This guide is aimed at OpenNebula 5.12.x users and administrators who want to upgrade to the latest version. The following sections summarize the new features and usage changes that should be taken into account, or are prone to cause confusion. You can check the upgrade process in the corresponding section.
In the new OneFlow server, the state poweroff sets the service state to warning.
The information stored vm_info has been reduced to avoid large documents. The information available now is: ID UID GID UNAME GNAME NAME. If you want to retrieve more information you can use the ID, e.g:
client = OpenNebula::Client.new vm = OpenNebula::VirtualMachine.new_with_id(client, id) # Retrieve VM information that previosly was stored in vm_info variable vm.info
When using ready_status_gate, the role will be considered as running when the VM has
READY=YES in the template. This has been implemented using ZMQ, so the old attribute
"running": true has been removed.
MySQL can be configured to use the BINARY clause in SELECT operations. This make object names to be case sensitive (as sqlite or PostgreSQL). The default behavior is to not use this feature to be backward compatible. Note that if you are using LDAP or AD as authentication backends this feature is not recommended.
New configuration option for SQLite TIMEOUT defines timeout in milliseconds for acquiring lock to DB. You may consider increasing this value if you have following error messages in monitor.log or oned.log:
SQL command was: ... error: database is locked
New default restricted attributes¶
HUGEPAGE_SIZE attributes from
TOPOLOGY are now restricted by default.
The monitoring system has been redesigned to improve its scalability and to better support different deployment scenarios. This redesign introduces some incompatibilities:
- Custom probes need to be copied to the new locations, see Monitoring Guide to learn about the new locations.
- In order to speed-up DB access, monitoring and VM/Host data has been separated. The XPATH of some data had to be modified to accommodate the change. You may need to adapt any custom integration to the new XPATHS. In particular:
/HOST/HOST_SHARE/[DISK_USAGE,FREE_DISK,MAX_DISK,USED_DISK]was moved to
/HOST/HOST_SHARE/[USED*, FREE*]was moved to monitoring object
/HOST/TEMPLATE/[NETRX, NETTX]was moved to monitoring object
- No monitor information is sent in listing API call, neither hosts nor VMs.
- Configuration of monitoring probes and parameters has been moved to its own file, you may need to adapt/migrate your custom modifications from
- The monitoring system now may use TCP transport. You’ll need to open incoming connections to port 4124 and TCP to the front-end, in addition to the UDP one.
Accounting and Showback¶
All the states that make Virtual Machines to remain in the host are taken in account to calculate the accounting and showback records. So now, the following states are also computed:
Azure driver was revamped to support Azure Resource Manager, there are new attributes requried to be defined for the Azure host. Please refer to the Azure driver.
New OneGate commands¶
onegate vnet show <ID> and
onegate vrouter show have been added. In order to be able to use them the corresponding permissions needs to be added at
:permissions: ... :vrouter: :show: true :vnet: :show_by_id: true