This section describes how to upgrade from Abiquo 4.7.0 or above to the current Abiquo version. Please contact Abiquo Support for further information.
Table of Contents | ||
---|---|---|
|
Numberedheadings | ||
---|---|---|
| ||
Warning | ||
Major upgradeThe upgrade to Abiquo 5.2.0 is a major upgrade with upgrades of third-party software packages that are critical platform dependencies You must make snapshots of ALL servers before you start the upgrade |
Note |
---|
|
Prepare branding
If you are upgrading from Abiquo 4.7.x, the new UI in Abiquo 5.0 requires new branding. See Abiquo branding guide
...
Prevent cloud users from performing VM operations
In the UI in the Infrastructure view, select each physical machine and click Disable
Using the API, set the state of each physical machine to DISABLED
...
Check for operations in progress on the platform
Before you shut down the platform servers you should check that no operations are in progress on the platform.
Check that the Abiquo RabbitMQ queues are empty on the Abiquo Monolithic Server, Abiquo Server or Datanode server
The number of messages in all queues must be 0.
Code Block # rabbitmqctl list_queues messages name
Code Block # rabbitmqctl list_queues messages name Listing queues ... 0 abiquo.am.notifications 0 abiquo.bpm.notifications 0 abiquo.datacenter.requests.ADatacenter.bpm 0 abiquo.datacenter.requests.ADatacenter.virtualfactory 0 abiquo.ha.tasks 0 abiquo.nodecollector.notifications 0 abiquo.pcrsync.messages 0 abiquo.pcrsync.parking-expect-no-consumers 0 abiquo.scheduler.fast.requests 0 abiquo.scheduler.requests 0 abiquo.scheduler.slow.requests 0 abiquo.tracer.traces 0 abiquo.virtualfactory.notifications 0 abiquo.virtualmachines.definitionsyncs 0 abiquo.vsm.eventsynk ...done.
On the V2V Server, check for any active conversions by checking for the V2V or Mechadora processes
Code Block $ ps aux | grep v2v $ ps aux | grep mechadora
When user VM operations are blocked and all of the above checks show that no tasks are running, it is safe to halt the platform.
...
Disable old billing dashboards and delete last bills
In Abiquo 5.2, the billing dashboard feature is now in the core Abiquo platform.
If you are using the billing dashboard scripts from previous versions, disable them and delete last bills before you continue with the upgrade.
On the server where you were running the billing packages, remove the Cron jobs to run the billing scripts.
Code Block # rm /etc/cron.d/abiquo-billing-ec2 # rm /etc/cron.d/abiquo-billing-arm
Delete the last bills from the previous version of the billing dashboards (from the kinton.billing_consolidation and kinton.billing_register tables), see commands below. Remove the bills for Azure and Amazon providers, for the number of months covered by the abiquo.enterprise.property.billing.monthoffset system property, which has a default value of 2 months. Later the new billing dashboards feature will regenerate these bills:
...
Code Block SELECT COUNT(*) FROM kinton.billing_consolidation; SELECT COUNT(*) FROM kinton.billing_register;
...
Back up the main platform elements
To perform a basic backup of the platform, run the following backups:
Note |
---|
Before you begin, stop platform services, and check you have enough space on your destination systems. |
Code Block | ||
---|---|---|
| ||
# On DB server, back up the Abiquo MySQL DBs mysqldump --routines --triggers kinton > kinton.sql-$(date +%Y%m%d_%H%M%S) mysqldump --routines --triggers kinton_accounting > kinton_accounting.sql-$(date +%Y%m%d_%H%M%S) # On the Monitoring server, back up Watchtower DB mysqldump --routines --triggers watchtower > watchtower.sql-$(date +%Y%m%d_%H%M%S) # On the Abiquo API Server or UI server, back up the UI tar cvfz /var/www/html/ui.tgz-$(date +%Y%m%d_%H%M%S) /var/www/html/ui # If you are using SAML authentication with multiple IdPs, on ALL Abiquo API Servers, back up this file cp /opt/abiquo/tomcat/webapps/api/WEB-INF/classes/springresources/security/saml/security-saml-generated-beans.xml /opt/abiquo/tomcat/webapps/api/WEB-INF/classes/springresources/security/saml/security-saml-generated-beans.xml-$(date +%Y%m%d_%H%M%S) # On ALL servers, back up the /opt/abiquo folder tar cvfz /opt/abiquo.tgz-$(date +%Y%m%d_%H%M%S) /opt/abiquo # On Abiquo Monolithic or API Server, Remote Services, or Datanode, back up Redis redis-cli save cp -a /var/lib/redis /var/lib/redis-$(date +%Y%m%d_%H%M%S) |
Note |
---|
If you are using HTTPS-SSL between your Tomcat servers, foreach Tomcat server: From the |
...
Stop platform services
This section describes how to stop platform services on all servers.
Warning |
---|
Before you begin, Check for operations in progress on the platform and wait for all operations to complete. If there are operations in progress, DO NOT STOP the platform services because this can cause serious issues with your platform. |
To stop platform services:
Stop the API on the API server or Monolithic server
Code Block systemctl stop abiquo-tomcat
Stop the UI on the API server or Monolithic server or dedicated UI server
Code Block systemctl stop httpd
StopRemote services server
Code Block systemctl stop abiquo-tomcat
Stop the database on the Monolithic server or Database server or Datanode server.
Code Block systemctl stop mariadb
For a datanode configuration, you will also need to stop the Galera cluster. For more details, see Stop and start HA configuration
Stop RabbitMQ (on the monolithic server or API Server or Datanode)
Code Block systemctl stop rabbitmq-server
V2V server -
You do not need to stop anything because the BPM remote service is run on-demand onlyStop monitoring server
Code Block systemctl stop abiquo-delorean systemctl stop abiquo-emmett systemctl stop kairosdb systemctl stop cassandra
On the monitoring server, check if Cassandra is really dead
Code Block ps auwx | grep cassandra
Get the process number for Cassandra (the first number in the output of the previous command), and kill it. In this example, Cassandra is process
12345
.Code Block kill -9 12345
All processes on platform servers should now be halted.
...
Make snapshots and backups of all platform servers
The upgrade to Abiquo 5.2.x is a major upgrade with upgrades of third-party software packages that are critical platform dependencies
You MUST make snapshots of all servers in your platform before you upgrade to Abiquo 5.2.x.
...
Prepare yum repositories on all servers
First, check that you have the repository URL and credentials.
To upgrade to a version with a patch number of zero, for example, version 5.2.0
Expand | ||||
---|---|---|---|---|
| ||||
|
OR To upgrade to a version with patch number that is not zero, for example, version 5.2.1, enable both repositories:
Code Block yum-config-manager --enable abiquo-base yum-config-manager --enable abiquo-updates
Optionally add your username and password to the Abiquo repos
Code Block yum-config-manager --save --setopt=abiquo-*.username=MYUSERNAME yum-config-manager --save --setopt=abiquo-*.password=MYPASSWORD
Don't forget to use a backslash to escape any shell special characters. For more details, see https://www.oreilly.com/library/view/learning-the-bash/1565923472/ch01s09.html
Clean yum and make cache
Code Block |
---|
yum clean all && yum makecache |
If you did not make snapshots of all servers already, then do this now
Warning |
---|
Major upgradeThe upgrade to Abiquo 5.2.x is a major upgrade with upgrades of third-party software packages that are critical platform dependencies You must make snapshots of ALL servers before you start the upgrade |
...
Upgrade monolithic server
These steps are for a Monolithic Abiquo Server, with API and Remote Services on a single server.
Abiquo will upgrade to new versions of RabbitMQ and MySQL that require a manual upgrade path of dependencies.
Note |
---|
These steps are for a server that contains RabbitMQ AND MySQL |
Remove the Erlang and Galera packages.
Code Block yum remove erlang* galera yum clean all && yum makecache
Install new RabbitMQ and MariaDB server
Code Block yum install rabbitmq-server MariaDB-server
Before you confirm, check that the packages will be installed from the abiquo-base repository
rabbitmq-server: 3.8.2.1
MariaDB-server: 10.4.10.1
Upgrade Abiquo to 5.2.x
Code Block yum update
Before you confirm, check the following packages will be installed from the abiquo-base repository:
jdk: 11.0.6u10
redis: 5.0.7.1
After the update, check versions
Check that the symbolic link to the latest version of Java points to Java 11 on Monolithic Server or Datanode
Code Block # cd /usr/java # ll latest -> /usr/java/jdk-11.0.6+10
Check that the correct Tomcat version will be used on the Monolithic Server
Code Block # cd /opt/abiquo/tomcat/bin # sh version.sh Using CATALINA_BASE: /opt/abiquo/tomcat Using CATALINA_HOME: /opt/abiquo/tomcat Using CATALINA_TMPDIR: /opt/abiquo/tomcat/temp Using JRE_HOME: /usr Using CLASSPATH: /opt/abiquo/tomcat/bin/bootstrap.jar:/opt/abiquo/tomcat/bin/tomcat-juli.jar NOTE: Picked up JDK_JAVA_OPTIONS: --add-opens=java.base/java.lang=ALL-UNNAMED --add-opens=java.base/java.io=ALL-UNNAMED --add-opens=java.rmi/sun.rmi.transport=ALL-UNNAMED Server version: Apache Tomcat/9.0.33 Server built: Mar 11 2020 09:31:38 UTC Server number: 9.0.33.0 OS Name: Linux OS Version: 3.10.0-1160.15.2.el7.x86_64 Architecture: amd64 JVM Version: 11.0.6+10 JVM Vendor: AdoptOpenJDK
On the Monolithic Server in the JDK folder, check certificate migration for AM download. The API Server certificate should be listed for Java 11 with its FQDN. You may need to enter the storepass option to supply the password and you can use the alias option to search for the hostname or FQDN of your server
Code Block # cd /usr/java/jdk-11.0.6+10/lib/security # keytool -list -keystore cacerts -alias (hostname or fqdn)
For example
Code Block sudo keytool -list -keystore cacerts -alias abiquo.example.com
Enable and start the new services
Code Block systemctl enable mariadb systemctl enable rabbitmq-server systemctl start mariadb systemctl start rabbitmq-server
Upgrade MySQL (you might need to specifiy the user and password in the command line):
Code Block mysql_upgrade
The mysql_upgrade step may detect errors that trigger messages such as the following:
Code Block kinton_accounting.ACCOUNT_PERIOD_USAGE_VW Error : Column count of mysql.proc is wrong. Expected 21, found 20. Created with MariaDB 100129, now running 100410. Please use mysql_upgrade to fix this error
These messages are expected and this same mysql_upgrade process will automatically fix these errors in its next stages.
Check that the mysql_upgrade process completes correctly.
Check that your hostname is in your DNS or in your /etc/hosts file
Upgrade the Abiquo API databases
Code Block abiquo-db [-h DB hostname] [-P DB port] [-u user] [-p password] update
If the liquibase update fails with a message similar to the following:
Code Block Liquibase update Failed: Validation Failed: 1 change sets check sum src/X.X.X-XX/kinton-X.X.X.xml::ABICLOUDPREMIUM-XXXX-XxxxxxxxXXxXxxxxxxXxxxxx::XXXXXXXXX is now: 7:ee2fa6e058ec76c7abf801567898917d For more information, use the --logLevel flag
Do the following steps
Clear the database checksums
Code Block abiquo-db clearCheckSums
Retry the above abiquo-db update command.
On the monolithic server, change file owners to tomcat user
Code Block chown -R tomcat /opt/abiquo
On the monolithic server running Remote Services that mount the NFS repository (AM, V2V), change file owners to tomcat user
Code Block chown -R tomcat /opt/vm_repository chmod -R a+r /opt/vm_repository
...
Upgrade monitoring server
This step is for Watchtower monitoring servers in all installations.
To continue using Java 8 (for Cassandra), remove the JDK through the package manager. If you use yum, it will be delete all the dependencies and you will have to reinstall them
Code Block rpm -e --nodeps jdk
Check the jdk to install is version 8
Code Block rpm -qa | grep jdk
Check that Java is correctly installed:
If you execute this command and get that error, please execute the command from the next bullet:
Code Block $ java -bash: java: command not found
Fix it by executing this command, then selecting the option "java-1.8.0-openjdk.x86_64 (/usr/lib/jvm/java-1.8.0-openjdk-1.8.0.292.b10-1.el7_9.x86_64/jre/bin/java)":
Code Block $ sudo alternatives --config java There are 3 programs which provide 'java'. Selection Command ----------------------------------------------- 1 /usr/java/jdk1.8.0_144/bin/java + 2 /usr/java/jdk-11.0.6+10/bin/java * 3 java-1.8.0-openjdk.x86_64 (/usr/lib/jvm/java-1.8.0-openjdk-1.8.0.292.b10-1.el7_9.x86_64/jre/bin/java) Enter to keep the current selection[+], or type selection number: 3
As for the Abiquo DB steps, remove Galera, which will also remove MariaDB
Code Block yum remove galera
Clean yum and make cache
Code Block yum clean all && yum makecache
Install MariaDB
Code Block yum install MariaDB-server
Update Abiquo
Code Block yum update
Check that the Java package is now "javajdk" to use Java 1.8.0u144 with Cassandra:
Code Block # rpm -qa 'javajdk*' javajdk-1.8.0u144-6691.el7.x86_64
Upgrade Monitoring server
If you have a HA datanode that runs the watchtower database, skip directly to Step 10 to update watchtower database
Enable and start the services that were reinstalled
Code Block systemctl enable mariadb systemctl start mariadb
Upgrade MySQL
Code Block mysql_upgrade
The mysql_upgrade step may detect errors that trigger messages such as the following message (note: this example is from the Abiquo database server).
Code Block kinton_accounting.ACCOUNT_PERIOD_USAGE_VW Error : Column count of mysql.proc is wrong. Expected 21, found 20. Created with MariaDB 100129, now running 100410. Please use mysql_upgrade to fix this error
These messages are expected and this same mysql_upgrade process will automatically fix these errors in its next stages.
Check that the mysql_upgrade process completes correctly.
Update the Abiquo Watchtower database
Code Block watchtower-db [-h DB hostname] [-P DB port] [-u user] [-p password] update
...
Upgrade steps from 4.7.x to 5.0.x
Start with your original version and perform the steps to the final version.
Expand | ||
---|---|---|
| ||
Run Redis database script On each Remote Services server:
Post-upgrade steps:
|
...
Upgrade steps from 5.0.x to 5.1.x
Start with your original version and perform the steps to the final version.
Expand | ||||||||
---|---|---|---|---|---|---|---|---|
| ||||||||
Click for Mandatory Run script to update job names on Veeam server When upgrading from 5.0.x or 5.1.x to 5.1.2 or above, if you are using Veeam, run the following application that will update the job names on the Veeam server to allow duplicate VM names. The instructions for running the tool are in the following file.
To run the tool, do these steps:
|
Expand | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|
| ||||||||||
Run mandatory database scripts on Redis database and Watchtower SQL database When upgrading from any version lower than 5.1.2, do the following steps once (e.g. during upgrade to 5.1.2 OR upgrade to 5.2.x) On the Abiquo Server, run the following tool that will generate two scripts to run on the Redis database and Watchtower SQL database. These scripts change the use of the Abiquo VM Identifier (name attribute) as the UID to the ID of the VM as the new UID. The instructions for running the tool are in the following file:
To run the tool and scripts, do these steps:
Run Redis database script On each Remote Services server:
|
...
Upgrade steps from 5.1.x to 5.2.x
Start with your original version and perform the steps to the final version.
When upgrading from 5.1.0 or 5.1.1 to 5.1.2 or above, follow the steps from the Upgrade steps from 5.0.x to 5.1.x block.
Expand | ||||
---|---|---|---|---|
| ||||
Run Redis Script to update Azure VSM subscriptions If you have VMs deployed in Azure, on the Remote Services server, run this script that will use the Redis address and port configured in abiquo.properties. The script will add the VM's Resource Group to the VSM Redis subscription of the VM. We assume that VMs are in the same Resource Group as their virtual datacenter. This is certainly true if the VMs were deployed from Abiquo. In any case, if you previously onboarded a VM in a different RG, the virtual machine definition synchronization process will update it. This script gets the resource group from the VDC provider ID with the format resourceGroup/vdcproviderId, e.g. abiquo-westeurope/abqvnet-vdcname.
|
Expand | ||||
---|---|---|---|---|
| ||||
Configure HTTPS TLS between Tomcat servers If you are using HTTPS TLS between the Tomcat servers, do these steps. For details of this configuration, see Configure Abiquo Tomcat with HTTPS for Remote RS.
|
Expand | ||
---|---|---|
| ||
Run Redis database script These scripts can always be found on the Abiquo Server. On each Remote Services server (including Monolithic sever) or Datanode server run the Redis scripts.
|
Expand | ||
---|---|---|
| ||
Remove deprecated billing dashboard scripts In Abiquo 5.2, the billing dashboard feature has been integrated into the Abiquo platform. If you were using the billing dashboard in a previous version, do these steps.
|
Upgrade steps for 5.2 versions
These steps apply to upgrades starting from version 5.2.0 and above.
Expand | ||
---|---|---|
| ||
Abiquo 5.2.0 or later to latest 5.2.x version
|
...
Configure Abiquo after the upgrade
Before you start the Abiquo tomcat server, add Abiquo configuration properties to the abiquo.properties file.
By default the abiquo.properties file is found in the /opt/abiquo/config/ folder.
See Changes to Abiquo configuration propertiesConfigure the user interface. The default UI location is /var/www/html/ui.
Optional: Add custom labels and translations in the lang_xx_XX_custom.json files in the lang folder
Add custom configuration to client-config-custom.json. See Configure Abiquo UI
If your API is not in the same domain as the UI, set the API endpoint pointing to your Abiquo API server:Code Block { "config.endpoint": "http://myAPIserver/api" }
Reporting changes: To upgrade the Abiquo Reports database for the upgrade to Abiquo 4.7.x+, contact Abiquo Support for the file and procedure.
...
Start Abiquo servers and services
To start the Abiquo platform servers and services, do these steps:
On the Abiquo server, restart the HTTP daemon to refresh the user interface files, and bring up the Tomcat server.
Code Block service httpd start service abiquo-tomcat start
On the Remote services server, start the Tomcat server
Code Block service abiquo-tomcat start
On the Monitoring server do these steps.
Edit the file
/opt/kairosdb/conf/kairosdb.properties
to update the name of the following variable and to remove the port from it:Replace the line
kairosdb.datastore.cassandra.host_list=192.168.888.999:9160
.With this line
kairosdb.datastore.cassandra.cql_host_list=192.168.888.999
. Please note the newcql_
prefix forhost_list
.
Edit the file
/etc/cassandra/conf/cassandra.yml
OR/etc/cassandra/default.conf/cassandra.yaml
(whichever exists) to remove a variable:Remove the line starting with
kairosdb.datastore.cassandra.datapoint_ttl
and save and close the file.
Edit the file
/opt/kairosdb/conf/kairosdb.properties
to add a new variable (ref: Internal JIRA SUP-333):Add
kairosdb.datastore.cassandra.datapoint_ttl = 15768000
On the Monitoring server, start the Cassandra service
Code Block sudo service cassandra start
WAIT about 5 minutes until the service is up and running
Start the KairosDB service
Code Block sudo service kairosdb start
Start the other services in this order
Code Block sudo service abiquo-emmett start sudo service abiquo-delorean start
On the V2V server: restart the Tomcat server:
Code Block service abiquo-tomcat restart
In Abiquo, re-enable the physical machines!
Clear your browser cache to prevent glitches in the UI