From 87462ceed214319d2d2505e2fa9c77897c57de49 Mon Sep 17 00:00:00 2001 From: ms4ons Date: Wed, 30 Sep 2026 21:32:27 +0200 Subject: [PATCH 1/4] B OpenNebula/one#8073: Fix migration states Signed-off-by: ms4ons --- .../virtual_machines_operation/virtual_machines/vm_instances.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/content/product/virtual_machines_operation/virtual_machines/vm_instances.md b/content/product/virtual_machines_operation/virtual_machines/vm_instances.md index 6330b776..f938cb56 100644 --- a/content/product/virtual_machines_operation/virtual_machines/vm_instances.md +++ b/content/product/virtual_machines_operation/virtual_machines/vm_instances.md @@ -915,7 +915,7 @@ Datastore Migration allows the transfer of a VM's disk images and associated fil Depending on the state of the VM and the underlying storage drivers, OpenNebula supports two primary methods: -* **Cold Storage Migration**: Performed when the VM is in a POWEROFF or UNDEPLOYED state. The disks are moved physically between datastores before the VM is resumed. +* **Cold Storage Migration**: Performed when the VM is in a POWEROFF or SUSPENDED state. The disks are moved physically between datastores before the VM is resumed. * **Live Storage Migration**: Performed while the VM is RUNNING. OpenNebula coordinates with the hypervisor to mirror disk writes to the new destination in real-time, ensuring zero downtime for the workload. From c415c781da5b48b6ba34d13c836b53e5002f3f78 Mon Sep 17 00:00:00 2001 From: ms4ons Date: Wed, 30 Sep 2026 21:33:16 +0200 Subject: [PATCH 2/4] B OpenNebula/one#8073: Add explanation of VM states for cluster optimization Signed-off-by: ms4ons --- content/product/cloud_system_administration/scheduler/drs.md | 2 ++ 1 file changed, 2 insertions(+) diff --git a/content/product/cloud_system_administration/scheduler/drs.md b/content/product/cloud_system_administration/scheduler/drs.md index 3109d86c..83c363b6 100644 --- a/content/product/cloud_system_administration/scheduler/drs.md +++ b/content/product/cloud_system_administration/scheduler/drs.md @@ -26,6 +26,8 @@ OpenNebula DRS employs an integer linear programming (ILP) solver to optimize Cl A VM might be migrated to another Host or another datastore, but can't change both Host and datastore simultaneously. +Migration recommendations can be made for the VMs in the following [states]({{% relref "product/operation_references/configuration_references/vm_states#vm-states" %}}): `ACTIVE`/`RUNNING`, `SUSPENDED`, or `POWEROFF`. VMs in the `UNDEPLOYED` and `STOPPED` states are not considered. If the Cluster contains a VM in any other state, resource optimization is skipped. + ## Configuration and Usage **To enable OneDRS**: In Sunstone, go to **Infrastructure -> Clusters**, select the relevant Cluster then click **Enable OneDRS** in the **OneDRS** tab. Alternatively, in the Cluster template set the `ONE_DRS` configuration attribute. From d1e7dcf7af4ee9f478b8b09d4fbd3f5dafafd6a9 Mon Sep 17 00:00:00 2001 From: ms4ons Date: Fri, 2 Oct 2026 23:13:19 +0200 Subject: [PATCH 3/4] F OpenNebula/one#7930: Add balancing by storage space Signed-off-by: ms4ons --- content/product/cloud_system_administration/scheduler/drs.md | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/content/product/cloud_system_administration/scheduler/drs.md b/content/product/cloud_system_administration/scheduler/drs.md index 83c363b6..ed148cd9 100644 --- a/content/product/cloud_system_administration/scheduler/drs.md +++ b/content/product/cloud_system_administration/scheduler/drs.md @@ -77,6 +77,7 @@ The Load Balancing policy can combine multiple performance indicators: - **CPU Usage**: Load distribution based on actual CPU utilization of the VM. - **CPU Capacity**: Allocation based on requested CPU (the VM template attribute). - **Memory Usage**: Balancing based on requested memory. +- **Storage Space**: Balancing based on the storage space of the datastores. - **Disk I/O**: Consideration of read/write operations. - **Network Traffic**: Optimization based on network throughput. @@ -89,9 +90,12 @@ ONE_DRS=[ DISK_WEIGHT="0.5", MEMORY_WEIGHT="0", NET_WEIGHT="0", + STORAGE_WEIGHT="0", ... ] ``` +**Note:** Storage space balancing assumes that each datastore represents an independent storage capacity. Datastores sharing the same underlying storage resource report the same total, used, and free space. Therefore, they should be avoided. For accurate balancing, configure datastores so that each one maps to a distinct storage capacity resource. + ### Predictive DRS OneDRS allows balancing based on monitored or predicted values. A **prediction weight** between `0` and `1` determines the influence of forecasts: From 2095b8e605ca63503d47f3dcbb3e4e2fe0b0372f Mon Sep 17 00:00:00 2001 From: mattrowe-opennebula Date: Mon, 5 Oct 2026 12:45:45 +0200 Subject: [PATCH 4/4] Fix link --- content/product/cloud_system_administration/scheduler/drs.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/content/product/cloud_system_administration/scheduler/drs.md b/content/product/cloud_system_administration/scheduler/drs.md index ed148cd9..25c176b3 100644 --- a/content/product/cloud_system_administration/scheduler/drs.md +++ b/content/product/cloud_system_administration/scheduler/drs.md @@ -26,7 +26,7 @@ OpenNebula DRS employs an integer linear programming (ILP) solver to optimize Cl A VM might be migrated to another Host or another datastore, but can't change both Host and datastore simultaneously. -Migration recommendations can be made for the VMs in the following [states]({{% relref "product/operation_references/configuration_references/vm_states#vm-states" %}}): `ACTIVE`/`RUNNING`, `SUSPENDED`, or `POWEROFF`. VMs in the `UNDEPLOYED` and `STOPPED` states are not considered. If the Cluster contains a VM in any other state, resource optimization is skipped. +Migration recommendations can be made for the VMs in the following [states]({{% relref "product/operation_references/configuration_references/vm_states" %}}): `ACTIVE`/`RUNNING`, `SUSPENDED`, or `POWEROFF`. VMs in the `UNDEPLOYED` and `STOPPED` states are not considered. If the Cluster contains a VM in any other state, resource optimization is skipped. ## Configuration and Usage