Upgrading Single Front-end Deployments

If you are running a 5.12.x version or older, please check these set of steps (some additional ones may apply, please review them at the end of the section).

Upgrading from 5.6.x+

Step 1. Check Virtual Machine Status

Before proceeding, make sure you don’t have any VMs in a transient state (prolog, migr, epil, save). Wait until these VMs get to a final state (running, suspended, stopped, done). Check the Managing Virtual Machines guide for more information on the VM life-cycle.

Step 2. Set All Hosts to Offline Mode

Set all Hosts to offline mode to stop all monitoring processes.

Step 3. Stop OpenNebula

Stop OpenNebula and any other related services you may have running: OneFlow, EC2, and Sunstone. It’s preferable to use the system tools, like systemctl or service as root in order to stop the services.

Warning

Make sure that every OpenNebula process is stopped. The output of systemctl list-units | grep opennebula should be empty.

Step 4. Back-up OpenNebula Configuration

Back-up the configuration files located in /etc/one and /var/lib/one/remotes/etc. You don’t need to do a manual backup of your database; the onedb command will perform one automatically.

cp -ra /etc/one /etc/one.$(date +'%Y-%m-%d')
cp -ra /var/lib/one/remotes/etc /var/lib/one/remotes/etc.$(date +'%Y-%m-%d')

Step 5. Upgrade to the New Version

Ubuntu/Debian

apt-get update
apt-get install --only-upgrade opennebula opennebula-sunstone opennebula-gate opennebula-flow opennebula-provision python3-pyone

CentOS/RHEL

yum upgrade opennebula opennebula-sunstone opennebula-gate opennebula-flow opennebula-provision python-pyone python3-pyone

Important

When upgrading an existing deployment which could be running OpenNebula older than 5.10.0 anytime in the past, you might need to upgrade also required Ruby dependencies with script install_gems if you are not yet using the shipped Ruby gems (i.e., when symbolic link /usr/share/one/gems doesn’t exist on your Front-end)!

If unsure, run /usr/share/one/install_gems and the script warns if action is not relevant for you. For example:

/usr/share/one/install_gems
WARNING: Running install_gems is not necessary anymore, as all the
required dependencies are already installed by your packaging
system into symlinked location /usr/share/one/gems. Ruby gems
installed by this script won't be used until this symlink exists.
Remove the symlink before starting the OpenNebula services
to use Ruby gems installed by this script. E.g. execute

    # unlink /usr/share/one/gems

Execution continues in 15 seconds ...

Step 6. Update Configuration Files

If you haven’t modified any configuration files, you can skip this step and proceed to the next one.

In HA setups it is necessary to replace in the file /etc/one/monitord.conf the default value auto of MONITOR_ADDRESS attributed to the virtual IP address used in RAFT_LEADER_HOOK and RAFT_FOLLOWER_HOOK in /etc/one/oned.conf.

Community Edition

In order to update the configuration files with your existing customizations you’ll need to:

  1. Compare the old and new configuration files: diff -ur /etc/one.YYYY-MM-DD /etc/one and diff -ur /var/lib/one/remotes/etc.YYYY-MM-DD /var/lib/one/remotes/etc. You can use graphical diff-tools like meld to compare both directories; they are very useful in this step.
  2. Edit the new files and port all the customizations from the previous version.

Enterprise Edition

If you have modified configuration files, let’s use onecfg to automate the configuration file upgrades.

Before upgrading OpenNebula, you need to ensure that the configuration state is clean without any pending migrations from past or outdated configurations. Run onecfg status to check the configuration state.

A clean state might look like this:

onecfg status
--- Versions ------------------------------
OpenNebula:  5.8.5
Config:      5.8.0

--- Available Configuration Updates -------
No updates available.

Unknown Configuration Version Error

If you get error message about an unknown configuration version, you don’t need to do anything. The configuration version will be automatically initialized during the OpenNebula upgrade. The configuration of the current version will be based on the former OpenNebula version.

onecfg status
--- Versions ------------------------------
OpenNebula:  5.8.5
Config:      unknown
ERROR: Unknown config version

Configuration Metadata Outdated Error

If the configuration tool complains about outdated metadata, you have not run a configuration upgrade during some of the past OpenNebula upgrades. Please note, configuration must be upgraded or processed with even OpenNebula’s maintenance releases.

The following invalid state:

onecfg status
--- Versions ------------------------------
OpenNebula:  5.8.5
Config:      5.8.0
ERROR: Configurations metadata are outdated.

needs to be fixed by reinitialization of the configuration state. Any unprocessed upgrades will be lost and the current state will be initialized based on your current OpenNebula version and configurations located in system directories.

onecfg init --force
onecfg status
--- Versions ------------------------------
OpenNebula:  5.8.5
Config:      5.8.5

--- Available Configuration Updates -------
No updates available.

After checking the state of configuration, in most cases running the following command without any extra parameters will suffice, as it will upgrade the probes based on the internal configuration version tracking of the currently installed OpenNebula.

onecfg upgrade
ANY   : Backup stored in '/tmp/onescape/backups/2020-6...
ANY   : Configuration updated to 6.0.0

If you get conflicts when running onecfg upgrade refer to the onecfg upgrade basic usage documentation on how to upgrade and troubleshoot the configurations, in particular the onecfg upgrade doc and the troubleshooting section.

Step 7. Upgrade the Database Version

Important

Users of the Community Edition of OpenNebula can upgrade from the previous stable version if they are running a non-commercial OpenNebula cloud. In order to access the migrator package a request needs to be made through this online form.

Make sure at this point that OpenNebula is not running. If you installed from packages, the service may have been started automatically. Simply run the onedb upgrade -v command. The connection parameters are automatically retrieved from oned.conf.

Step 8. Check DB Consistency

First, move the 6.0 backup file created by the upgrade command to a safe place. If you face any issues, the onedb command can restore this backup, but it won’t downgrade databases to previous versions. Then, execute the onedb fsck command:

$ onedb fsck
MySQL dump stored in /var/lib/one/mysql_localhost_opennebula.sql
Use 'onedb restore' or restore the DB using the mysql command:
mysql -u user -h server -P port db_name < backup_file

Total errors found: 0

Step 9. Start OpenNebula

Now you should be able to start OpenNebula as usual by running as root:

systemctl start opennebula

At this point OpenNebula will continue the monitoring and management of your previous Hosts and VMs. As a measure of caution, look for any error messages in oned.log, and check that all drivers are loaded successfully. You may also try some show subcommand for some resources to check everything is working (e.g. onehost show, or onevm show).

Step 10. Restore Custom Probes

If you have any custom monitoring probes, follow these instructions, to update them to the new monitoring system

Step 11. Update the Hypervisors

Warning

If you’re using vCenter please skip to the next step.

First we need to let OpenNebula know that the hypervisors can be monitored again, but we still don’t want to deploy any VMs on them until they are fully updated. Instead, we will switch them to DISABLE state:

onehost disable <host_id>

After that, update the virtualization, storage and networking drivers. As the oneadmin user, execute:

onehost sync

Then log in to your hypervisor Hosts and update the opennebula-node packages:

Ubuntu/Debian

apt-get install --only-upgrade opennebula-node-kvm
service libvirtd restart # debian
service libvirt-bin restart # ubuntu

If upgrading the LXD drivers on Ubuntu

apt-get install --only-upgrade opennebula-node-lxd

CentOS

yum upgrade opennebula-node-kvm
systemctl restart libvirtd

Step 12. Enable Hosts

Enable all Hosts, disabled in step 2.

Important

Due to the new monitoring system, the firewall of the Front-end (if enabled) must allow TCP and UDP packages incoming from the Hosts on port 4124.

After following all the steps, please review the corresponding guide:

Testing

OpenNebula will continue the monitoring and management of your previous Hosts and VMs.

As a measure of caution, look for any error messages in oned.log, and check that all drivers are loaded successfully. After that, keep an eye on oned.log while you issue the onevm, onevnet, oneimage, oneuser, onehost list commands. Try also using the show subcommand for some resources.

Restoring the Previous Version

If for any reason you need to restore your previous OpenNebula, simply uninstall OpenNebula 6.0, and install again your previous version. After that, update the drivers if needed, as outlined in Step 11.