* kvm: Add a configuration setting to switch between multicast and evpn VXLAN modes
Using the 'network.vxlan.mode' setting you can switch between the multicast (default) and evpn VXLAN modes on a KVM Agent.
When nothing is configured CloudStack will default to multicast by using the modifyvxlan.sh script in the background.
If this setting is set to 'evpn' the KVM Agent will execute the 'modifyvxlan-evpn.sh' script which will configure the VXLAN
devices for EVPN (usually with FRRouting with BGP) mode.
This removes the need to manually replace a shell script on the hypervisor to switch modes.
Existing environments are not touched by this and it is safe to add this setting a an environment already using EVPN for the
VXLAN deployment.
* Add network.vxlan.mode to agent.properties
Make sure there is an example in the agent.properties file so people
can easily discover this configuration setting exists
Per review on PR #13090:
- Refactor common fence-action case into kvmha-fence.sh sourced by both
kvmheartbeat.sh and kvmspheartbeat.sh (per @sureshanaparti).
- Add 'custom' fence action that invokes an operator-supplied script
(kvm.heartbeat.fence.custom.script, default /etc/cloudstack/agent/
heartbeat-fence-custom.sh) with the heartbeat script name as arg;
falls back to hard-reboot if the script is missing/non-executable
(per @NuxRo).
- Rename canonical action 'reboot' -> 'hard-reboot' for clarity; keep
'reboot' accepted as alias so existing deployments don't break (per
@DaanHoogland). Default behavior unchanged: sysrq-trigger reboot,
required where a stale NFSv3 mount blocks systemctl reboot.
The KVM agent's storage heartbeat scripts (kvmheartbeat.sh and
kvmspheartbeat.sh) hard-code an immediate kernel-level reboot via
'echo b > /proc/sysrq-trigger' when a heartbeat write to primary storage
times out. This bypasses all OS-level shutdown protections, drops every
running VM on the host instantly, and triggers HA cascades onto
surviving hosts.
For NFS shared storage the binary "heartbeat-write-failed = host-is-dead"
heuristic is reasonable. For LINSTOR/DRBD or other replicated local
storage, the same disk serves application I/O, replication I/O and
heartbeat I/O simultaneously - so a transient I/O contention spike can
time out the heartbeat write without the host actually being unhealthy.
The result is false-positive sysrq fencing.
Adds a new agent.properties option:
kvm.heartbeat.fence.action = reboot | graceful-reboot
| restart-agent | log-only
Default value is "reboot" so existing deployments keep their current
behavior. Operators on replicated storage backends can choose a less
destructive action:
- graceful-reboot: 'systemctl reboot' instead of sysrq, allowing VMs
a chance to shut down cleanly
- restart-agent: restart cloudstack-agent only, preserving running VMs
- log-only: log + alert, no automatic action
The existing 'reboot.host.and.alert.management.on.heartbeat.timeout'
boolean continues to function as a complete Java-side bypass.
Refs: https://github.com/apache/cloudstack/issues/13089
This PR introduces the initial implementation of Veeam integration support for KVM in CloudStack by adding a UHAPI-compatible server and image server components.
Veeam Backup & Replication interacts with virtualization platforms using its Universal Hypervisor API (UHAPI). To enable backup and restore workflows for CloudStack-managed KVM environments, this change introduces a UHAPI server that exposes CloudStack resources through a UHAPI-compatible interface.
In addition to the control plane APIs, an image server component is introduced to handle the data transfer operations required during backup and restore workflows.
The integration consists of two main components:
1. UHAPI Server (Control Plane) named CloudStack Veeam Control Service
A lightweight UHAPI server runs inside the CloudStack management server and exposes endpoints under:
/ovirt-engine
- /api - For APIs
- /sso - For authentication
- /services/pki-resource - For certificates
This server provides inventory discovery APIs required by Veeam and translates CloudStack resources into the structures expected by UHAPI.
The server:
- exposes infrastructure inventory
- handles authentication and session tokens
- maps CloudStack resources to UHAPI-compatible representations
2. Image Server (Data Plane) named CloudStack Image Service
A separate image server component is introduced to handle backup and restore data transfer operations.
This component:
- serves disk image data during backup
- receives image data during restore operations
- exposes endpoints used by Veeam worker components
- integrates with CloudStack storage to read and write VM disk data
The separation between both these components server ensures that:
- metadata APIs and control operations remain lightweight
- bulk image transfer operations are handled independently
Signed-off-by: Abhishek Kumar <abhishek.mrt22@gmail.com>
Co-authored-by: Abhisar Sinha <63767682+abh1sar@users.noreply.github.com>
Co-authored-by: abh1sar <abhisar.sinha@gmail.com>
Co-authored-by: Wei Zhou <weizhou@apache.org>
This PR aligns the use of terminology, renaming VM / virtual machine references to 'Instance' and also capitalising the terms Templates, Network, Snapshot, User, Account in CloudStack APIs, error and log messages, events, tooltips, etc. Many typos, grammar and spelling mistakes were fixed, also terms like IPv4, VPN, VPC, etc. were properly capitalised. Some error messages were cleaned for better readability. The test cases, expecting some exception strings were adjusted accordingly.
Here is the wiki page, describing the changes in details:
https://cwiki.apache.org/confluence/display/CLOUDSTACK/Object+Naming+and+Title+Case+Convention
---------
Co-authored-by: Manoj Kumar <manojkr.itbhu@gmail.com>
Co-authored-by: Harikrishna <harikrishna.patnala@gmail.com>
* Add source VM name on virt-v2v migration log entries
* Improve the feedback by displaying the running importing tasks
* Add source VM name prefix on more conversion logs
* Improve listing and also list completed tasks
* Pass extra parameters to virt-v2v if administrator allows via global setting
* Add Force converting directly to storage pool option
* Refactor based on review comments
* Add properties for env vars for the instance conversion
* Add separate component for Import VM Tasks
* applying copilot suggestions from code review
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* Fix importing unmanaged instances due to incorrect internal name
* Add VM prefix on each log operation for conversion
* Log the original VM name instead of the cloned VM in case of cloning
* Allow searching storage pool by UUID after conversion to support SharedMountPoint
* Fix search pools logic
* Improve UI and add checks for force convert to pool parameter
* Support Local storage when forceconverttopool is set to true
* Add config key to for allowed extra params and add validation
* Fix params lists
* Fix compile error
* Remove extra stubbings
* Fix extra params execution
---------
Co-authored-by: Abhishek Kumar <abhishek.mrt22@gmail.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Suresh Kumar Anaparti <sureshkumar.anaparti@gmail.com>
It adds a configuration called create.full.clone to the agent.properties file. When set to true, all QCOW2 volumes created will be full-clone. If false (default), the current behavior remains, where only FAT and SPARSE volumes are full-clone and THIN volumes are linked-clone.
This PR allows attaching of GPU devices via PCI, mdev or VF to an Instance for KVM.
It allows the operator to discover the GPU devices on the KVM host and create a Compute Offering with GPU support based on the available GPU devices on the host. Once the operator has created the Compute offering, it can be used by users to launch Instances with GPU devices.
* [PowerFlex/ScaleIO] Added wait time after SDC service start/restart/stop, and retries to fetch SDC id/guid
* Added agent property 'powerflex.sdc.service.wait' for the time (in secs) to wait after SDC service start/restart/stop
* code improvements
* Cumulative enhancements fix for ScaleIO: MDM add/remove, Host prepare/unprepare, validate Storage Pool can be created in Agent.
- Implemented validation to fail Host disconnect from Storage Pool if there are Volumes attached and SDC client MDM removal requires scini service to be restarted
- Implemented Storage Pool validation by checking whether MDM addresses from configuration file and from memory (using CLI) matches, otherwise file ModifyStoragePool command.
- Introduced configuration key to apply timeout after making MDM changes for ScaleIO: powerflex.mdm.change.apply.timeout.ms (default 1000ms)
- Implemented logic to apply timeout after making MDM changes for ScaleIO in prepare and unprepare logic
- Added detection of MDM removal support via CLI
- If MDM removal support via CLI supported then use CLI, fall back to edit drv_cfg.txt and restart scini instead
Co-authored-by: Suresh Kumar Anaparti <suresh.anaparti@shapeblue.com>
Co-authored-by: mprokopchuk <mprokopchuk@apple.com>
* Management Server - Prepare for Maintenance and Cancel Maintenance improvements:
- Added new setting 'management.server.maintenance.ignore.maintenance.hosts' to ignore hosts in maintenance states while preparing management server for maintenance. This skips agent transfer and agents count check for hosts in maintenance.
- Rebalance indirect agents after cancel maintenance, using rebalance parameter in cancelMaintenance API
- Force maintenance after maintenance window timeout, using forced parameter in prepareForMaintenance API.
- Propagate 'indirect.agent.lb.check.interval' setting change to the host agents.
* rebases fixes
* code improvements, cleanup
* [UI] Set rebalance true by default in cancel maintenance dialog
* Update MS state after executing cluster cmd in the target MS, and some code improvements
* code improvements
* Ensure the host lb algorithm 'shuffle' is applied once before disabling the indirect agent lb check background task
* KVM incremental snapshot feature
* fix log
* fix merge issues
* fix creation of folder
* fix snapshot update
* Check for hypervisor type during parent search
* fix some small bugs
* fix tests
* Address reviews
* do not remove storPool snapshots
* add support for downloading diff snaps
* Add multiple zones support
* make copied snapshots have normal names
* address reviews
* Fix in progress
* continue fix
* Fix bulk delete
* change log to trace
* Start fix on multiple secondary storages for a single zone
* Fix multiple secondary storages for a single zone
* Fix tests
* fix log
* remove bitmaps when deleting snapshots
* minor fixes
* update sql to new file
* Fix merge issues
* Create new snap chain when changing configuration
* add verification
* Fix snapshot operation selector
* fix bitmap removal
* fix chain on different storages
* address reviews
* fix small issue
* fix test
---------
Co-authored-by: João Jandre <joao@scclouds.com.br>
James Peru Mmbono
Wido den Hollander
jmsperu
Abhishek Kumar
Wei Zhou
Suresh Kumar Anaparti
Harikrishna
Daan Hoogland
João Jandre
Edward-x
vladimirpetrov
dahn
John Bampton
Rohit Yadav
Nicolas Vazquez
Bernardo De Marco Gonçalves
Vishesh
Pearl Dsilva