STRATISYSTEM USER GUIDE
General user guide for StratiSTOR, StratiSERV, and HyperSERV.
STRATISYSTEM OS — GENERAL SYSTEM USER GUIDE — REV 4.10
StratiSYSTEM OS is SteelDome’s universal infrastructure operating system. It provides a common foundation for storage, virtualization, containers, private cloud, AI, and high-performance computing on supported industry-standard x86 hardware.
Services are activated according to the deployment’s requirements. Administrators can deploy dedicated storage, dedicated compute, or HyperSERV, while maintaining consistent provisioning, management, monitoring, and lifecycle practices.
| Product role | Purpose |
|---|---|
| StratiSTOR | Scale-out block, file, and object storage, with data protection and availability policies matched to the workload. |
| StratiSERV | Virtual machines, containers, Kubernetes infrastructure, workload placement, and compute lifecycle management. |
| HyperSERV | StratiSERV compute and StratiSTOR storage operating together on a shared infrastructure foundation. |
Deployment patterns include single-node edge systems, multi-node clusters, independent compute and storage tiers, and appropriately designed multi-site environments. Optional services extend this foundation with OpenStack private cloud, GPU workloads, and parallel storage for AI and HPC.
This guide describes administration of StratiSYSTEM OS, including StratiSTOR storage, StratiSERV compute, HyperSERV, networking, data protection, management services and optional AI and cloud components. Available functions depend on installed services, licenses, supported hardware and configuration. Follow the applicable release notes and advanced guides for specialized procedures.
About the illustrations: The screenshots illustrate the existing management workflows. Labels and available controls can vary with the installed build and enabled services; use the applicable release notes and deployment configuration when following a procedure.
STRATISTOR STORAGE SERVICES
StratiSTOR provides distributed block, file, and object storage for enterprise applications, private cloud, backup, media, and data-intensive workloads. Supported configurations combine scale-out capacity, replication or erasure coding, health monitoring, recovery, and protocol services on industry-standard hardware. Protection and performance depend on the selected topology, backend, and workload.
STRATISERV COMPUTE SERVICES
StratiSERV provides enterprise virtualization and container infrastructure, including VM lifecycle management, host clustering, migration, placement policies, templates, snapshots, backup, and recovery. Supported configurations extend the compute foundation with Kubernetes, virtual TPM, Secure Boot, GPU access, and optional OpenStack private-cloud services.
HYPERSERV DEPLOYMENTS
HyperSERV combines StratiSERV compute and StratiSTOR data services on shared infrastructure. Reserve resources for storage and cluster services before allocating capacity to VMs and containers. Plan sufficient CPU, memory, storage, and network headroom for workload growth, maintenance, and the intended failure scenarios.
DEPLOYMENT PROFILES, RESOURCES, AND AVAILABILITY
Select a deployment profile before assigning hardware. Size CPU, memory, storage, and networking for the enabled services, expected workload, recovery activity, and capacity required during a node outage.
| Deployment profile | Planning requirement |
|---|---|
| Single node | Provide appropriate disk and component protection. Loss of the node interrupts services hosted on that node. |
| Two nodes | Use the supported quorum and fencing design for each enabled cluster service. Validate what remains available after either node fails. |
| Three or more nodes | Distribute control services and data across suitable failure domains and retain capacity for recovery. |
| Dedicated StratiSTOR and StratiSERV tiers | Size storage and compute separately, including the network between them. |
| HyperSERV | Reserve resources for storage and cluster services before allocating the remaining capacity to workloads. |
| Multi-site | Define latency, bandwidth, data placement, quorum location, recovery objectives, and behavior during site isolation. |
Memory planning must include storage processes, metadata services, caching, management services, the host operating system, and guest workloads. Dense storage nodes and GPU or Kubernetes workloads require explicit budgets. A process memory target is a tuning objective and should not be treated as a guaranteed ceiling on total host consumption.
For StratiSTOR, plan raw capacity separately from usable capacity. Replication or erasure coding, metadata, snapshots, uneven device utilization, and recovery reserve all affect the capacity available to applications. Expansion planning should account for the time and resources required to restore protection after a failure.
StratiSTOR monitor quorum requires a majority of the configured monitors. Three monitors require two available members; five require three. Monitor count and total node count are separate design decisions. Quorum availability and data availability must both be satisfied.
An external quorum participant contributes only to the cluster service for which it is configured. Evaluate StratiSTOR, StratiSERV, and Kubernetes availability independently. A witness does not supply missing storage replicas or compute capacity.
HARDWARE EXPANSION AND CAPACITY PLANNING
Scale-up and scale-out can be performed incrementally by adding appropriate compute, storage, or HyperSERV resources. Mixed vendor or processor generations may be useful for capacity expansion, but they are not automatically interchangeable for live migration, GPU passthrough, processor-feature profiles or consistent performance. Validate each new host against the cluster's compatibility matrix and placement policies.
Document the current and post-failure resource budget. For storage, account for usable versus raw capacity, data protection, fullness variance between devices, recovery headroom and rebalance traffic. For compute, include guest memory, storage/management processes, migration targets and any GPU memory reservations. Do not infer zero-downtime operation for every expansion, conversion, or host retirement.
HARDWARE AND RESOURCE PLANNING
| Resource | Planning guidance |
|---|---|
| CPU | Use supported 64-bit Intel or AMD hardware. Size core count, clock frequency, virtualization capabilities, and NUMA topology for storage activity and workload density. |
| Memory | Budget host and management overhead, storage processes, metadata, caching, VMs, containers, and recovery activity. Verify the complete node budget instead of applying one RAM value to every deployment. |
| Storage | Select supported NVMe, SSD, HDD, or external block devices. Define OS-disk protection, data placement, usable capacity, metadata requirements, and recovery reserve. |
| Network | Select redundant physical links, bandwidth, traffic separation, routing, and fabric features appropriate to the workload. Validate client and switch capacity as part of the design. |
| Accelerators | Match GPU, DPU, adapter, firmware, and driver combinations to the supported workload profile. Include device placement and workload mobility restrictions. |
Use the approved sizing profile for the installed release and the Features and Specifications reference. Requirements for compact edge deployments, dense StratiSTOR nodes, StratiSERV compute clusters, and HyperSERV environments differ. Record the sizing assumptions and tolerated failure conditions before deployment.
MAIN MANAGEMENT CONSOLE
LOGIN PAGE
The main management console can be accessed via any standard web browser at https://[any cluster node address]:3199
NOTE: Installation can generate a self-signed management certificate. Verify the endpoint and certificate through the deployment procedure, and install a trusted certificate for production use. Certificate configuration is described in the advanced configuration guide.
Use the administrative account and credentials supplied for this deployment. Replace bootstrap credentials according to the deployment procedure and use individual administrative identities where supported.
MAIN DASHBOARD
Once logged in, you are presented with the main dashboard which provides a snapshot across four different cards for the overall health of the node, CPU and memory utilization, general system information and high-level system configuration.
Each card contains a drilldown link which will provide more specific information about that respective card topic.
SYSTEM ANALYSIS AND OBSERVABILITY
Use the Analysis view, where enabled, to inspect processor topology, memory configuration, virtualization capabilities and system-readiness details. Use the Performance views to review CPU, memory, network and storage trends for the selected host or cluster. Compare historical usage with current service health before diagnosing a resource issue.
When diagnosing a virtual-machine or accelerated-workload issue, correlate host metrics, storage latency, physical-port activity, cluster logs, CPU locality and device health rather than relying on one dashboard card. GPU monitoring may report device discovery, utilization, temperature and readiness when supported by the installed driver and module.
SELECTOR
In the upper left-hand corner, you will find the server selector dropdown. This allows the administrator to toggle between different hosts within the cluster providing easy access to administrative functions for that specific host.
METRICS AND HISTORY
The view metrics and history drilldown provides detailed information related to CPU, memory, disk, and network activity.
Additionally, the graphical display of the metrics can be seen at the bottom. The administrator can select the timeframe that is displayed in the graph by selecting the dropdowns located on the upper left-hand corner of the graph section.
SYSTEM LOGS
The log view displays all system logs in real-time. There are various filters which can be used to narrow the log view to specific messages, event severities or event timeframes.
STORAGE OPERATIONS
On the storage view, the administrator can see the real-time performance metrics for all storage related activity within a selectable time window. All known storage devices will be displayed in the lower pane along with real-time storage related logging.
The action button in the upper right-hand corner of the storage device pane allows the administrator to perform functions such as mapping NFS volumes, creating LVM volumes, attaching iSCSI disks, and much more.
You can also perform individual disk operations such as mounting, formatting, shrinking, growing, deactivating and deleting disk objects.
NETWORK OPERATIONS
On the network view, the administrator can see the real-time performance metrics for all network related activity within a selectable time window. All known network interfaces will be displayed in the lower pane along with real-time network related logging.
From this view, the administrator can modify network interface configurations and create supported network devices such as VPNs, bonds, bridges, and VLANs. Existing installations may also show legacy teaming configurations; use the supported bonding workflow for new deployments.
Additionally, if required, the firewall services can be enabled and rules defined.
NETWORK OPERATIONS (VPN)
If VPN services are required, the administrator can use the native WireGuard VPN capability to establish point-to-point secure high-speed interconnects with other WireGuard systems anywhere in the world over any routable network (see advanced configuration guide for more details).
WIREGUARD VPN VALIDATION
After configuring WireGuard, verify peer identities, allowed address ranges, endpoint reachability, routing and firewall policy. Confirm a live handshake and test the intended application traffic in both directions. Record peer configuration and maintain a separate recovery path for management access during network changes.
NETWORK OPERATIONS (NETWORK BONDING/TEAMING)
If network bonding is required, select a supported bond mode that matches the host, switch, and workload design. Active-backup provides link failover; load-distributing modes have additional topology requirements. IEEE 802.3ad LACP requires the corresponding switch configuration. Refer to the advanced configuration guide for the supported setup procedure.
NOTE: Bonding should be used in place of teaming as teaming is deprecated and will be removed in future releases.
A bond can distribute eligible traffic across links, but it does not guarantee that a single connection uses the combined link rate. Verify the selected traffic pattern and failure behavior. Application multichannel and storage multipathing must be configured and tested for their respective services.
NETWORK OPERATIONS (BRIDGING)
A StratiSERV deployment commonly uses a bridge such as br0 to connect guest interfaces to the intended network. The exact bridge layout depends on the enabled services and deployment profile. Verify management, guest, and cluster dependencies before changing or removing an existing bridge.
Additional bridges can be created as necessary for any and all unused network interfaces available on the node. For more information related to advanced bridging techniques, see the advanced configuration guide for more details.
NETWORK OPERATIONS (VLAN)
If the node is connected to a network which utilizes network trunking, VLANs can be created in order to communicate with specific layer-2 VLAN segments. When a VLAN is created, the system creates a new virtual network adapter which is tagged to the specified VLAN ID. This can be used directly or further attached to a bridge allowing multiple devices to use the VLAN interface.
To configure the VLAN, specify the parent network interface, VLAN ID and name of the VLAN. For more information related to trunks and VLANs, see the advanced configuration guide for more details.
NETWORK OPERATIONS (FIREWALL)
If you require a firewall configuration, the firewall service can be enabled and rules defined on a per cluster node basis.
The first step is to unmask the firewall service in the Services View (from the left pane). Find the firewalld service and using the action button (┆), select Allow running (unmask). Now return to the Networking View and you can now enable/disable the firewall service.
Once enabled, you will be able to add/modify zones and add/delete services for your specific configuration requirements.
NETWORK OPERATIONS — MULTIPLE NICS AND PERFORMANCE
StratiSYSTEM OS supports network designs ranging from compact edge deployments to high-throughput storage, virtualization, AI, and HPC environments. Plan management, application, storage, migration, and overlay traffic according to bandwidth and availability requirements.
For systems with multiple network adapters, record each physical port, negotiated link speed, network role, addressing, routing, and redundancy design. Include NUMA locality and PCIe connectivity when tuning high-throughput workloads.
Bonding, application multichannel, and storage multipathing have different purposes. Explain which mechanism is used by each service and how to verify that traffic uses the intended interfaces. Several IP addresses on one adapter share that adapter’s physical bandwidth.
Performance validation should include physical-interface traffic, errors and drops, CPU utilization, memory pressure, storage throughput, and workload latency during a sustained test. Where supported, evaluate receive queues, CPU affinity, service worker allocation, and NUMA placement as a coordinated configuration.
Use the supported persistent network configuration workflow and verify settings after a reboot. Record any required MTU, routing, bonding, VLAN, or accelerated-network configuration consistently across clients, hosts, and switches.
PERSISTENT NETWORK CHANGES AND TIME SYNCHRONIZATION
Use the supported NetworkManager profile to persist bonds, bridges, VLANs, addresses, routes and MTU choices; avoid ad-hoc interface changes that disappear after reboot. For planned maintenance, verify the management path before bringing down a network profile. Commands such as nmcli connection reload, nmcli connection up <profile> and nmcli device reapply <interface> may be appropriate during troubleshooting when performed with out-of-band access available.
Accurate time synchronization is operationally important for distributed storage, authentication, certificate validation, scheduled backup jobs and MFA. A running time daemon is not proof of synchronization. Verify the synchronized state and upstream reachability if event times, Kerberos authentication or scheduled jobs behave unexpectedly.
VIRTUAL MACHINES
If virtual machine services are enabled on the cluster, you will have a Virtual Machine view listed in the pane on the left (along with a StratiSERV view, which we will cover in more detail later in this guide).
The virtual machines view provides access to all virtual machine related functions such as creating, configuring and operating virtual machines.
VIRTUAL MACHINES (STORAGE POOLS)
When first creating a virtual machine, you must define at least one virtual disk for the virtual machine to use as the base operating system disk. You can also allocate additional disks as necessary for each virtual machine.
The first step is to open the Storage Pools view within the Virtual Machines view. This is located at the top of the Virtual Machines view pane. From here, you will see all available storage pools present on the node.
If you are running virtual machine services along with StratiSTOR storage services, you should see at least one storage pool named “machines”. If you are running StratiSERV independent of StratiSTOR, a default storage pool may be the only one listed. You can create additional storage pools from any available on-prem storage devices including SMB, NFS, iSCSI, NVMeoF, and Fibre-channel based storage devices.
VIRTUAL MACHINES (VIEW STORAGE VOLUMES)
For this example, we will select the machines storage pool and then select the Storage Volume tab to see all available virtual disk images present within the storage pool.
VIRTUAL MACHINES (CREATE STORAGE VOLUMES)
Create a storage volume for the VM operating system or application data. Provide the name and capacity, then select a format supported by the chosen storage backend and intended operations. QCOW2 is a common file-based option; RAW and native StratiSTOR block volumes are supported in the applicable workflows.
When ready, click on CREATE to create the storage volume.
VIRTUAL MACHINES (CREATE VIRTUAL MACHINE)
The next step is to create the virtual machine. To do this, click on the Virtual Machines link at the top of the storage pools pane (or click on Virtual Machine view in the far left pane). Now click on the CREATE VM button on the far right side of the virtual machines view.
The Create Virtual Machine wizard will appear. From this view, provide a name for the machine, installation type (download from online repo and local install media (ISO) are the two most common), the operating system type (which should be auto-detected), storage location (select the machines storage pool), the volume (which is the virtual storage volume you created in the previous step), and the memory allocation.
When ready click on Create and Edit. This will allow for further configuration prior to starting the virtual machine.
STRATISERV GUEST SECURITY — UEFI, SECURE BOOT, AND VIRTUAL TPM
StratiSERV supports modern guest configurations using UEFI firmware, Secure Boot, and virtual TPM capabilities in supported deployments. Select the guest security profile before installing the operating system and verify that the selected guest, firmware, and drivers are compatible.
For Windows deployments, make the supported VirtIO driver media available during installation. Load the appropriate storage driver when Windows Setup requests it, then install the required network and guest integration components. SATA disks and emulated network adapters can serve as compatibility options when required by the installation workflow.
After installation, verify the guest’s boot mode, Secure Boot state, TPM visibility, storage and network drivers, and guest-agent communication. Record the supported guest operating-system and driver versions in the deployment documentation.
Recovery procedures for these VMs must account for virtual disks, the VM definition, UEFI variable state, and virtual TPM state. Verify how the selected backup, restore, clone, and migration workflows handle each item. Retain guest recovery keys through the organization’s approved process and test recovery before relying on it.
Guest Secure Boot, virtual TPM, and confidential computing address different requirements. Confidential computing requires a validated combination of processor capabilities, firmware, host software, and guest support. Use the supported Intel TDX or AMD SEV configuration and verify its operational limitations for the installed release.
ENCRYPTED GUESTS AND RECOVERY STATE
Windows guests may use BitLocker; Linux guests may use LUKS or other supported filesystem encryption. Keep guest recovery material outside the guest and backup repository under a controlled key-management process. File-level restore of an encrypted volume generally requires that the operator supply the appropriate recovery password or key; encryption is not removed from the original guest by a read-only inspection.
A restore of only a guest disk may not be sufficient for a bootable recovery of some guests. Record and recover the VM definition, firmware variables, virtual TPM state and guest encryption/recovery metadata as required by the selected backup mode. Test encrypted-guest recovery before deploying policies widely.
VIRTUAL MACHINES (IMPORT VIRTUAL MACHINE)
Rather than creating a new virtual machine, if you need to import a QCOW2 or RAW virtual machine image from a foreign system, you can perform this by clicking on the IMPORT VM button from the Virtual Machine pane.
Here you will be presented with the Import Virtual Machine wizard. The storage and related virtual machine image must be accessible on the local node in order to properly import an existing virtual machine.
ADDITIONAL MIGRATION SOURCES
In addition to importing local disk images, StratiSERV provides a VMware import workflow and may offer other source-specific import methods when the corresponding integration is installed. Source connection, export and conversion requirements vary. Verify guest firmware, disk controllers, networking and security state before application cutover.
VIRTUAL MACHINES (MODIFY VIRTUAL MACHINE)
Immediately following the creation of a new virtual machine or if you are modifying an existing virtual machine, the virtual machine details view will be shown as seen below.
This will allow the administrator to modify the existing virtual machine configuration for items such as CPU, memory, disk and network.
NOTE: If you are modifying a running virtual machine, some changes require a power down/reboot before the changes can take effect. The interface will indicate this as necessary.
VIRTUAL MACHINES (MODIFY VIRTUAL MACHINE MEMORY)
To change the memory allocation for a specific VM, click on the EDIT link next to the Memory field in the VM overview pane. From here you will be presented with current and maximum memory values allocated to this VM.
Current memory identifies the allocation configured for the guest; it is not a guaranteed minimum reservation. The maximum value defines the applicable allocation limit for the VM configuration. Dynamic adjustment and memory hot-plug depend on guest support and the configured mechanism. Verify that the host has sufficient capacity before increasing an allocation.
VIRTUAL MACHINES (MODIFY VIRTUAL MACHINE CPU)
To change the CPU configuration for a specific VM, click on the EDIT link next to the CPU field in the VM overview pane. From here you will be presented with vCPU maximum allocated value, the current vCPU count (not to exceed the maximum value), the number of sockets, cores per socket, and threads per core the vCPU should be configured with. The standard configuration is to allocate a single socket and the core per socket to match the number of vCPUs defined above.
NOTE: You may need to adjust these vCPU settings to be compliant with specific software licensing requirements within the VM.
VIRTUAL MACHINES (MODIFY VIRTUAL MACHINE WATCHDOG)
A virtual watchdog can trigger a configured action when its timer expires. The guest must have a compatible watchdog driver and an active service that refreshes the timer. Configure and test the guest-side service before relying on the watchdog for recovery.
NOTE: Guest integration drivers, the guest agent, and a watchdog service perform different functions. Installing a driver bundle alone does not prove that watchdog recovery is active. Use the supported guest-specific configuration and test the configured action.
For guest integration drivers, use the version approved for the installed StratiSERV release and guest operating system. The VirtIO Windows download is available for reference; confirm compatibility before replacing an installed driver set.
VIRTUAL MACHINES (MODIFY VIRTUAL MACHINE VSOCK)
The VSOCK feature is a special communications capability that allows the guest VM to communicate directly with the underlying hypervisor host.
NOTE: This feature requires vsock compatible tools such as socat installed in the guest to facilitate the communications channel between the guest and the host.
VIRTUAL MACHINES (MODIFY VIRTUAL MACHINE DISKS)
Within the Virtual Machine overview pane, the administrator can modify the virtual disk configuration for a specific VM. Add and Edit functions are available in the Disk section of this pane.
VIRTUAL MACHINES (ADD VIRTUAL MACHINE DISKS)
To add a virtual disk image to this virtual machine, click on the ADD DISK button. This will display the Add Disk wizard view. Select Create New. Provide the storage pool, name, size, and format for the new virtual volume. Optionally, you can define the cache mode and bus. Generally the default settings will be appropriate for new disk volumes.
NOTE: Windows can use a supported VirtIO storage controller when the matching driver is loaded from the VirtIO driver media during installation. SATA is a compatibility option when required by the guest or installation workflow. For Linux, verify driver support in the selected distribution and installation image before choosing the device model.
VIRTUAL MACHINES (ADD VIRTUAL MACHINE NETWORKS)
To add a virtual network to this virtual machine, click on the ADD NETWORK INTERFACE button. For typical configurations, select Bridge To LAN as the interface type, then the br0 device, and finally the driver model for the virtual interface.
NOTE: Use a supported VirtIO network adapter when the matching guest driver is available. Install or supply the driver for Windows as part of the guest setup workflow. An emulated adapter such as E1000 can be used as a temporary compatibility option. Verify guest connectivity after changing the adapter model.
VIRTUAL MACHINES (ADD VIRTUAL MACHINE HOST DEVICE PASSTHROUGH)
If you have specialized hardware that needs to be dedicated to the guest VM, you can add a host device by clicking on the ADD HOST DEVICE button. You will be presented with USB and PCI based devices which can be mapped directly through to the guest VM.
NOTE: Any devices mapped to the guest VM will be inaccessible by the host machine. Multiple devices may need to be mapped depending on the device selected. Please see the specific hardware manufacturer recommendations for how to properly perform this function.
Direct device assignment can restrict live migration, automatic placement, and recovery on another host. Validate IOMMU grouping, device availability, drivers, and the supported maintenance workflow before assigning a GPU, PCI device, or other host-specific resource.
VIRTUAL MACHINES (ADD VIRTUAL MACHINE SNAPSHOT)
Virtual machine snapshots are a powerful feature which allows an administrator to rollback to a previous point in time in the event a change was made within the VM and needs to be reverted.
To perform a snapshot, the VM can be offline or running. Provide the name and description of the snapshot. If the VM is online, you will also need to specify the memory state path (which should default to the location where the VM is running). When ready, click on CREATE.
It may take a moment to save the memory state to the disk, but the VM is usable during this time.
To revert to a previous snapshot state, simply click on REVERT
To delete a snapshot state, simply click on DELETE
Snapshot support and included state depend on the disk backend, VM configuration, and selected snapshot mode. Verify application consistency, available capacity, and handling of UEFI and virtual TPM state. Snapshots should be paired with a tested backup and recovery procedure.
STRATISERV CLUSTER MANAGER (HOSTS AND VMS)
If you have deployed HyperSERV (StratiSTOR with StratiSERV) or just StratiSERV by itself, you will see the StratiSERV virtual cluster management options in the far left menu pane.

This management plugin provides access to all necessary virtual machine and container clustering support operations.
The main Hosts and VMs view shows the current state of all hosts and virtual machines in the cluster.
NOTE: If StratiSERV is the only service installed, you will see an additional info icon next to the CLUSTER HOSTS title used for opening the license manager.
HOSTS AND VMS — HOST STATUS AND MAINTENANCE
The Cluster Hosts table can be used to perform host based operations such as entering and exiting maintenance mode.

The view can be filtered using the filter line located just above the host table on the left side of the pane.
HOSTS AND VMS - MAINTENANCE MODE
To place a host into maintenance mode, simply click on the hosts’ action button (┆), and select Enter Maintenance Mode. If the host is already in maintenance mode, you will be presented with the option to Exit Maintenance Mode.
Before entering maintenance mode, verify destination capacity, shared storage access, CPU compatibility, network mappings, placement rules, and device assignments. Eligible VMs can be evacuated to other hosts. Workloads that cannot migrate require the supported shutdown or maintenance procedure. Confirm the final state of every workload before servicing the host.

KUBERNETES-AWARE DRAIN AND ROLLBACK
For maintenance on a host running virtual machines or Kubernetes, confirm the availability of destination hosts and review whether workloads can be evacuated or drained safely. Check control-plane quorum, pod disruption budgets and persistent-volume access before starting maintenance. Do not continue if the status of a required cluster service cannot be confirmed.
If a maintenance action fails, review the host and workload state, including VM placement, Kubernetes cordon/drain status and any services stopped for maintenance. Verify that applications and storage have returned to a healthy state before retrying or rebooting the host.
HOSTS AND VMS - VM FUNCTIONS
The Cluster Virtual Machines table can be used to perform VM based operations such as opening a remote console (to a running VM), performing power functions, live migrating a VM, live storage migrating a VM, enabling/disabling HA mode, cloning a VM, resizing a VM disk, removing/deleting a VM from inventory, marking a VM as a template, and editing additional settings.

HOSTS AND VMS - REMOTE CONSOLE
The remote console feature is a break-out window that provides the administrator with console access to the running virtual machine.

HOSTS AND VMS - POWER FUNCTIONS
To modify the power state of the VM, select the appropriate power function from the list of options. The list will always reflect the options available based on the current power state of the VM.

HOSTS AND VMS - MIGRATE VM
To live migrate (or offline migrate) a VM from one host to another, select from the list of available hosts in the cluster and click CONFIRM. The list of available hosts will only include hosts which are online and not in maintenance mode.

STRATISERV CPU COMPATIBILITY, PLACEMENT, AND FAILOVER
StratiSERV supports clustered VM operations including migration, workload placement, high availability, templates, backup, and recovery. Each operation depends on the VM’s storage, CPU profile, device assignments, and destination-host capabilities.
Where supported, configure a per-VM CPU compatibility profile to provide a common CPU feature baseline across eligible hosts. Validate the destination host before migration. The presence of a compatibility setting does not establish that arbitrary processor models or different CPU vendors are interchangeable.
Before evacuating a host, check destination capacity, storage access, network mappings, affinity policies, and attached devices. GPU, PCI, and other direct device assignments can restrict mobility. Use the supported maintenance procedure for workloads that cannot migrate.
High availability can restart eligible workloads after a host failure. Application recovery time includes failure detection, fencing where required, VM restart, and guest application recovery. Live migration, host-failure recovery, and application-level availability should each have separate validation criteria.
For VM replication, select a supported source and target combination and define the synchronization schedule, consistency method, recovery-point objective, and recovery procedure. Recovery to another platform may require disk conversion, driver changes, or VM reconfiguration. Run a recovery test on each intended destination.
DEVICE-ASSIGNMENT AND GPU READINESS
Use the available GPU inventory and health views to check supported NVIDIA or AMD device detection, driver state and hardware health. Device discovery alone does not establish vGPU licensing or live-migration support. Verify hardware, firmware, host and guest drivers, IOMMU settings and resource reservations before enabling GPU passthrough or GPU-backed containers.
Before changing GPU or device assignments, verify that no running workload depends on the affected device. Shut down or migrate eligible workloads and use the supported host-maintenance procedure for changes that cannot be performed online.
HOSTS AND VMS - MIGRATE VM DISKS
To migrate the VMs’ virtual disk(s), select the disks from the left side of the panel and the destination location on the right side of the pane.

NOTE: For the shared-storage migration workflow, the destination storage must be available to each eligible host with the required access permissions. Storage migration and host migration have separate prerequisites. Verify storage, network, CPU, and device compatibility for the specific operation before proceeding.
HOSTS AND VMS - CLONE VM
To clone a VM, ensure the VM is powered off. In the Clone VM view, provide a unique name for the VM. When ready to begin the clone process, click the CLONE button.
Cloning can be performed on VMs which utilize QCOW2, RAW, or StratiSTOR (RBD) backed disks.

NOTE: If you are cloning a VM that has not been sysprep’d and the original VM is powered on while the clone is powered on, there is a possibility of identity collision on the network (i.e. hostname and/or IP address).
Prepare a unique guest identity for the clone, including the operating-system identity, hostname, network configuration, and any guest security state required by the supported cloning workflow. Verify application licensing and recovery behavior before placing the clone into service.
HOSTS AND VMS - CONVERT VIRTUAL DISK FORMAT
To convert a VM’s disk image, first shut down the VM. Then select the VM, choose Convert VM, and open the Convert Virtual Disk Format wizard. In the wizard, select the disk or disks to convert and choose a destination format—QCOW2, RAW, VHD, VHDX, VMDK, or RBD—then review the summary and confirm.

After conversion, VHD/VHDX/VMDK images are exported to the location you specify and the VM’s configuration is not changed. For QCOW2/RAW/RBD, the VM is updated to use the new image (the original disk reference is replaced).
If converting to RBD, the wizard writes to the selected StratiSTOR pool/image name; for file-based formats, it writes to the filesystem path you provide. Ensure you have adequate free capacity at the destination.

HOSTS AND VMS - RESIZE VM DISKS
To resize a virtual machine disk image, first ensure that the VM is powered off. Next, select the virtual volume to be resized, provide the new desired virtual volume size with units (i.e. G for gigabytes, T for terabytes). When ready to proceed, click the CONFIRM button.

HOSTS AND VMS - REMOVE VM FROM INVENTORY
The remove VM options allows you to either unregister or delete the VM. Unregistering the VM will remove it from VM inventory but leave the virtual disks unaltered. Deleting the VM will remove the VM from inventory and delete all associated virtual disks attached to the VM.

HOSTS AND VMS - CONVERT VM TO TEMPLATE
Before designating a VM as a template, prepare the guest with the supported operating-system generalization procedure, such as Sysprep for Windows. Remove deployment-specific identity where required and verify the supported handling of firmware and virtual TPM state. Power off the VM before converting it to a template.

The conversion process is immediate and the VM will be removed from the VM inventory pane and will now be visible from the Template View.
HOSTS AND VMS - VM SETTINGS
There are additional VM settings which can be enabled or disabled based on the VM configuration requirements. They can be applied with the VM powered on or off. However, if the changes are made while the VM is powered on, you will need to shutdown the VM completely and then power it back on in order for the changes to take effect.

- The Guest Agent Channel Support is used for enabling the virtio guest agent to communicate and exchange information with the hypervisor such as IP address and liveliness conditions. [RECOMMENDED]
- The High Resolution Video support enables guest VMs with full GUI based operating systems to run at resolutions as high as 4K. [OPTIONAL]
- Confidential Computing uses supported Intel TDX or AMD SEV capabilities to protect guest execution and memory. A validated combination of processor, firmware, host software, and guest support is required. Review migration, device, backup, and recovery limitations for the selected configuration. [OPTIONAL]
CONTAINERS
If container (Kubernetes) services are enabled on the cluster, you will see the Containers tab at the top of the StratiSERV Manager view.
NOTE: The info icon beside the container-management title opens the dedicated management endpoint. If certificate trust prevents the embedded view from loading, verify that endpoint and establish trust through the approved certificate procedure. Return to the embedded view and refresh it after correcting the trust configuration.
Once logged into the container management interface, select the local or remote environment registered with that management service. Available operations depend on the environment type, permissions, and enabled integration.
STRATISERV KUBERNETES INFRASTRUCTURE
StratiSERV provides Kubernetes infrastructure for containerized applications, with StratiSTOR-backed persistent storage available through the supported storage integration. HyperSERV deployments can host compute and storage services within the same environment, subject to resource and availability planning.
Plan control-plane and worker roles, API access, cluster DNS, pod and service address ranges, ingress, load-balancer addresses, and persistent storage before deploying applications. Keep application, management, storage, and overlay networks consistent with the site’s addressing and isolation requirements.
Validate the Kubernetes environment before placing application workloads into service:
- Verify control-plane and worker health and access to the supported management interface.
- Configure namespaces, workload permissions, resource requests, limits, and scheduling rules.
- Select an appropriate StratiSTOR-backed storage class and verify its access modes, reclaim policy, and protection behavior.
- Deploy a test application and confirm service access, DNS resolution, and persistent data across a pod restart.
- Validate planned node maintenance and recovery behavior for stateful workloads.
Select the intended local or externally managed cluster before making changes. Verify which backup process protects application data and which protects Kubernetes configuration and control-plane state.
CONTAINER RUNTIMES AND STATEFUL APPLICATIONS
Choose an approved Kubernetes distribution and compatible container runtime. Docker and Podman can coexist on StratiSYSTEM hosts for different purposes; installation or update tooling should not remove an existing container runtime required by another platform service, including StratiSTOR orchestration. A Ceph-backed persistent-volume class must be tested for snapshot, expansion and recovery behavior before production deployment.
During a node drain, confirm that disruption budgets, volume access modes, local data, control-plane quorum and stateful workloads can survive the planned placement change. Cluster VM HA and Kubernetes application availability are distinct recovery mechanisms.
OPENSTACK PRIVATE CLOUD
StratiSYSTEM OS supports an optional OpenStack private-cloud layer for environments that require tenant projects, self-service infrastructure, quotas, and cloud APIs. StratiSERV provides the compute foundation and StratiSTOR supplies supported storage services within the selected cloud architecture.
An OpenStack deployment introduces additional service, networking, and lifecycle requirements. Define controller and compute roles, identity services, image storage, block storage, tenant networking, external connectivity, endpoint names, certificates, and API availability as part of the deployment plan.
Before making the cloud available to tenants, validate project access, image upload, instance launch, network connectivity, persistent-volume attachment, instance deletion, and resource accounting. Confirm the health of the cloud services as well as the underlying infrastructure.
Use the management interface responsible for the selected cloud workload or resource. Coordinate tenant operations, infrastructure maintenance, and recovery through the appropriate management plane.
TEMPLATES
If there are any VMs which have been designated as templates, they will appear in the Current Templates table in the left pane. Clicking on the template will display the template details in the right pane.

There is an action button (┆) associated with each template which allows the administrator to either deploy a new VM from the template or to unregister the image as a template.
Unregistering the VM as a template will remove it from the templates table and return the VM back to the standard VM inventory which can be accessed from the HOSTS AND VMS tab.
DEPLOY FROM TEMPLATE
If deploying from a template, you will be asked to provide a unique name for the VM that is created from this template. When ready to proceed, click on the DEPLOY button.
Deploying from template supports file and block-based (RBD) disk image types.

PERFORMANCE
The performance view provides a real-time cluster-wide view of performance for CPU, memory, disk, and network. The top center of the performance view shows the total number of CPU cores and threads, RAM and whether resource scheduling is enabled.
In the upper-right corner of the performance view, there is a server selector which allows the operator to select a specific host in the cluster. This will filter the performance data shown in the graphs to this specific host.

PERFORMANCE - RESOURCE SCHEDULER
Next to the Cluster Performance title there is an info icon (i). Clicking on this icon will open the Distributed Resource Scheduler pane. The administrator can enable the scheduler to monitor the cluster hosts for CPU and/or memory pressure. If CPU and/or memory pressure thresholds are maintained for a period defined by the aggressiveness value, the cluster will automatically start balancing the VMs workloads out to remove the pressure.

The aggressiveness value determines the time frame by which the pressure is evaluated. Low is over a ten (10) minute period. Medium is over a five (5) minute period and High is over a one (1) minute period.
Resource scheduling remains subject to host eligibility, available capacity, CPU compatibility, storage access, affinity policy, and device-assignment restrictions. Review workloads that cannot move, and reserve sufficient capacity for the intended host-failure and maintenance scenarios.
SHARED DISKS
Shared disks are similar to VMware’s VMFS datastores. They are used when external block storage devices are attached to the cluster hosts such as from a Fibre-channel or iSCSI storage array.
NOTE: If block devices are currently attached and the table is empty, press the RESCAN button to scan all systems for devices. If the table remains empty, then investigate the host connection to the array.
SHARED DISKS - CREATE FENCE DEVICE
The fence device is used by the cluster software to determine the appropriate actions to take in the event of a failure.
NOTE: It is highly recommended to create a fencing device for each external block storage device/array that is being used as a shared disk (a small 1GB disk volume is sufficient).
Fencing prevents an isolated or failed host from continuing unsafe access to shared resources. Validate the complete fencing and quorum design for the deployment. Merely allocating a small shared disk does not establish that fencing is configured and operational.
SHARED DISKS - CREATE FILESYSTEM
When at least one fence device exists, the administrator can create a new filesystem. This filesystem is synonymous with a VMware datastore where the VMs will reside.
Click on the action button (┆) for the block device you want to use as a datastore for VMs and select Create Filesystem.
SHARED DISKS — NAME AND CREATE THE FILESYSTEM
To create the cluster filesystem, provide a unique name for the filesystem. The creation process will take 1-2 minutes to complete depending on the size of the volume being prepared.
Once the filesystem creation process has completed, the new filesystem will appear as a storage pool where VMs can now be created and/or migrated to.
SHARED DISKS - STATUS INDICATORS
The green colored host names indicate the hosts which are currently using the block device.
The magenta colored host names indicate the hosts which are not currently using the block device.
If there is a block device which contains both green and magenta host names, this indicates that a block device is operational and in use by the green colored hosts, but the magenta colored hosts have not yet been configured to use that device. In this case, you will see an additional action item named Extend Filesystem. Selecting this option will activate the block device on the remaining hosts.
EXPANSION OF CLUSTERED FILESYSTEM CAPACITY
Supported shared-disk deployments can extend cluster filesystem access to additional eligible hosts and may support online GFS2 capacity growth once the underlying shared LUN is safely expanded. This is distinct from merely adding an access path to another host. Validate the filesystem, shared-device identity, multipath consistency, fencing, free space and active mount state on all participating nodes before using an Extend or Grow action.
Do not use ordinary local-filesystem growth commands or independently format the shared LUN on each host. Consult the release-specific shared-disk procedure and back up critical guest data before changing its layout.
AFFINITY GROUPS
Affinity groups can be defined which contain two or more VMs. Affinity groups ensure that all VMs in that group will remain together on the same host.
NOTE: The affinities will be maintained even if resource scheduling thresholds are exceeded.
Anti-Affinity groups can be defined which contain two or more VMs. Anti-Affinity groups ensure that all VMs in that group will remain on the different hosts as long as there is a sufficient number of operational hosts to maintain the separation.

BACKUPS
The backup facilities provide a convenient way to protect the VM data without the use of other third-party software.
At least one backup repository needs to be defined before a backup job can be created. The backup repository is any storage location (local or remotely attached) that the hosts are currently configured to use. This can be block, file or cloud based storage. It is highly recommended that all hosts have access to the same location so that backups can occur no matter which host a VM happens to reside on. If the repository is only available on certain hosts, this will be indicated by the Online column showing PARTIAL instead of YES.

Define the intended recovery point, recovery time, application-consistency method, and protected failure scenarios. A complete VM recovery plan must account for disks, the VM definition, and any required firmware or virtual TPM state. Confirm what the selected backup workflow includes and test restoration.
BACKUP - REPOSITORIES
To add a new repository click on the ADD REPO button. Provide a unique name and location for the repository. When ready, click ADD to complete the process.

NOTE: Supported repositories can use attached block storage, file services, or supported cloud-backed storage. Provider support and authentication methods depend on the installed connector and release. Validate repository access, capacity, performance, retention, and recovery from every host that may run the protected VM.
BACKUP - ADD NEW BACKUP JOB
Now that a backup repository exists, you can create a new backup job by clicking on the ADD JOB button. This will invoke the new backup job view.

The backup job view consists of a unique name for the job, at least one VM to backup, the backup schedule, the backup retention period, the backup repo to use for this job, and backup options such as purging and notification settings.
JOB EXECUTION, ELIGIBILITY AND NEW VM DISKS
Use Run Job to start an enabled backup job immediately without changing its schedule. Disabled jobs must be re-enabled before execution. Check the job status and completion result to confirm the backup started and finished successfully.
When changing a schedule (including short intervals such as 15 minutes), inspect the next run, service health and logs, then verify at least one scheduled execution. A saved schedule alone does not demonstrate that the scheduler is running. Scheduled and manually initiated runs should use the same eligibility checks and repository permissions.
A VM may gain another virtual disk after its first backup. Review the job's VM and disk inventory and verify the new disk is included in the next applicable full/incremental recovery point. Do not assume older restore points contain disks that did not exist when they were taken. Review backups after controller changes, VM relocation or storage-format conversion.
BACKUP - BACKUP JOB
Once at least one backup job exists, the job will appear in the backup jobs table. Each job has an action item which allows the administrator to run the job immediately, edit the job, disable the job scheduler, and remove the job.

BACKUP - BACKUP JOBS
The bottom pane in the Backup view is a tree view outlining the details about each protected VM along with details about each disk in the VM.
The Refresh Backups button polls the system for any new updates which may have occurred since the last time the view loaded.
BACKUP CHAIN HEALTH AND SAFEGUARDS
In the backup view, review the repository, protected VM, available restore points and disk inventory before deleting or expiring data. Preserve required full and incremental chains. Do not select a standby replication destination as an independent production backup source unless that is an intentional part of the protection design.
If a background backup completes but the UI shows stale state, refresh the manifest or backup list and check the authoritative job record. Investigate repository access, scheduler status, disk discovery, retention and verification results rather than deleting backup data to resolve a display issue.
BACKUP - RESTORE A VM
Selecting a VM disk in the tree view will allow the administrator to perform two additional functions. The first is the Restore VM option. Selecting this option will present the administrator with a location selector for the restored VM files.
The second option is to verify the backup integrity for a specific disk.
After restoring a VM, verify boot, guest drivers, network identity, application function, and access to persistent data. Include guest security-state recovery where applicable. Perform recovery tests in an isolated environment and record the observed recovery time and recoverable data point.
RESTORE VIRTUAL MACHINE — MODES AND VALIDATION
Open Restore Virtual Machine and choose the recovery operation appropriate to the restore point and target. Available choices can include complete VM recovery, recovery to another location, file/folder recovery and restore-point verification. Confirm the source, destination, data being overwritten and applicable guest security state before starting the operation.
| Recovery task | Operator checks |
|---|---|
| Recover the original VM or a new VM | Source restore point, complete disk set, destination path/name, VM inventory conflicts, firmware/vTPM and guest identity |
| Recover to an alternate VM or target | Target selection, disk mapping, power state, existing target data, overwrite confirmation and rollback plan |
| Restore files or folders | Guest disk/partition, filesystem and encryption detection, selected guest paths, host output directory, name conflicts |
| Verify a restore point | Integrity result and remaining required application/boot tests; file-level verification is not proof of bootability |
SELECTIVE GUEST FILE AND FOLDER RECOVERY
Choose the restore point, select Restore files or folders, inspect the guest volume and use the guest-path browser to select individual files or entire directories. The browser reads backup content without booting the protected VM. Choose a destination directory on the StratiSERV host and review the conflict/overwrite option before starting. The guest path is the recovery source; the host output directory is the recovery destination.
For standard NTFS and supported Linux filesystems, validate the volume and browse the directory tree. For BitLocker-encrypted Windows volumes or LUKS-encrypted Linux volumes, the interface may request the recovery key or passphrase, then reload the accessible volume view. The exact unlock modes depend on the installed backend. Do not save sensitive keys in logs, tickets or unprotected local files.
If the restore reports partial success, examine the failed path list and the associated filesystem or I/O errors. Files recovered successfully may remain in the selected output directory; do not regard the operation as complete until every required file has been checked and any failures addressed.
VM REPLICATION AND DISASTER RECOVERY
StratiSERV VM replication supports scheduled copying of eligible virtual machines to designated recovery infrastructure in configured deployments. A replication job is not the same as a backup job: it maintains a recovery copy and must have an explicit recovery process, conflict policy and consistency/RPO target. Validate application-level recovery rather than assuming that a transferred disk alone meets the business objective.
- Identify the source VM, destination host or cluster, storage mapping and required recovery point.
- Inspect the destination VM inventory before the first job. A same-named or unmanaged target VM can prevent creation; do not delete a target automatically.
- If a replacement option is offered, use it only after confirming that the identified destination VM may be overwritten. Replacement can permanently remove the destination VM’s existing data.
- Choose and verify the replication schedule. After an immediate run or scheduled cycle, review the status, errors, destination VM state and copied disk inventory.
- Exclude recovery replicas from ordinary source-backup selection where supported, so the destination does not recursively become a new protected source.
- Test a controlled recovery or failover in isolation, including network identity, encryption keys, VM definition, application state, reverse replication and return-to-service.
VM IMPORT
The Import tab allows the administrator to import VMs from other third-party platforms such as VMware and Microsoft Hyper-V directly into StratiSERV.

VM IMPORT — VMWARE AUTHENTICATION
To perform the import process with a VMware platform, click on the Authenticate button. You must provide the credentials that will be used to connect to either the VMware ESXi host or vCenter Server.
VM IMPORT — VMWARE PREREQUISITES
Once authenticated, the list of all known VMs will appear in the table below. The VMs must be powered off, have no snapshots in place, and the ESX host must have SSH enabled. The import option will not be available until all three conditions are met.
Once the conditions are met the Import VM option will appear in the action button (┆) for the given VM.
VM IMPORT — SELECT THE VMWARE VM
To import a VM, click on the action button (┆) for the specific VM you want to import and select Import VM.
VM IMPORT — VMWARE DESTINATION AND TRANSFER
The final step is to provide the ESX host root credentials. You can test the credentials by clicking on the TEST CONNECTION button. Select the location for the imported VM and click the IMPORT button when ready to proceed.
Following import, verify firmware mode, guest drivers, storage and network mappings, application function, and any required guest security state. Keep the source recovery option available until the imported workload has passed its validation and cutover checks.
AI, GPU, AND HIGH-PERFORMANCE COMPUTING
StratiSYSTEM OS combines supported GPU compute, high-speed networking, and data services for AI training, inference, scientific computing, analytics, and rendering. Select the compute and storage architecture according to the application’s throughput, latency, capacity, and data-protection requirements.
For GPU workloads, select dedicated device assignment, a supported shared-GPU configuration, or container-based GPU access according to the application and deployment profile. Record the compatible GPU models, firmware, drivers, software stack, and any required vendor licensing. Validate resource allocation and workload operation after installation.
Supported deployments can integrate NVIDIA GPU platforms, DOCA, and BlueField acceleration. Supported accelerator configurations can include NVIDIA vGPU and GPUDirect, AMD ROCm, and AMD MxGPU capabilities. Each requires an explicit hardware and software compatibility profile.
For HPC, supported Lustre deployments provide parallel filesystem capabilities for workloads requiring coordinated access to large datasets. Before placing a Lustre deployment into service, verify client access, data and metadata capacity, network design, monitoring, availability, and the recovery procedure for the selected implementation.
StratiSTOR data services and supported Lustre configurations address different workload designs. Validate the complete path between storage, network, CPU, and GPU when selecting an accelerated configuration. A GPU or high-speed adapter alone does not establish that every application uses an accelerated data path.
AI MODEL DEPLOYMENT AND INFERENCE SERVICES
Selected StratiSYSTEM installations can use the SteelDome AI Deployment Tool (deployai) to prepare AI inference resources, obtain and verify compatible model files, build or register the supported inference runtime, and operate a browser-based AI workspace. This is an optional, separately configured module, not evidence that every cluster includes a ready-to-run AI service.
| Component | Role in the data path |
|---|---|
| Operator and WebUI | Selects models, submits prompts and manages conversation/knowledge workflows through authenticated HTTPS access when enabled |
| Model service (llama.cpp) | Loads a compatible GGUF model and exposes an inference API; each served model can retain a separate API endpoint |
| StratiSTOR/shared filesystem | Stores model weights, downloaded artifacts, knowledge files and deployment state as configured; shared storage is required for the supported distributed-RPC workflow |
| Optional web search, RAG and memory | Retrieves external search results and approved internal documents, and provides persistent context according to configured privacy controls |
| CPU/GPU workers | Execute model inference using eligible hardware and the selected single-node, independent-node or distributed mode |
CHOOSING THE DEPLOYMENT MODE
Single-node inference: Run a model service on one StratiSYSTEM host with local or shared model storage. Independent multi-node inference: Run separate model-service instances on selected hosts; these do not automatically form an HA service. Distributed native RPC inference: For supported CPU-only llama.cpp configurations, a single inference request can use cooperating nodes. Configure reachable worker endpoints, compatible runtime binaries and shared model storage. This mode does not imply distributed GPU inference.
MULTIPLE MODELS IN ONE WEB UI
Configure multiple supported model endpoints in a single Open WebUI instance to make the models available in one selection interface. Confirm that each model endpoint responds before making it available to users. Existing WebUI credentials, conversations, TLS and configured knowledge services remain associated with the primary WebUI instance. Adding a model endpoint does not merge databases from separately deployed WebUI instances.
CONTAINERS, SEARCH, RAG AND SECURITY
Supported setups offer Podman or Docker execution. --container-runtime selects the runtime; --container-install off means that the selected runtime must already be installed and should not be changed by the AI deployment. A system may retain both Podman and Docker, including on hosts running StratiSTOR services.
Optional services include web search, retrieval-augmented generation (RAG), persistent memory, an API gateway and browser TLS. Incremental RAG synchronization can add/update changed sources while retaining previously indexed material absent from the current source set; explicit mirror behavior can remove entries no longer in source. Review the configured mode before large document updates. Secure credentials and use a trusted certificate for any externally accessed endpoint.
For AI deployments with sensitive data, document model provenance, resource limits, allowed search sources, knowledge-base access, web-interface permissions, backups and log retention. For a complex model, consult the dedicated StratiSYSTEM AI Deployment User Guide rather than inferring an exact command line from this general guide.
STRATISTOR CLUSTER MANAGER (MAIN DASHBOARD)
The StratiSTOR Management plugin provides access to all necessary storage clustering support operations.
The main Cluster Status view shows the current state of the local node and all cluster-wide services, storage manifest, and any health related notifications in real-time.
Review service and device availability, monitor quorum, data protection state, recovery progress, metadata-service health, and capacity distribution. An available management screen does not by itself establish that every data service is healthy. Resolve sustained inactive data states or blocked recovery before planned maintenance.
LICENSING MANAGER
The license manager can be located by clicking on the info icon on the Cluster Raw Capacity line in the status table.
NOTE: If you have deployed HyperSERV (StratiSTOR and StratiSERV), the license manager is only available on this tab. This license manager handles both products.
PRODUCT LICENSING AND SERVICE ACTIVATION
StratiSYSTEM OS licensing follows the deployed product role. Supported capabilities are included within that role without separate per-feature licensing.
| Product | Licensing basis |
|---|---|
| StratiSTOR | Total raw storage capacity managed. |
| StratiSERV | Total compute node count. |
| HyperSERV | Managed raw storage capacity and compute node count. |
An optional service is a deployment choice. Activation still requires the appropriate hardware, resources, supported configuration, and any independently applicable third-party software licenses.
STRATISTOR DATA SERVICES AND PROTECTION
StratiSTOR provides block, file, and object storage for supported deployment configurations. Select the access protocol, storage backend, and protection policy together so that availability, performance, and data-management behavior match the application.
| Service | Access examples | Configuration and validation |
|---|---|---|
| Block | iSCSI and NVMe-oF | Volume creation, initiator access, path redundancy, permissions, and recovery behavior. |
| File | SMB and NFS | Filesystems, shares or exports, identity mapping, access rules, locking, and client requirements. |
| Object | S3-compatible endpoints | Endpoint configuration, credentials, buckets, policies, and supported API behavior. |
| Web and transfer | HTTPS and SFTP | Service scope, authentication, certificates, permissions, and permitted filesystem paths. |
Protocol availability and behavior depend on the enabled service and backend. Verify supported combinations before exposing the same dataset through multiple services.
Choose replicated or erasure-coded data protection according to workload and failure-domain requirements. Verify metadata protection separately where applicable. Pool planning should cover placement, usable capacity, recovery reserve, and the supported placement-group sizing or autoscaling method.
Configure SDCache, compression, metadata placement, and encryption according to the supported storage backend. Account for their resource requirements and verify that the intended settings are active. Do not imply that a capability in one backend automatically applies to every storage service.
Use StratiSTOR health information to monitor service availability, protection state, capacity pressure, recovery progress, and device errors. Investigate sustained degraded or inactive data states before planned maintenance. Temporary maintenance settings must be recorded and returned to their intended operating state after the work is complete.
MONITORING DATA PLACEMENT, REBALANCING AND RECOVERY
During normal operation and after adding, removing or replacing a storage device, monitor cluster health, OSD/drive utilization, placement-group states and recovery activity. Large per-device fullness differences, stuck recovery and stalled reads require investigation before aggressive manual weight changes. Use the supported management interface and the documented StratiSTOR procedures to assess the root cause.
For service validation, review monitor quorum, metadata-service availability, pool protection settings, client I/O and the state of SMB/NFS/S3 gateways used by applications. Where used, inspect data and metadata pools separately: an available filesystem name does not guarantee that every metadata or data service is healthy. Rebalancing after an expansion can affect latency and recovery headroom; plan capacity accordingly.
SDCACHE LIFECYCLE AND BACKING STORAGE
Where enabled, SDCache applies configured memory and/or flash tiers to supported workloads. Review cache-to-backing-device relationships, persistent job state and shutdown sequencing during node maintenance. Do not treat a cache as a separate data-protection copy; data durability still depends on the protected backing storage and the selected cache mode.
BLOCK SERVICES
The Block Services tab provides access to all block storage pools and devices created from the pools. To add a new block device pool, click on the ADD POOL button.

NOTE: Block pools can be renamed and deleted.
NOTE: Block devices can be renamed, deleted, mapped to a virtual machine (if VM services are enabled), and mapped to the local cluster node.
BLOCK SERVICES - ADD BLOCK POOL
The Add Block Pool view allows the administrator to define the pool name, Data Pool Placement Group (PG) count, Metadata Pool Placement Group (PG) count (if using erasure-coding), protection scheme (either replicated or erasure-coded), and the replication factor (if using replicated protection scheme).

NOTE: If using erasure-coding for the data pool, the metadata pool must be a replicated pool and will be created as such automatically.
NOTE: If using erasure-coding for the protection scheme, the cluster is set with a default EC profile as shown in the view. If this needs to be changed to a different scheme definition, you must perform this at the command line or within the advanced storage administration web interface (see the advanced configuration guide for more details).
Choose the protection profile and failure domain before creating the pool. Use the supported placement-group sizing or autoscaling policy for the installed release. Raw capacity, usable capacity, and capacity required to restore protection are different values; retain the recovery reserve defined by the deployment design.
BLOCK SERVICES - RENAME BLOCK POOL
The Rename Block Pool view allows the administrator to change the block pool name. Enter the new pool name and click RENAME when ready to proceed.

BLOCK SERVICES - ADD BLOCK DEVICE
The Add New Block Device view allows the administrator to create a new block device from an existing block device pool.
To create a new block device, provide the block device name, capacity in GB or TBs, the primary pool to use, the secondary data pool if data and metadata separation are required (otherwise leave empty), object size, stripe unit size, and stripe count.

NOTE: If unsure about the object size, stripe unit size, and stripe count values, accept the defaults as these are optimal for most scenarios.
FILE SERVICES
The File Services tab provides access to all filesystem and file sharing services. To add a new filesystem, click on the ADD FILESYSTEM button.
FILE SERVICES - ADD NEW FILESYSTEM
To add a filesystem for supported SMB or NFS services, click ADD FILESYSTEM. Provide its name, data and metadata pool settings, protection scheme, and replication factor where applicable. Filesystem metadata requires replicated protection; the data pool can use the supported replicated or erasure-coded profile. Use the placement-group settings or autoscaling policy prescribed for the deployment.

NOTE: If using erasure-coding for the protection scheme, the cluster is set with a default EC profile as shown in the view. If this needs to be changed to a different scheme definition, you must perform this at the command line or within the advanced storage administration web interface (see the advanced configuration guide for more details).
FILE SERVICES - DELETE FILESYSTEM
To delete a filesystem, click in the action button (┆) for the specific filesystem and click Delete. The Delete Filesystem confirmation view will appear. This requires the operator to type CONFIRM (in all caps). When ready to proceed, click on the Confirm button.

NOTE: Ensure that all IO has stopped to the filesystem and that the filesystem has been globally unmounted before deleting the filesystem.
FILE SERVICES - ADD SMB GROUP
If the SMB protocol is enabled on the cluster, you will see the SMB tab in the lower portion of the File Services pane.
To create a new SMB user group, click on the ADD GROUP button. The Add SMB Group view will appear. Provide a unique name for the new SMB group and click the ADD button when ready to proceed.
NOTE: SMB services can also utilize Active Directory integration for account access and management. The local cluster accounts and AD accounts can be simultaneously assigned to a share for access. AD integration can be done from the provisioning engine (see the advanced systems guide for more information).
FILE SERVICES - ADD SMB USER
To create a new SMB user, click on the ADD USER button. The Add SMB User view will appear. Provide a unique user name for the new SMB user, password, and assigned SMB user groups. Click the ADD button when ready to proceed.
FILE SERVICES - ADD SMB SHARE
To create a new SMB share, click on the ADD SHARE button. The Add SMB Share view will appear. Provide an SMB share name, the filesystem which will be the basis for the file share, a subdirectory within the filesystem to map to, and the SMB groups which should be provided access to the share. When ready to proceed, click on the ADD button.
NOTE: The default subdirectory is ‘/’, which is the root of the filesystem. If you designate a different subdirectory, the system will check to see if the subdirectory exists. If the directory doesn’t exist, the system will create it.
STRATISTOR NFS ACCESS
Use the supported NFS service for the selected StratiSTOR filesystem and deployment. Export configuration, identity mapping, availability, and locking behavior must match the intended clients and application.
- Select the filesystem and path to be exported, and verify that the underlying storage is healthy.
- Configure the export through the supported service interface, including allowed client networks, access mode, identity mapping, and applicable security settings.
- Use the documented service address and export path to mount the export from an authorized client.
- Verify file creation, read/write permissions, ownership, and any application locking requirements.
- For deployments designed for failover, validate client behavior during the supported maintenance or failure test.
When sharing a dataset through both SMB and NFS, verify that the selected service combination supports the required permissions, identity mapping, and locking behavior.
STRATISTOR S3 OBJECT SERVICES
StratiSTOR provides S3-compatible object access in supported configurations. Use the object-service endpoint and credentials defined for the deployment. Object buckets, object access policies, and file shares have distinct management and permission models.
- Verify the object-service endpoint, certificate trust, network access, and service health.
- Create or select the account and credentials appropriate to the application through the supported object-management workflow.
- Create the required bucket and apply the intended access policy and supported data-management settings.
- Configure the client with the endpoint, credentials, and any deployment-specific connection settings.
- Validate upload, download, listing, and deletion with test data, then confirm the required protection and recovery behavior.
STRATISTOR HIGH-PERFORMANCE FILE SERVICES AND SMB MULTICHANNEL
Deployment scope: Use this service only with a server release, storage backend, authentication model, and client configuration documented as supported for the deployment. Availability of a protocol in another StratiSTOR service does not establish support in this service.
Where enabled in a supported configuration, the high-performance StratiSTOR file service provides an optimized data path for demanding file workloads. SMB 3 multichannel can establish multiple transport channels between compatible clients and the server. Actual channel use depends on the client operating system, server configuration, physical network topology, and workload.
Multiple physical adapters can provide additional aggregate capacity when client routing and server configuration use those paths. Multiple addresses on one adapter share that adapter’s bandwidth. Inspect active channels and per-interface traffic during a sustained workload to confirm the intended paths are in use.
Select the backend and availability design explicitly. Local storage, distributed StratiSTOR storage, and other supported backends can have different protection, recovery, and performance characteristics. Verify the supported SMB, NFS, and object modes independently.
Record the client and server versions, physical links, CPU and NUMA placement, storage backend, cache behavior, data set, I/O pattern, duration, and units when reporting performance. A result from a local RAID-0 backend does not establish distributed-storage throughput or fault tolerance.
SMB IDENTITY AND MULTI-NODE SERVICE CHECKS
SMB domain membership and authorization may be managed separately from the Cockpit administrator's directory login. When using Active Directory for SMB, validate domain join, name resolution, share ACLs, multi-node identity consistency and client credentials for the actual file service. Where CTDB or a clustered SMB gateway is enabled, test virtual IP ownership, failover and client reconnection.
SMB multichannel can improve aggregate throughput when supported by both client and server. Its presence should be verified at the client and service level rather than inferred from the number of NICs. Published or lab throughput numbers are workload-specific and should not be treated as guaranteed per-session performance.
STRATISTOR FIBRE CHANNEL GATEWAY
Deployment scope: This section applies only to deployments with an explicitly supported StratiSTOR Fibre Channel gateway configuration. Confirm the gateway release, adapter and firmware compatibility, initiator support, and documented failure behavior before use.
The StratiSTOR Fibre Channel gateway presents selected StratiSTOR block volumes to supported Fibre Channel initiators. A validated multi-gateway configuration can expose multiple paths to one logical volume for management by the initiator’s multipath software.
Configure the supported target-mode adapters, fabric zoning, initiator access controls, and logical-unit mappings. Maintain a consistent logical-unit identity across paths to the same volume and use the documented multipath policy. Validate path loss, gateway restart, and recovery using the supported test procedure.
Path visibility alone does not establish active-active access, safe concurrent writers, or shared persistent reservations across gateways. Applications that require shared-disk coordination must use the explicitly supported locking, reservation, and failure-handling design.
WEB SERVICES
The web services engine provides an alternative way to access any filesystem via a secure high-performance web browser. This is provided on a per cluster-node basis which allows each node to present a different filesystem from the underlying cluster if necessary.
The web service pane displays the current nodes’ portal name, filesystem, binding address, binding port, and certificate information (default self-signed certificates are automatically generated).
WEB SERVICES - SERVICE CONTROL
The web service engine state can be controlled by clicking on the SERVICE CONTROL button. The current state along with stop, start and restart options will be displayed in the service control view.
WEB SERVICES - CONFIGURATION
The web services configuration can be modified by clicking on the EDIT CONFIG button. This will open the web engine config view which will allow the administrator to modify the portal name, binding server address, binding server port, the filesystem root path, the portal admin password, and the certificates used for SSL communication.
NOTE: The Filesystem Path will always mount to the root of the selected filesystem. The administrator can modify the scope of the filesystem that is made accessible by the web engine from the web engine configuration console.
GLOBAL TASK MANAGEMENT
The global task management system is a cluster-wide service which allows administrators to automate and perform any task or group of tasks that require execution on a regular basis, such as migrating data from one location to another or perform general system administrative functions.
The tasks can be scheduled to run on one or all the nodes in the cluster as necessary. The automation tasks can be written in any language such as python, bash shell script, powershell, and many more.
SCHEDULED JOB VERIFICATION
The global scheduler and service-specific backup/replication schedulers may have different ownership and execution logs. After creating, editing, disabling or re-enabling a task, confirm its next scheduled run and check the final execution result. Do not assume that a graphical “enabled” state or the presence of a recurring task establishes that work has executed successfully.
GLOBAL TASK MANAGEMENT - CREATE TASK
To create a task, click on the Schedule tab and then click the Add New Event button. This will open the task wizard where the administrator can define what the task is and how it is to be executed.
For recurring tasks, define the intended execution host or scope, credentials, timeout, retry behavior, logging, and notification path. Prevent unintended overlap where a previous run can still be active. Test the task and inspect its result before enabling the recurring schedule.
REMOTE SUPPORT
The StratiSYSTEM platform has call-home remote support capability which allows SteelDome support personnel to connect and assist with support requests.
When enabled, the service forms a secure reverse tunnel which doesn’t require corporate firewall changes unless outbound access is blocked.
The service can be enabled when required and disabled when not.
REMOTE SUPPORT - SERVICE CONTROL
The remote support service state can be controlled by clicking on the SERVICE CONTROL button. The current state along with stop, start and restart options will be displayed in the service control view.
REMOTE SUPPORT - CONFIGURATION
The remote support service requires a one-time time-based support token. Once SteelDome personnel generate this token, the administrator will have five minutes to enter and activate the service or else the token will expire and become invalid.
This process only needs to be completed once. From that point forward, access can be controlled by the administrator by starting and stopping the service as required.
UPDATES
The StratiSYSTEM platform can be updated directly from the web interface using the Updates tab. The current version and new version (if available) will be displayed in this pane.
For a connected update, verify access to the required software sources. For a prepared offline environment, use the approved local resources and release-specific update procedure. Availability of an offline deployment method does not imply that every upgrade path is available offline.
NOTE: Review the component scope and maintenance requirements of each update. Management-interface changes, operating-system updates, and storage, virtualization, Kubernetes, or OpenStack changes can have different restart and reboot requirements. Use simultaneous or rolling updates only as prescribed for that release; do not assume that every update is limited to the management layer.
ONLINE AND PREPARED OFFLINE UPDATE SOURCES
On the Updates page, select the approved connected repository or offline update bundle when the corresponding update method is available. For offline updates, provide the release-specific bundle and required credentials, then review package verification and prerequisite results. Confirm the target release, affected nodes, enabled services and maintenance sequence before starting.
Review the update details and any required prerequisites before applying packages. Monitor the update log through completion and confirm that all affected services have returned to a healthy state. In offline environments, verify the required local packages and update artifacts are present before beginning maintenance.
MAINTENANCE AND OFFLINE OPERATION
StratiSYSTEM OS supports connected environments and prepared offline deployments using local software resources. The applicable installation or update procedure depends on the installed release, target release, enabled services, and available software bundle.
Before applying an update, review its component scope, prerequisites, compatibility changes, restart requirements, and recovery procedure. Distinguish management-interface updates from changes to the operating system, storage services, virtualization services, Kubernetes, or OpenStack.
For offline environments, prepare the approved packages, repositories, container images, and other required artifacts before the maintenance window. Verify the completeness and integrity of the bundle and the availability of any required local infrastructure services.
Use the release-specific maintenance sequence. Check service health and destination capacity before evacuating or restarting a node, and validate the node before advancing to the next step where rolling maintenance is supported. A failed health check should be resolved before the maintenance sequence continues.
After the update, verify software versions, cluster membership, storage health, workload access, scheduled tasks, monitoring, and persistent configuration. Record any remaining operational action in the maintenance log.
MANAGEMENT INTERFACE HEALTH AFTER UPDATES
If a StratiSYSTEM management page does not load after an update, confirm that the underlying service is running and review the system journal and management-service logs. Check any reported configuration or access errors before attempting a package reinstall. Make configuration-file changes only through an approved maintenance or support procedure.
After an update or recovery, verify access to management endpoints, local and directory authentication, MFA where configured, time synchronization, storage health and scheduled-job execution. Confirm that the workloads and services required by the deployment are operational before ending the maintenance window.
ZFS SERVICES - DASHBOARD
The ZFS Dashboard provides a summarized view of the current pool and dataset configuration within the system. It displays vital information about pool health, including the RAID level, total provisioned space, used and available capacity, and error metrics such as read, write, and checksum errors. Dataset details are also presented, listing each filesystem and volume associated with the pool, their usage statistics, compression and deduplication status, encryption settings, mount points, and snapshot counts. This serves as a central point for monitoring the operational state of ZFS services.

ZFS SERVICES - CLOUD STORAGE
The Cloud Storage tab presents the operator with the ability to define cloud-backed storage service providers to be used as block devices. These block devices can be used as backing disks for ZFS pools or any other function within the platform.

To begin, click on the Add Provider button. This will open the Add Cloud Provider modal window. Enter the provider friendly name and provider specific Access Key ID and Secret Access Key pairs. When finished, click on the Save Provider button.

Next, define a cloud provider connection. Select the provider from the dropdown list, provide a unique bucket name for this connection, mount path where buckets will be mounted as block device, the cache directory where the cache files for the connection will be stored, the designated cache size in MBs, the region where the bucket originates from the provider, and the size of the block device to be created from the bucket.

Before using a cloud-backed device, verify the supported provider and connector, cache requirements, network dependencies, and recovery behavior. Confirm the required durability and consistency characteristics for the intended application and avoid concurrent access patterns that are not supported by the selected backend.
ZFS SERVICES - STORAGE
The Storage section visualizes the structural layout of ZFS storage using a hierarchical tree view. Each pool is displayed as a top-level node, under which associated filesystems and volumes are listed. This representation helps users intuitively understand the relationship between datasets and their parent pool, offering quick access to context-specific actions and configurations. It supports easy navigation and management of datasets within a graphical interface.

From the action menu for each object in the tree, the operator can:
|
|
|
ZFS SERVICES - SNAPSHOTS
The Snapshots tab lists all snapshots associated with a selected dataset. Each entry includes the snapshot name, creation timestamp, data size used, whether the snapshot is held (protected from deletion), and whether it is currently mounted. This view enables users to audit the state of point-in-time copies for rollback or recovery purposes and provides controls for managing individual snapshot lifecycles.

From the action menu for each snapshot in the table, the operator can:
- Create Snapshots – Manually generate a point-in-time copy of a dataset or volume.
- Delete Snapshots – Permanently remove a selected snapshot from the system.
- Hold Snapshots – Prevent deletion of a snapshot by placing a hold on it.
- Release Snapshots – Remove a previously applied hold, allowing deletion.
- Rollback to a Snapshot – Revert the dataset or volume to the state captured by the snapshot.
- Schedule a Snapshot – Define an automated recurring snapshot creation rule.
- Show Differences Between Snapshots – Display changes between two snapshots or snapshot and current state.
- Promote Snapshots – Change a clone’s origin relationship through the supported promotion operation; review the effect on dependent datasets before proceeding.
- Mount Snapshots – Temporarily mount a snapshot for read-only access or inspection.
- Unmount Snapshots – Detach a previously mounted snapshot from the filesystem.
- Rename Snapshots – Change the name of an existing snapshot.
- Clone Snapshots – Create a writable clone from a snapshot for testing or development.
A snapshot hold prevents deletion while the hold remains in place. It is not an independent backup or an immutable retention guarantee against an administrator who can release the hold. Keep recovery copies and access controls appropriate to the protection requirement.
ZFS SERVICES - EXPORTS
The Exports tab lists all NFS exports that are active on the current system. Exports allow you to share ZFS-backed filesystems over NFS, enabling seamless access from remote systems on your network. Each export defines the filesystem path, the range of allowed client endpoints, and specific NFS options controlling access, performance, and security. This configuration provides fine-grained control over how datasets are accessed, ensuring both flexibility and efficiency in network-based storage environments.

Exports can be mounted by all nodes within the cluster, or by any authorized NFS client on the local network, providing a versatile and high-performance way to share data across systems. Because ZFS can be backed by any physical disk type—including local disks, cluster-attached storage, SSDs, HDDs, or even a mix of different devices—it allows storage pools to be created from diverse storage resources. This flexibility means exports are not limited to cluster disks, enabling administrators to aggregate capacity and performance from a wide range of hardware into a unified pool. These pools can host datasets that take full advantage of ZFS’s advanced features, including inline compression, deduplication to eliminate duplicate data blocks and save space, snapshots for point-in-time recovery, replication for disaster recovery, and end-to-end data integrity verification.

Once an export is defined in the ZFS Services interface, it becomes immediately available for mounting within the cluster. Using the Storage tab and the New NFS Mount option, administrators can quickly mount the export to any node, ensuring consistent access paths, centralized management, and optimal utilization of all available storage resources.
ZFS SERVICES - SNAPSHOTS (SCHEDULING)
The Snapshot Schedules interface provides an overview of all active snapshot schedules. Each row includes the scheduling interval defined in CRON-like terms (minute, hour, day, etc.), the prefix used in naming the snapshots, retention count (number of snapshots to keep), and the target dataset. This feature helps ensure consistent data protection policies are applied by automating the creation of snapshots at regular intervals.

The “New Snapshot Schedule” dialog allows users to define a custom recurring snapshot task for a specific dataset. Users can select snapshot intervals, such as every 15 minutes, set retention limits to control snapshot accumulation, and define a naming prefix for organizational clarity. This configuration ensures that snapshots are created and retained according to predefined policies, facilitating robust data protection and recovery strategies.

ZFS SERVICES - STATISTICS
The Stats section presents real-time ZFS performance metrics, offering insight into read/write IOPS, throughput, and ARC (Adaptive Replacement Cache) behavior. Metrics such as ARC hit ratio, ARC size, and configured limits are displayed both as numerical values and graphical bars, helping users evaluate system performance and identify bottlenecks or inefficiencies in caching and disk operations.

ZFS SERVICES - SMART INFO
The Smart Info tab provides an overview of hardware health data collected from connected block devices. Each device is listed with its model, serial number, pool association, interface type, and SMART attributes such as health status, temperature, and error counts. In this example, most traditional devices show unknown SMART data, while NVMe-based SDCache devices report health as "Healthy" with active SMART statistics. This information is critical for identifying failing drives before data loss occurs.

By selecting a device from the SMART Info list, users can view partition-level details including partition names, sizes, mount points, filesystem types, and usage levels. This enables a deeper understanding of how each physical device is partitioned and utilized by the system, and can assist in tasks such as troubleshooting or planning storage layout changes.

ZFS SERVICES - SETTINGS
The Settings section allows configuration of global ZFS management behavior. Users can adjust the console refresh interval, statistics polling frequency, and specify an email address for system-generated notifications. These settings provide control over how frequently information is updated in the UI and how administrators are alerted to important ZFS events or conditions.

ZFS SERVICES - NEW POOL WIZARD
The pool creation wizard guides users through defining a new ZFS pool. Users can specify the pool name, select the RAID level (e.g., RAIDZ1), define a mount point, and choose advanced options including encryption type, compression algorithm, deduplication hash, checksum method, and block size. Additional comments can be added for documentation purposes. This tool simplifies the process of building secure, high-performance ZFS pools tailored to specific workload needs.

The second step in the ZFS Pool Creation wizard allows users to select both required and optional devices to be included in the new pool. Users begin by choosing the main data devices, which will form the core of the pool's RAID group based on the selected RAID level (e.g., RAIDZ1, RAIDZ2, etc.). Additionally, users can assign drives to specialized roles under optional device categories, such as:
- Spare Devices, which act as hot spares to automatically replace failed disks.
- Log Devices, used to accelerate synchronous write operations via ZFS Intent Log (ZIL).
- Cache Devices, which extend the ARC (Adaptive Replacement Cache) for improved read performance via L2ARC.
- Deduplication Devices, which store deduplication metadata to improve performance when deduplication is enabled.
Each optional category supports different RAID level scopes where applicable, allowing administrators to tailor the redundancy and performance characteristics of auxiliary storage tiers within the ZFS pool.

In the final step of the pool creation wizard, a complete summary of the configuration is presented for user confirmation before committing changes. This summary includes:
- Pool Name, compression algorithm (e.g., ZSTD), encryption level (e.g., 256-bit GCM), and deduplication hash (e.g., SHA256).
- Checksum algorithm used for data integrity validation, typically SHA-256 for cryptographic strength.
- Autotrim and Autoreplace settings, which help maintain performance and automate failed disk replacement.
- Disk Block Size and any administrative comments for internal documentation.
The list of main data devices is shown with the corresponding RAID layout (e.g., RAIDZ1 for single disk fault tolerance), followed by any assigned optional devices for cache, logs, deduplication, and spares. This comprehensive overview ensures the user can validate the complete storage layout and protection strategy before the pool is provisioned.

FILE BROWSER
The File Browser view provides a convenient way to access the node and cluster storage resources directly from the web console. The administrator can create files, directories, change permissions, and upload and download files directly through the browser window.
FILE BROWSER - FILE AND DIRECTORY FUNCTIONS
There are several filtering and action buttons available depending on what type of object you are working on within the storage system.
ACCOUNTS
Account Management supports local administrative accounts and, in compatible builds, cluster-aware management and enterprise identity workflows. Local operating-system accounts can still have node-specific scope; verify the selected host and the actual propagation behavior before assuming changes were applied to every cluster node. Administration identities, SMB users, storage-client credentials and OpenStack tenants may use different identity stores and authorization policies.
Host accounts, SMB identities, storage-client credentials, and private-cloud tenant accounts can have different scopes and management interfaces. Use individual identities and task-appropriate privileges where supported. Verify which systems are affected when changing an account or credential.
ACCOUNTS - CREATE NEW ACCOUNT
To create a new account click on the Create New Account button. The administrator can define all the parameters necessary to create the new account such as full name, user name, home directory, shell type, and password.
ENTERPRISE IDENTITY SERVICES — COCKPIT ACTIVE DIRECTORY
Where enterprise directory integration is enabled, open Accounts → Cockpit AD to configure domain membership and define the directory groups authorized to administer the management console. Cockpit directory authentication is separate from SMB domain integration. Retain a controlled emergency local-access method.
- Confirm persistent DNS resolution, domain controller reachability and synchronized system time on every participating host.
- Join or reuse the domain membership through the supported interface; verify trust and status per host rather than assuming one node's membership applies cluster-wide.
- Set allowed administrative groups and explicitly test both authorized and unauthorized users.
- Test local-account fallback and the planned response to temporary directory outages.
If authentication fails, check the domain membership and trust status, directory lookup, DNS resolution and time synchronization on the affected node. Use the advanced identity guide or approved support process for command-line remediation.
MULTIFACTOR AUTHENTICATION FOR THE MANAGEMENT CONSOLE
Where multifactor authentication is configured for the management console, enroll the intended administrator identities using the supported authentication provider and method. Verify that valid credentials and a valid second factor are required. MFA for the management console does not automatically protect SSH, storage protocols, BMC access or application logins; configure those services separately where required.
Plan policy consistency across nodes, time synchronization, enrollment, administrator recovery, break-glass access and provider outage behavior. Verify a valid second factor succeeds, an invalid factor fails and unauthorized directory users remain denied. A saved configuration is not proof that the second factor is being enforced.
SERVICES
The Services view provides a real-time status of all services currently running on this cluster node. The services can be filtered using the filter bar.
Service details can be viewed by clicking on any of the services. The full service detail including service state control will be visible to the administrator.
DIAGNOSTICS
System diagnostic reports can be generated at the direction of SteelDome support personnel. The Diagnostic Reports view will show all previous generated reports for this cluster node.
To run a report, click on the Run Report button.
DIAGNOSTICS - RUN NEW REPORT
The Run New Report view will appear. The administrator will need to provide the report label, optional password for encrypting the report, and other report specific options such as obfuscation and verbosity settings.
When ready to run the report, click on the Run Report button. The report will take a few minutes to complete. Once completed this can be downloaded and provided to SteelDome personnel, or the support personnel can access the report via the remote support service.
ROUTINE HEALTH CHECKS AND RECOVERY VALIDATION
Monitor the health of the operating platform and each enabled service. Routine checks should cover StratiSTOR protection and capacity, StratiSERV hosts and workloads, Kubernetes services where enabled, network health, host resource pressure, and recent backup outcomes.
For StratiSTOR, examine device and service availability, data protection state, recovery progress, metadata-service health, and capacity distribution. For StratiSERV, check host eligibility, HA protection, migration restrictions, placement rules, and workload resource pressure. HyperSERV operations should consider the combined effect of compute activity and storage recovery.
Snapshots, replication, and backups meet different recovery needs. Verify what each mechanism protects, where recovery data is stored, and the failure scenarios it addresses. Include VM configuration and security state as well as disks when defining a complete recovery procedure.
Validate backup repository access from the required hosts, monitor job failures, and perform scheduled restores into an isolated test environment. Recovery tests should verify application function and the recovery objectives established for the workload.
When investigating an incident, preserve diagnostic evidence before a reboot or service restart when operationally practical. Include the relevant time window, cluster and service status, resource pressure, storage and network errors, crash information, and recent changes.
RECOMMENDED WEEKLY OPERATIONS REVIEW
- Confirm StratiSTOR quorum, device health, balanced utilization, recovery/scrub state, metadata services and client protocol endpoints.
- Confirm StratiSERV host availability, VM HA/replication health, sufficient migration headroom and destination policy consistency.
- Check backup job last/next run, repository access, disk coverage, retention, integrity tests and at least one documented recovery test.
- Review Kubernetes/OpenStack service health where enabled, plus GPU/AI model-service status and shared filesystem access.
- Review directory login, second-factor enforcement, certificate expiry, time sync, remote-access status and administrative audit records.
- Validate updates and platform configuration changes against release notes; retain evidence of any incident or incomplete operation.
TERMINAL ACCESS
The Terminal view provides administrators with terminal access to the local cluster node. This is equivalent to SSH access but through the web browser window without the need for SSH terminal software.
Terminal access can provide more detailed access to underlying system functions and should only be used under direction of SteelDome support personnel.
OPTIONAL SERVICES AND CONFIGURATION REQUIREMENTS
Before using an optional service, verify that the service is installed and enabled, the required license is active, the hardware and operating-system components are compatible, and the selected management interface is available. Test the expected behavior before placing a workload into production.
| Service category | Examples | Administrator checks |
|---|---|---|
| Core platform services | Host management, networking, storage pools, VMs, backups, accounts and updates | Confirm service health, permissions, available capacity and recovery procedures. |
| Optional platform services | Kubernetes, OpenStack, AI model services, directory authentication and MFA | Verify installation, configuration, access policy, prerequisites and service-specific health. |
| Hardware-specific capabilities | GPU passthrough and sharing, accelerated networking, specialized gateways and multi-site services | Confirm hardware and driver compatibility, topology, failure handling and documented limitations. |
Use the StratiSYSTEM OS release notes, hardware compatibility information and advanced product guides for supported configurations and detailed maintenance procedures.
MANAGEMENT CONSOLES AND ACCESS
Access the StratiSYSTEM OS management interface using the hostname or address and credentials supplied for the deployment. Service-specific management endpoints depend on which services are enabled and how the environment was configured.
Use individual administrative identities where supported and assign privileges appropriate to the service and task. Host administration, storage-client access, application access, and tenant access can use different identity systems; verify the scope of each identity before assigning access.
Replace initial bootstrap credentials according to the deployment procedure and configure trusted certificates for production management endpoints. Restrict management access to the intended networks and maintain a current endpoint inventory.
Remote support access remains under the administrator’s control. Enable it when required for a support session, verify its state, and stop the service when that access is no longer required.
The following addresses are examples from the illustrated deployment. Substitute the hostname or address and endpoint configuration assigned to your environment. Optional services are available only when enabled. Use the credentials supplied or created for each service; no shared default password is implied.
| Service | Example address |
|---|---|
| StratiSYSTEM Cluster Manager | https://stratistor1-1:3199 https://stratiserv1-1:3199 |
| StratiSERV Container Manager | https://stratistor1-1:3197 |
| StratiSYSTEM Operations Manager | https://stratistor1-1:3195 |
| StratiSERV VDI Management | https://stratistor1-1:12443 |