
  parted postgresql-client-9.1 python python2.7 python-boto python-psutil \
  rsync sudo tgt vblade vlan vtun postgresql-9.1 apache2-threaded-dev \
  bzr drbd8-utils gcc kvm libsys-virt-perl libxml-simple-perl make openntpd \
  python-libvirt python-pygresql qemu-kvm unzip at

# install axis2
wget http://archive.apache.org/dist/ws/axis2/1_4/axis2-1.4-bin.zip
sudo unzip axis2-1.4-bin.zip -d /opt

We have to disable apparmor on Ubuntu, so as to avoid conflicts with Eucalyptus.

WARNING: Removing apparmor could lower your system’s security, but it has to be done since it conflicts with Eucalyptus in a few ways. A workaround would require tuning the /etc/apparmor.d/ configuration files.

# we must disable apparmor to avoid DHCP/libvirt daemon issues
sudo service apparmor stop
sudo service apparmor teardown
sudo apt-get remove apparmor

# restart libvirt-bin after removing apparmor
sudo service libvirt-bin restart

Eucalyptus needs an ethernet bridge device on which instance network adapters can attach to. This way, your box will be able to talk to an instance and viceversa.

WARNING: I am using 10.1.0.1 as the bridge’s IP address here. Please make sure it does not conflict with your existing network configuration, or change it otherwise.

# add a br0 bridge interface to your /etc/network/interfaces file
# WARNING: no need to attach your eth0 or any other card to this bridge
sudo sh -c 'echo "\nauto br0\niface br0 inet static\naddress 10.1.0.1\nnetmask 255.255.0.0\nbridge_stp off\nbridge_ports none" >>/etc/network/interfaces'

# bring the bridge up
sudo ifup br0

# make sure the bridge is up
sudo brctl show br0
ifconfig br0

NOTE: Please restart your system if a new kernel was installed in the upgrade/dist-upgrade process.

sudo reboot now

Step 2: Download Eucalyptus 3.3.0.1 from GitHub

# download eucalyptus 3.3.0.1 from github
git clone https://github.com/eucalyptus/eucalyptus.git
pushd eucalyptus/
git checkout 3.3.0.1
popd

Step 3: Configure Eucalyptus for building

# configure eucalyptus
export JAVA_HOME="/usr/lib/jvm/java-7-openjdk-amd64"
export JAVA="$JAVA_HOME/jre/bin/java"
export EUCALYPTUS="/"

pushd eucalyptus/
./configure \
  --prefix=$EUCALYPTUS \
  --with-axis2c=/usr/lib/axis2 \
  --with-axis2c-services=/usr/lib/axis2/services \
  --with-apache2-module-dir=/usr/lib/apache2/modules \
  --with-axis2=/opt/axis2-1.4
popd

Step 4: Patch a few build issues

You can download the patch file here: storage-makefile.patch

wget --user-agent=Mozilla http://alinush.github.io/files/storage-makefile.patch
mv storage-makefile.patch eucalyptus/
pushd eucalyptus/
patch -p1 < storage-makefile.patch
popd

You can download the patch file here: clc-makefile.patch

wget --user-agent=Mozilla http://alinush.github.io/files/clc-makefile.patch
mv clc-makefile.patch eucalyptus/
pushd eucalyptus/
patch -p1 < clc-makefile.patch
popd

Step 5: Build Eucalyptus

# build eucalyptus with make
export JAVA_HOME="/usr/lib/jvm/java-7-openjdk-amd64"
export JAVA="$JAVA_HOME/jre/bin/java"
export EUCALYPTUS="/"

pushd eucalyptus/
make
popd

Step 6: Pre-installation requirements

There are a few things we must do, before we can install Eucalyptus with sudo make install.

# pre-installation requirements (add eucalyptus user/group)
sudo addgroup eucalyptus
sudo adduser eucalyptus --ingroup eucalyptus
sudo usermod -d /var/lib/eucalyptus/ eucalyptus
sudo ln -s /lib/udev/scsi_id /usr/bin/scsi_id

Step 7: Install Eucalyptus

export JAVA_HOME="/usr/lib/jvm/java-7-openjdk-amd64"
export JAVA="$JAVA_HOME/jre/bin/java"
export EUCALYPTUS="/"

# install eucalyptus with make
pushd eucalyptus/
sudo make install
popd

Step 8: Post-installation requirements

# post-installation requirements (set permissions, copy XML fault files)
sudo chown -R eucalyptus:eucalyptus /etc/eucalyptus /var/lib/eucalyptus/ /var/log/eucalyptus
sudo chmod +s /usr/lib/eucalyptus/euca_rootwrap -v

sudo adduser `id -un` libvirtd
sudo adduser eucalyptus libvirtd
sudo adduser `id -un` kvm
sudo adduser eucalyptus kvm

sudo mkdir -p /var/lock/subsys
sudo mkdir -p /var/run/eucalyptus
sudo chown -R eucalyptus:eucalyptus /var/lock/subsys /var/run/eucalyptus

sudo mkdir -p /etc/eucalyptus/faults/en_US
sudo cp eucalyptus/util/faults/en_US/* /etc/eucalyptus/faults/en_US/

Step 9: Configure Eucalyptus by editing /etc/eucalyptus/eucalyptus.conf

You should start off with this eucalyptus.conf file from here and edit it to match your network configuration.

wget --user-agent=Mozilla http://alinush.github.io/files/eucalyptus-3.3.conf
sudo mv /etc/eucalyptus/eucalyptus.conf /etc/eucalyptus/eucalyptus.conf.orig
sudo cp eucalyptus-3.3.conf /etc/eucalyptus

# edit /etc/eucalyptus/eucalyptus.conf in your favorite editor 
vim /etc/eucalyptus/eucalyptus.conf

# the settings you might need to change are:
VNET_PUBLICIPS="10.0.0.100-10.0.0.253"
VNET_DNS="10.0.0.1"

VNET_SUBNET="10.1.0.0"
VNET_NETMASK="255.255.255.0"
VNET_ADDRSPERNET="32"

WARNING: Here i am assuming you are behind a NAT router (your typical WiFi router with DNS and DHCP), whose address is 10.0.0.1. If that’s not the case, please change the VNET_PUBLICIPS and the VNET_DNS to match your router’s configuration. Ideally, you would also have your router configured to ONLY assign addresses below 10.0.0.100 since we are reserving the rest for Eucalyptus instances. However, as long as you have less than 100 physical machines on your network this should not pose a problem.

WARNING: Also know that the Eucalypus virtual network (VNET_SUBNET) is in the 10.1.0.0/24 subnet, so you will need to change that as well in eucalyptus.conf if it conflicts with your network.

Step 10: Initialize the Cloud Controller (CLC)

Now, we are ready to initialize the CLC.

# some pre-initialization setup
sudo service postgresql restart
sudo killall dnsmasq

sudo mkdir -p /var/lock/subsys
sudo mkdir -p /var/run/eucalyptus
sudo chown -R eucalyptus:eucalyptus /var/lock/subsys /var/run/eucalyptus

# initialize the Cloud Controller (CLC)
sudo /usr/sbin/euca_conf --setup
# check the logs while this is being executed (see below)
sudo /usr/sbin/euca_conf --initialize

# mark the logs as readable and writable by everyone
sudo chmod -R a+rw /var/log/eucalyptus/

While the euca_conf --initialize command is running, you should check the CLC logs to see what’s happening.

# WARNING: 'euca_conf --initialize' can either loop infinitely if something is wrong
# or return with an error.
# To be sure, you can tail -f cloud-output.log to see what's going on
tail -f /var/log/eucalyptus/cloud-output.log

The last logs before a sucessful euca_conf --initialize on my machine were:

2013-06-18 15:41:02 INFO | -> Setup done for persistence context: eucalyptus_cloud
2013-06-18 15:41:02 INFO | Trying to load config for com.eucalyptus.util.StorageProperties from //etc/eucalyptus/cloud.d/scripts/storageprops.groovy
2013-06-18 15:41:05 INFO | -> Setup done for persistence context: eucalyptus_storage
2013-06-18 15:41:08 INFO | -> Setup done for persistence context: eucalyptus_autoscaling
2013-06-18 15:41:09 INFO | -> Setup done for persistence context: eucalyptus_records
2013-06-18 15:41:17 INFO | -> Setup done for persistence context: eucalyptus_cloudwatch
2013-06-18 15:41:17 INFO | -> Setup done for persistence context: eucalyptus_general
2013-06-18 15:41:21 INFO | -> Setup done for persistence context: eucalyptus_auth
2013-06-18 15:41:22 INFO | -> Setup done for persistence context: eucalyptus_config
2013-06-18 15:41:24 INFO | -> Setup done for persistence context: eucalyptus_loadbalancing
2013-06-18 15:41:27 INFO | -> Setup done for persistence context: eucalyptus_walrus
2013-06-18 15:41:28 INFO | -> Setup done for persistence context: eucalyptus_dns
2013-06-18 15:41:29 INFO | -> Setup done for persistence context: eucalyptus_faults
2013-06-18 15:41:31 INFO | -> Setup done for persistence context: eucalyptus_reporting
2013-06-18 15:41:31 INFO | :1371584491778:Bootstrap:bootstrap:COMPONENT_REGISTERED:Component eucalyptus=available service=not-local
| :1371584491778:Bootstrap::bootstrap:COMPONENT_REGISTERED:
2013-06-18 15:41:31 INFO | :1371584491780:ServiceConfigurations:bootstrap:COMPONENT_REGISTERED:ServiceConfiguration eucalyptus arn:euca:eucalyptus:192.168.1.102/ 192.168.1.102:8773:/services/Eucalyptus:vm-local:host-local:PRIMORDIAL
2013-06-18 15:41:31 INFO | Added registration for this cloud controller: ServiceConfiguration eucalyptus arn:euca:eucalyptus:::192.168.1.102/ 192.168.1.102:8773:/services/Eucalyptus:vm-local:host-local:PRIMORDIAL
2013-06-18 15:41:31 INFO | Postgres command : '/usr/lib/postgresql/9.1/bin/pg_ctl' 'status' '-D//var/lib/eucalyptus/db/data'
2013-06-18 15:41:31 INFO | stdout: pg_ctl: server is running (PID: 17976)
2013-06-18 15:41:31 INFO | stdout: /usr/lib/postgresql/9.1/bin/postgres "-D" "/var/lib/eucalyptus/db/data" "-h0.0.0.0/0" "-p8777" "-i"
2013-06-18 15:41:31 INFO | Postgres command : '/usr/lib/postgresql/9.1/bin/pg_ctl' 'stop' '-mf' '-D//var/lib/eucalyptus/db/data'
2013-06-18 15:41:38 INFO | stdout: waiting for server to shut down.......... done
2013-06-18 15:41:38 INFO | stdout: server stopped
2013-06-18 15:41:38 INFO | Executing Pre-Shutdown Hooks...
2013-06-18 15:41:38 INFO | Executing Shutdown Hooks...
2013-06-18 15:41:38 INFO | Executing Shutdown Hook: com.eucalyptus.component.BasicService$1@180d47cf
2013-06-18 15:41:38 WARN | Parsing common file /usr/share/eucalyptus/faults/en_US/common.xml
2013-06-18 15:41:39 WARN | Parsing common file /etc/eucalyptus/faults/en_US/common.xml
2013-06-18 15:41:39 WARN | SHUTDOWN Service: arn:euca:eucalyptus::db:192.168.1.102/
2013-06-18 15:41:39 INFO | Executing Post-Shutdown Hooks...
2013-06-18 15:41:39 INFO | Executing Post-Shutdown Hook: com.eucalyptus.system.Threads$ThreadPool$1@2bb3e57c
2013-06-18 15:41:39 WARN | SHUTDOWN:Eucalyptus.SYSTEM Stopping thread pool...
2013-06-18 15:41:39 INFO | Executing Post-Shutdown Hook: PostgresqlBootstrapper$3@2f9f464e
2013-06-18 15:41:39 INFO | Executing Post-Shutdown Hook: com.eucalyptus.system.Threads$ThreadPool$1@3cf6fa16
2013-06-18 15:41:39 WARN | SHUTDOWN:Eucalyptus.bootstrap:Futures Stopping thread pool...

Step 11: Start the Eucalyptust cloud services

Now, let’s start the Cloud Controller (CLC), Walrus, Cluster Controller (CC), Storage Controller (SC) and Node Controller (NC).

# openntpd needs to be started for time synchronization
sudo service openntpd restart

# if JDBC cannot write to a temp file in /tmp, then eucalyptus CLC will fail
sudo chmod 0777 /tmp

# for the 3rd time, these directories have to exist!
sudo mkdir -p /var/lock/subsys /var/run/eucalyptus
sudo chown -R eucalyptus:eucalyptus /var/lock/subsys /var/run/eucalyptus

# start the eucalyptus services
sudo service eucalyptus-cloud start
sudo service eucalyptus-cc start
sudo service eucalyptus-nc start

NOTE: You might get a message when starting the NC, complaining about the keys not being found. That’s okay, just ignore it for now.

Cannot find keys (node-pk.pem, node-cert.pem) in //var/lib/eucalyptus/keys

As always, please check the CLC logs while services are starting. You will not be able to check the CC and NC logs until later when we register them with the CLC.

# To be sure all is well, you can tail -f cloud-output.log to see what's going on
tail -f /var/log/eucalyptus/cloud-output.log

The last CLC logs will look like this. Notice that the Elastic Load Balancer (ELB) needs to be set up.

2013-07-06 15:36:33 INFO | Postgres command : '/usr/lib/postgresql/9.1/bin/pg_ctl' 'status' '-D//var/lib/eucalyptus/db/data'
2013-07-06 15:36:33 INFO | stdout: pg_ctl: server is running (PID: 1154)
2013-07-06 15:36:33 INFO | stdout: /usr/lib/postgresql/9.1/bin/postgres "-D" "/var/lib/eucalyptus/db/data" "-h0.0.0.0/0" "-p8777" "-i"
2013-07-06 15:36:35 ERROR | START:DISABLED arn:euca:eucalyptus::loadbalancing:10.0.0.3/ NOTREADY->DISABLED=NOTREADY [LoadBalancingPropertyBootstrapper.enable( ): returned false, terminating bootstrap for component: loadbalancing]
2013-07-06 15:36:35 ERROR | START:DISABLED arn:euca:eucalyptus::loadbalancing:10.0.0.3/ NOTREADY->DISABLED=NOTREADY [LoadBalancingPropertyBootstrapper.enable( ): returned false, terminating bootstrap for component: loadbalancing]
2013-07-06 15:36:35 ERROR | START:DISABLED arn:euca:eucalyptus::loadbalancing:10.0.0.3/ NOTREADY->DISABLED=NOTREADY [LoadBalancingPropertyBootstrapper.enable( ): returned false, terminating bootstrap for component: loadbalancing]
2013-07-06 15:36:36 ERROR | START:DISABLED arn:euca:eucalyptus::loadbalancing:10.0.0.3/ NOTREADY->DISABLED=NOTREADY [LoadBalancingPropertyBootstrapper.enable( ): returned false, terminating bootstrap for component: loadbalancing]
2013-07-06 15:36:36 ERROR | START:DISABLED arn:euca:eucalyptus::loadbalancing:10.0.0.3/ NOTREADY->DISABLED=NOTREADY [LoadBalancingPropertyBootstrapper.enable( ): returned false, terminating bootstrap for component: loadbalancing]
2013-07-06 15:36:37 ERROR | START:DISABLED arn:euca:eucalyptus::loadbalancing:10.0.0.3/ NOTREADY->DISABLED=NOTREADY [LoadBalancingPropertyBootstrapper.enable( ): returned false, terminating bootstrap for component: loadbalancing]
2013-07-06 15:36:39 INFO | Hosts.entrySet(): 10.0.0.3 finished.
2013-07-06 15:36:39 INFO | Hosts.entrySet(): 10.0.0.3 finished.'

Step 12: Register the Eucalyptus components

We have to register the CC, SC and Walrus with the CLC. And we also have to register the NC with the CC.

WARNING: Once again, this step assumes that you are behind a router whose address is 10.0.0.1. Adjust as necessary.

# set a root password on your machine, you will be asked for it later
sudo passwd

# generate an SSH key for root, and let root@localhost ssh into root@localhost
# (this is for convenience, so that you do not get prompted for the root
# password when you register the components later on)
if sudo su -c "test ! -f /root/.ssh/id_rsa"; then
    sudo su -c 'ssh-keygen -f /root/.ssh/id_rsa -t rsa -N ""'
fi

# here you will be prompted for the password you entered above in 'sudo passwd'
sudo su -c "ssh-copy-id root@localhost"

# WARNING: the assumption here is that you are behind a NAT'ed network (you are using a router)
# and your NAT-assigned IP address is 10.0.0.100. also, the IP address on the br0 bridge should be 10.1.0.1
cluster=cluster01
extaddr=10.0.0.100
intaddr=10.1.0.1

# register the components
sudo -u root /usr/sbin/euca_conf --register-walrus --partition walrus --host $extaddr --component walrus-$extaddr
sudo -u root /usr/sbin/euca_conf --register-cluster --partition $cluster --host $extaddr --component cc-$extaddr
sudo -u root /usr/sbin/euca_conf --register-sc --partition $cluster --host $extaddr --component sc-$extaddr
sudo -u root /usr/sbin/euca_conf --register-nodes "$intaddr"

You will get an warning upon registering the SC, which we will deal with later. Now we can check that everything has registered (it might take a few seconds for everything to be in the ENABLED state).

# check to be sure everything is registered
sudo euca_conf --list-clouds
sudo euca_conf --list-walrus
sudo euca_conf --list-clusters

# the --list-sc command will tell you your SC is NOTREADY, BROKEN or DISABLED.
# we will fix this later.
sudo euca_conf --list-sc

# the --list-nodes will not work until you have the eucalyptus 
# credentials sourced in your environment. so
# you can check the NC logs to make sure the NC is up
euca_conf --list-nodes 

tail -f /var/log/eucalyptus/nc.log

You should see messages like these popping up continuously in the NC logs:

2013-07-06 15:45:08 DEBUG 000003546 statfs_path | path '/var/lib/eucalyptus/instances/work' resolved
2013-07-06 15:45:08 DEBUG 000003546 statfs_path | to '/var/lib/eucalyptus/instances/work' with ID 0
2013-07-06 15:45:08 DEBUG 000003546 statfs_path | of size 99544817664 bytes with available 87483858944 bytes
2013-07-06 15:45:08 DEBUG 000003546 statfs_path | path '/var/lib/eucalyptus/instances/cache' resolved
2013-07-06 15:45:08 DEBUG 000003546 statfs_path | to '/var/lib/eucalyptus/instances/cache' with ID 0
2013-07-06 15:45:08 DEBUG 000003546 statfs_path | of size 99544817664 bytes with available 87483858944 bytes
2013-07-06 15:45:12 DEBUG 000000786 doDescribeResource | returning status=enabled cores=2/2 mem=7813/7813 disk=25/25 iqn=iqn.1993-08.org.debian:01:4eccde31f9c
2013-07-06 15:45:12 DEBUG 000000786 doDescribeInstances | invoked userId=eucalyptus correlationId=UNSET epoch=31 services[0]{.name=cc-10.0.0.3 .type=cluster .uris[0]=http://10.0.0.3:8774/axis2/services/EucalyptusCC}
2013-07-06 15:45:18 DEBUG 000000786 doDescribeResource | returning status=enabled cores=2/2 mem=7813/7813 disk=25/25 iqn=iqn.1993-08.org.debian:01:4eccde31f9c
2013-07-06 15:45:18 DEBUG 000000786 doDescribeInstances | invoked userId=eucalyptus correlationId=UNSET epoch=31 services[0]{.name=cc-10.0.0.3 .type=cluster .uris[0]=http://10.0.0.3:8774/axis2/services/EucalyptusCC}
2013-07-06 15:45:25 DEBUG 000000786 doDescribeResource | returning status=enabled cores=2/2 mem=7813/7813 disk=25/25 iqn=iqn.1993-08.org.debian:01:4eccde31f9c
2013-07-06 15:45:25 DEBUG 000000786 doDescribeInstances | invoked userId=eucalyptus correlationId=UNSET epoch=31 services[0]{.name=cc-10.0.0.3 .type=cluster .uris[0]=http://10.0.0.3:8774/axis2/services/EucalyptusCC}
2013-07-06 15:45:31 DEBUG 000000786 doDescribeResource | returning status=enabled cores=2/2 mem=7813/7813 disk=25/25 iqn=iqn.1993-08.org.debian:01:4eccde31f9c'

Step 13: Retrieve and ‘source’ your Eucalyptus admin user credentials

To give commands to the cloud, you will need to ‘download’ the admin credentials from Eucalyptus and source them in your environment.

# get your eucalyptus admin user credentials and source them in your current terminal/environment
userid=`id -u`
mkdir -p credentials/
sudo /usr/sbin/euca_conf --get-credentials admin.zip
sudo chown -R $userid:$userid *
unzip admin.zip -d credentials/

. credentials/eucarc

You will see a warning upon sourcing the credentials: "WARN: Load Balancing service URL is not configured." We will fix this later.

WARNING: From this point on, we assume your eucalyptus credentials are sourced in your environment. If they are not, please repeat this step. Now that the credentials are sourced, you will be able to --list-nodes in euca_conf.

euca_conf --list-nodes

Step 14: Fix the Storage Controller (SC) by setting it into ‘overlay’ mode

We have to set the Storage Controller (SC) into ‘overlay’ mode in order to get it in the ENABLED state. Right now it should be in the NOTREADY or DISABLED state. First we have to rename the tgt service to tgtd (seems like this is still an issue with Eucalyptus 3.3 in Ubuntu).

sudo service tgt stop
sudo mv /etc/init/tgt.conf /etc/init/tgtd.conf
sudo mv /etc/init.d/tgt /etc/init.d/tgtd
sudo service tgtd start

Now, we can configure the SC to be in ‘overlay’ mode.

sc=`euca-describe-properties | grep blockstoragemanager | cut -f 2`
echo $sc # should print 'cluster01.storage.blockstoragemanager'
euca-modify-property -p $sc=overlay # should print 'PROPERTY cluster01.storage.blockstoragemanager overlay was '

You can now check that the SC is enabled, but it will take a few seconds for it to appear as ENABLED.

sudo euca_conf --list-sc

Step 15: Prepare to launch an instance

Check to see how many instances you can launch. This will depend on the number of cores on your CPU, on your RAM and on your free hard disk space.

# check the cloud's availability, the services
euca-describe-availability-zones verbose

You should see a few available m1.small instances:

AVAILABILITYZONE |- m1.small 0002 / 0002 1 256 5

Next, check the status of all services. The ELB service will be in the NOTREADY state. As a result, the output will be huge sometimes, so just reissue the command a few times.

euca-describe-services

You will see a line like this for the load balancer service:

SERVICE loadbalancing eucalyptus 10.0.0.3 NOTREADY 32 http://10.0.0.3:8773/services/LoadBalancing arn:euca:eucalyptus::loadbalancing:10.0.0.3/

We must enable IP forwarding, to allow Eucalyptus to do some routing.

sudo su -c 'echo 1 >/proc/sys/net/ipv4/ip_forward'

Next, we have to generate an RSA keypair for SSH and store the public key part in the cloud, keeping the private key to ourselves.

# generate an SSH keypair in Eucalyptus (use euca-create-keypair if euca-add-keypair does not work)
euca-add-keypair mykey | grep -v ^KEYPAIR >mykey.pem

mykey.pem should look something like this: WARNING: To prevent you from using this key, I have changed the output considerably, please generate your own using this command, as opposed to being lazy and just copy & pasting, which will create serious security issues.

-----BEGIN RSA PRIVATE KEY-----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-----END RSA PRIVATE KEY-----

Once you have the file with the key make sure to change permissions so only you can read and write the file.

chmod 0600 mykey.pem

Step 16: Install the newer euca2ools 3.0 from GitHub sources

We must update euca2ools to the latest 3.0 version, since there seems to be a bug in eustore-install-image that does not allow the ‘eucalyptus’ admin user to install an OS image.

git clone https://github.com/eucalyptus/euca2ools.git
pushd euca2ools/
git checkout 3.0.0
popd

sudo apt-get remove euca2ools # WARNING: This is IMPORTANT!

pushd euca2ools/
sudo python setup.py install
popd

Step 17: Install an OS image from eustore in your cloud

Now we can download and install an OS image that we can run in the cloud as a VM.

eustore-describe-images

We want to install an EMI that’s KVM-compatible, so I selected this one:

1424900416 opensuse x86_64 starter OpenSUSE 12.2 x86_64 - KVM image. SUSE Firewall off. Root disk of 2.5G. Root user enabled. Working with kexec kernel and ramdisk. OpenSUSE minimal base package set.

Download & install the OpenSUSE EMI. This took 20 minutes on my machine to download, on a 4Mbit/s coonection:

# NOTE: use the --debug flag to see what's going on 
# (the command does not seem to print any output on my machine otherwise)
eustore-install-image -i 1424900416 -b osimages --debug
euca-describe-images

Step 18: Start an instance (a.k.a. a VM) in your cloud

First, we must load the KVM kernel module.

# start KVM for Intel
sudo modprobe kvm_intel
# ...or start KVM for AMD
sudo modprobe kvm_amd

Get the first EMI in the list of images (you only have one at this point).

emi=`euca-describe-images | grep IMAGE | grep emi-........ -o | head -n 1`

Start an instance! This will take a few minutes to start up, since the NC has to download the EMI from Walrus.

euca-run-instances -k mykey -t m1.small $emi

Wait a few minutes until euca-describe-instances marks your instance as running.

euca-describe-instances

Check out the console output of the instance.

inst=`euca-describe-instances | grep running | head -n 1 | cut -f 2`
euca-get-console-output $inst

Get the instance’s IP and connect to it.

inst_ip=`euca-describe-instances | grep running | cut -f 4`

# remove a previous host key (if any) at that address
ssh-keygen -R $inst_ip &>/dev/null

# you need to authorize SSH access in your default security group
euca-authorize -P tcp -p 22 -s 0.0.0.0/0 default

# now you can connect to your instance and check if internet
# connectivity works inside it
ssh -i mykey.pem root@$inst_ip

You should see the following output:

The authenticity of host '10.0.0.100 (10.0.0.100)' can't be established.
ECDSA key fingerprint is 57:76:26:5d:2e:82:98:0b:e8:9e:1e:2b:d3:6c:9f:7b.
Are you sure you want to continue connecting (yes/no)? yes
Warning: Permanently added '10.0.0.100' (ECDSA) to the list of known hosts.

Last login: Sat Jul 6 18:15:21 2013 from 10.1.0.97
Have lots of Eucalyptus fun...
ip-10-1-0-122:~ # # you can ping/wget google.com inside the instance
ip-10-1-0-122:~ # wget google.com
ip-10-1-0-122:~ # ping google.com

PING google.com (74.125.226.238) 56(84) bytes of data.

64 bytes from lga15s29-in-f14.1e100.net (74.125.226.238): icmp_seq=1 ttl=53 time=586 ms
64 bytes from lga15s29-in-f14.1e100.net (74.125.226.238): icmp_seq=2 ttl=53 time=607 ms
64 bytes from lga15s29-in-f14.1e100.net (74.125.226.238): icmp_seq=3 ttl=53 time=558 ms
64 bytes from lga15s29-in-f14.1e100.net (74.125.226.238): icmp_seq=4 ttl=53 time=507 ms
64 bytes from lga15s29-in-f14.1e100.net (74.125.226.238): icmp_seq=5 ttl=53 time=438 ms
64 bytes from lga15s29-in-f14.1e100.net (74.125.226.238): icmp_seq=6 ttl=53 time=472 ms
64 bytes from lga15s29-in-f14.1e100.net (74.125.226.238): icmp_seq=7 ttl=53 time=656 ms
^C
--- google.com ping statistics ---
7 packets transmitted, 7 received, 0% packet loss, time 11108ms
rtt min/avg/max/mdev = 438.535/546.710/656.002/71.824 ms"

Step 19: Attach a volume to your instance

Now we can play with EBS volumes.

euca-create-volume -z cluster01 -s 5
euca-describe-volumes

Once the volume becomes available, try attaching it to the instance.

vol=`euca-describe-volumes | grep available | head -n 1 | cut -f 2`
euca-attach-volume -i $inst -d /dev/vdb $vol

Check to see if the volume attached.

euca-describe-volumes

The output of euca-describe-volumes should look like this:

VOLUME vol-1E573DBC 5 cluster01 in-use 2013-07-06T22:27:04.372Z standard
ATTACHMENT vol-1E573DBC i-DC213D33 unknown,requested:/dev/vdb attaching 2013-07-06T22:31:55.979Z

Inside the instance, you can check dmesg logs to see if the volume attached.

ip-10-1-0-122:~ # dmesg

The output should look like this (note the /dev/vdb device created at the end):

[ 1114.557787] pci 0000:00:06.0: [1af4:1001] type 00 class 0x010000
[ 1114.557970] pci 0000:00:06.0: reg 10: [io 0x0000-0x003f]
[ 1114.558062] pci 0000:00:06.0: reg 14: [mem 0x00000000-0x00000fff]
[ 1114.559107] pci 0000:00:06.0: BAR 1: assigned [mem 0x80000000-0x80000fff]
[ 1114.559158] pci 0000:00:06.0: BAR 0: assigned [io 0x1000-0x103f]
[ 1114.559210] pci 0000:00:00.0: no hotplug settings from platform
[ 1114.559211] pci 0000:00:00.0: using default PCI settings
[ 1114.559268] pci 0000:00:01.0: no hotplug settings from platform
[ 1114.559287] pci 0000:00:01.0: using default PCI settings
[ 1114.559356] ata_piix 0000:00:01.1: no hotplug settings from platform
[ 1114.559358] ata_piix 0000:00:01.1: using default PCI settings
[ 1114.559414] uhci_hcd 0000:00:01.2: no hotplug settings from platform
[ 1114.559416] uhci_hcd 0000:00:01.2: using default PCI settings
[ 1114.559483] piix4_smbus 0000:00:01.3: no hotplug settings from platform
[ 1114.559485] piix4_smbus 0000:00:01.3: using default PCI settings
[ 1114.559540] virtio-pci 0000:00:03.0: no hotplug settings from platform
[ 1114.559542] virtio-pci 0000:00:03.0: using default PCI settings
[ 1114.559596] virtio-pci 0000:00:04.0: no hotplug settings from platform
[ 1114.559598] virtio-pci 0000:00:04.0: using default PCI settings
[ 1114.559652] virtio-pci 0000:00:05.0: no hotplug settings from platform
[ 1114.559654] virtio-pci 0000:00:05.0: using default PCI settings
[ 1114.559708] pci 0000:00:06.0: no hotplug settings from platform
[ 1114.559709] pci 0000:00:06.0: using default PCI settings
[ 1114.569109] virtio-pci 0000:00:06.0: enabling device (0000 -> 0003)
[ 1114.571647] ACPI: PCI Interrupt Link [LNKB] enabled at IRQ 11
[ 1114.579644] virtio-pci 0000:00:06.0: setting latency timer to 64
[ 1114.580116] virtio-pci 0000:00:06.0: irq 45 for MSI/MSI-X
[ 1114.580148] virtio-pci 0000:00:06.0: irq 46 for MSI/MSI-X
[ 1114.584510] vdb: unknown partition table

Feel free to write some data on your volume.

ip-10-1-0-122:~ # sudo mkfs.ext2 /dev/vdb
ip-10-1-0-122:~ # sudo mount /dev/vdb /mnt
ip-10-1-0-122:~ # echo hello >/mnt/hello
ip-10-1-0-122:~ # sudo umount /dev/vdb

Now you can safely detach the volume.

euca-detach-volume $vol

Feel free to reattach the volume and check that the data is still there.

Step 20: Terminate your instance

Terminate the instance using euca-terminate-instances

euca-terminate-instances $inst

Bonus Step: Get the Elastic Load Balancer to work

I have not used the Elastic Load Balancer at all, but I can show you how to enable it and hopefully you can figure out the rest. First, we have to download an elastic load balancer EMI. We use one provided as an rpm by Eucalyptus on their website.

wget http://downloads.eucalyptus.com/software/eucalyptus/3.3/centos/6/x86_64/eucalyptus-load-balancer-image-1.0.0-0.92.el6.x86_64.rpm

sudo apt-get install rpm

sudo rpm -i --nodeps eucalyptus-load-balancer-image-1.0.0-0.92.el6.x86_64.rpm

Next, we must install the ELB EMI.

git clone https://github.com/eucalyptus/load-balancer-image.git

pushd load-balancer-image/
./euca-install-load-balancer --install-default
popd

Your output should be similar to this:

Preparing to extract image...
-- Uploading ramdisk image --
Registered ramdisk image eri-85E93353
-- Uploading kernel image --
Registered kernel image eki-42933821
-- Uploading machine image --
Registered machine image emi-870E3685
-- Done --
PROPERTY loadbalancing.loadbalancer_emi emi-870E3685 was NULL

Load Balancer Support is Enabled

Now, we should check to see the ELB service is working.

euca-describe-services | grep loadbalancing

Hopefully, you see the same message.

SERVICE loadbalancing eucalyptus 10.0.0.3 ENABLED 31 http://10.0.0.3:8773/services/LoadBalancing arn:euca:eucalyptus::loadbalancing:10.0.0.3/

That’s all!

Enjoy your cloud!

This page used to be at http://alinush.org/2013/07/21/how-to-compile-and-install-eucalyptus-3-3-0-on-ubuntu-13-04-from-github-sources-cloud-in-a-box/