STEELDOME TECHNOLOGIES

PARTNER DEPLOYMENT GUIDE

The SteelDome Partner Onboarding and Technology Deployment Guide

STRATISYSTEM OS — PARTNER DEPLOYMENT GUIDE

SteelDome Technologies | StratiSTOR • StratiSERV • HyperSERV

Standard installation, cluster provisioning, storage and compute activation, operational validation, and customer handoff for authorized SteelDome partners, deployment engineers, and solution architects.

Document: Partner Deployment Guide, revision 1.4. Release-specific SteelDome-approved procedures take precedence.

BEFORE YOU BEGIN

This guide explains the standard deployment sequence for StratiSYSTEM OS. It is a field guide: understand the stage, follow the typical steps, and verify the result before moving on.

RELEASE-SPECIFIC SOURCE OF TRUTH: Use SteelDome-approved installation media, build-matched preparation and provisioning tools, node-bound licenses, and an approved site plan. Menu labels and available options can change between software releases.

CHOOSE YOUR DEPLOYMENT

Deployment What it delivers Typical starting point
StratiSTOR Software-defined storage for block, file and object data services. Dedicated storage nodes; enable only required protocols.
StratiSERV Virtual machines, containers and workload management. Dedicated compute nodes with an approved storage path.
HyperSERV StratiSTOR storage and StratiSERV compute on the same nodes. Converged three-node cluster; provision storage before compute.

STANDARD DEPLOYMENT SEQUENCE

PLAN INSTALL DEFINE PREPARE PROVISION ACTIVATE VALIDATE
Design Install OS Configure Prepare nodes Form cluster Enable services Accept and hand off

Standard sequence: plan, install, define, prepare, provision, activate, and validate. Complete each gate before continuing.

1 | UNDERSTAND THE PLATFORM

StratiSYSTEM OS is the common operating foundation across storage, compute and hyperconverged deployments. Partners deploy the same base platform and activate only the services needed for the customer environment.

DEPLOYMENT ROLES AND SHARED FOUNDATION

StratiSTOR StratiSERV HyperSERV
Software-defined storage

Block, file, object and storage protection

Compute and orchestration

Virtual machines, containers, workload lifecycle and HA

Converged infrastructure

StratiSTOR + StratiSERV on shared compute/storage nodes

STRATISYSTEM OS

Common platform foundation

Common deployment, networking, security and management foundation Activate only the required service roles

Figure 1. Storage, compute and hyperconverged service roles share the StratiSYSTEM OS foundation.

WHAT CHANGES BETWEEN DEPLOYMENTS

STANDARD CLUSTER PATTERNS

Pattern Recommended approach
Production HA Use three or more nodes for a conventional highly available cluster; verify quorum, redundancy and adequate spare capacity.
Single-node Use for evaluation, labs or explicitly accepted non-HA installations.
Hardware symmetry Keep software release, firmware policy, boot mode, networking and performance settings consistent. Document intentional differences.

2 | PLAN THE SITE

Most deployment delays originate from unresolved prerequisites. Confirm the hardware platform, networks, storage ownership and access information before booting any node.

HARDWARE AND PLATFORM READINESS

Area Field guidance
Node count Three or more for standard production HA; one for non-HA evaluation. A special two-data-node design with a separate quorum-only node is also supported.
OS media At least 200 GB OS capacity per node. Mirrored SSD OS storage is preferred.
Network ports 10 GbE or faster is strongly recommended for front-end and private traffic. Use redundant ports for production where feasible.
Remote access BMC/iDRAC/iLO or equivalent with working console and virtual media.
Firmware and BIOS Consistent boot mode, approved Secure Boot setting, High Performance profile and supported firmware.
Virtualization / GPUs Enable CPU virtualization and IOMMU where required. Prepare GPUs only for approved designs.

NETWORKS IN A TYPICAL DEPLOYMENT

A standard deployment separates routed client/service-facing networking from the private inter-node network. Management access can share the front-end network or use a dedicated interface.

NODE 1

Front-end + private NIC/bond

NODE 2

Front-end + private NIC/bond

NODE 3

Front-end + private NIC/bond

FRONT-END / br0 — Client, administration, service and VM traffic
CLUSTER PRIVATE — Isolated inter-node traffic; normally no default gateway

Figure 2. Example of a three-node, two-network design; actual NICs, addresses, VLANs, and switch topology are site-specific.

Network Used for Standard rule
Front-end / br0 Administration, client access, storage services and VM-facing traffic Use approved IP, prefix, gateway, DNS and MTU.
Cluster-private Inter-node communication and, when enabled, migration traffic Separate subnet; normally no default gateway or DNS.
Management / BMC Out-of-band server access and optional dedicated management Restrict access to approved administrators.
Service VIPs HA entry points for SMB, NFS, S3 and other protocols Reserve IPs and DNS names before deployment.

NETWORK DESIGN CHECKPOINT: NIC/bond settings must match switch-side LACP or failover configuration. Verify VLANs, routing, address uniqueness and end-to-end MTU before provisioning.

SITE CHECKLIST

DO NOT TREAT A DEVICE NAME AS IDENTITY: Linux /dev/sdX and /dev/nvmeXnY paths may change. Confirm controller, capacity, model, serial and WWN before any data-erasing operation.

3 | INSTALL STRATISYSTEM OS

Install the base operating system on each server. At this stage the goal is a correctly installed and reachable node; cluster services will be configured later.

INSTALL ON EACH NODE

  1. Boot the approved StratiSYSTEM OS ISO using remote virtual media or prepared USB media.
  2. Choose the language, keyboard, hostname and intended OS installation target.
  3. Identify the OS disk by serial and capacity. If protected hardware RAID already presents one OS virtual disk, do not add a second software-RAID layer.
  4. Where two independent OS SSDs are supported, use installer RAID 1. For standard physical nodes, choose standard partitioning rather than LVM unless release guidance specifies otherwise.
  5. Remove installer-created /home and swap partitions when required by the workflow; use remaining OS capacity for the root filesystem. Enable encryption only with an approved unlock procedure.
  6. Configure initial management/front-end IPv4, prefix, gateway and DNS. Apply approved IPv6 policy and defer other NICs to preparation when appropriate.
  7. Complete installation, remove media, reboot and verify console/SSH access. Repeat consistently on all nodes.

DESTRUCTIVE DISK SELECTION: Installer partition changes erase the selected disk. Never select a storage data device, cache member or protected SAN LUN as the OS destination.

FIRST-BOOT VERIFICATION

Run these commands on each node and compare against the approved site worksheet.

hostnamectl
ip -4 address show
ip route
lsblk -o NAME,SIZE,TYPE,FSTYPE,MOUNTPOINTS,MODEL,SERIAL

GATE 1 — INSTALLATION COMPLETE: Do not define the cluster until all nodes are reachable and their OS/data disk inventories have been checked.

4 | DEFINE THE CLUSTER

The deployment definition describes node identities, cluster networking and services for the preparation and provisioning tools. Begin with the JSON supplied for the installed release rather than creating a new definition from scratch.

SELECT THE PRODUCT-SPECIFIC FILE

Product Working deployment file
StratiSTOR /stratisystem-prep/stratistor/config/deployment/current_deployment.json
StratiSERV /stratisystem-prep/stratiserv/config/deployment/current_deployment.json
HyperSERV /stratisystem-prep/stratistor/config/deployment/current_deployment.json

HYPERSERV FOLLOWS THE STORAGE-FIRST ROUTE: Use the StratiSTOR deployment definition to establish storage. Add hypervisor services only after storage provisioning and health validation.

EDIT THE SITE-SPECIFIC ESSENTIALS

Configuration area What partners should confirm
Cluster identity and secrets ClusterNodeUserId and bootstrap credential fields are appropriate; protect secrets.
Front-end subnet FrontendAccessSubnet, service VIPs and alert recipients match the approved plan.
Private network ClusterPrivateNetwork and ClusterPrivateSubnet describe the intended isolated network.
Node inventory NodeConfig includes one accurate entry per node, with unique hostnames and IP addresses.
Interface mapping FrontendAccessInterface, FrontendAccessBridge (typically br0), ClusterPrivateInterface and MTUs match the physical topology.
Storage definition Use approved pool protection, device selection and failure-domain policy; avoid casually modifying EC and low-level performance fields.
Optional services Enable NFS/S3 HA, stretched configurations and other advanced services only when approved.

VERIFY AND DISTRIBUTE ONE APPROVED DEFINITION

  1. Confirm node identities with hostnamectl and interfaces with ip link. Check that node/VIP addresses are unique.
  2. Confirm the front-end default gateway uses the intended routed network. Private gateway and DNS are usually blank.
  3. Check end-to-end MTUs and approved replicated/EC protection settings.
  4. Distribute one identical, validated deployment definition to each node and compare hashes before provisioning.
# Example: validate approved JSON and obtain comparison hash
python3 -m json.tool /path/to/current_deployment.json >/dev/null
sha256sum /path/to/current_deployment.json

CHECK THE RELEASE VERSIONS FILE

The versions manifest at /stratisystem-prep/modules/versions/versions.json selects component versions used by the build. Confirm that it matches the SteelDome-approved release; do not independently alter package pins or upgrade versions without release instructions or support direction.

GATE 2 — DEFINITION ACCEPTED: Every node must have a valid, identical deployment file. Stop if any interface, hostname, address or storage-protection setting is in doubt.

5 | PREPARE THE NODES

The preparation engine applies node-bound licensing, installs prerequisites, configures approved hardware support and prepares servers for cluster integration. Run it on every participating node.

START THE PREPARATION ENGINE

cd /stratisystem-prep
source start.sh

COMPLETE THE STANDARD WIZARD

  1. Choose Licensing. Import the supplied license for this node and verify its machine ID; repeat on all nodes.
  2. Select the appropriate new-cluster preparation workflow for StratiSTOR or StratiSERV; for HyperSERV, follow the storage-first route.
  3. Confirm NTP/time servers, front-end and private interfaces, host identity and administrative access.
  4. Install the Cockpit networking option only when needed for advanced NIC/bond configuration.
  5. Select NVIDIA or AMD GPU preparation only for supported hardware and approved workloads.
  6. Review any non-OS disk-wipe option carefully. Skip it if shared LUNs, previously used disks or device ownership are uncertain.
  7. Check the final wizard summary against the site plan and start preparation on each node.

CONFIGURATION VALUES MAY BE UPDATED BY THE WIZARD: Compare displayed network values against the deployment definition before confirming. Keep console access available in case a network change disrupts connectivity.

ALLOW REBOOTS AND VERIFY COMPLETION

Preparation may reboot a node multiple times. Reconnect after the final reboot, confirm completion and inspect the log only if it failed.

ls -l /.prepcomplete
tail -n 200 /.preplog      # only if preparation fails

GATE 3 — EVERY NODE PREPARED: /.prepcomplete must exist on every member before provisioning. If missing, inspect /.preplog, resolve the first error and repeat only the supported workflow.

6 | PROVISION THE CLUSTER

Provisioning integrates prepared nodes into a cluster. The standard procedure includes a first-node role, peer-node roles and a mandatory dependency between Phase 1 and Phase 2.

OPEN THE CORRECT PROVISIONER

Deployment Command
StratiSTOR / HyperSERV storage cd /stratistor && ./start.sh
StratiSERV cd /stratiserv && ./start.sh

FOLLOW THE PHASE ORDER

Sequence What to do Continue when…
1 — Node 1 Choose option 1 to initialize the first node; start Phase 1. Node 1 reaches its Phase 1 checkpoint.
2 — Peer nodes On each peer node, choose option 2 and complete Phase 1. Every peer reports Phase 1 completed successfully.
3 — Node 1 again Only after every peer has finished Phase 1, continue Phase 2 on Node 1. Node 1 completes integration and cluster services are running.

PHASE ORDERING IS MANDATORY: Never start Node 1 Phase 2 while any peer is still running or has failed Phase 1. Resolve the peer problem first to avoid a partially integrated cluster.

SELECT THE NEXT PRODUCT PATH

If deploying… Next step
StratiSTOR Activate storage and SDCache, then required client data services.
StratiSERV Activate VM services, container services or both.
HyperSERV Activate StratiSTOR storage first, then the StratiSERV compute layer.

GATE 4 — CLUSTER INTEGRATED: Confirm all intended nodes are visible and the provisioning wizard reports success before enabling storage or workload services.

7 | ACTIVATE STORAGE AND SDCACHE

StratiSTOR and HyperSERV use SDCache as part of storage-device preparation. It can accelerate supported devices using RAM or disk cache and provide a consistent device presentation. Begin only after successful cluster provisioning.

CHOOSE THE APPROPRIATE CACHE APPROACH

Approach What it does Normal partner guidance
RAM-based Uses a configured share of node memory to accelerate selected devices. Straightforward option; preserve enough RAM for cluster services.
Disk-based Uses a dedicated cache disk or protected device group. Use only when included in the approved design; protected cache media is recommended.
Write-back Permits acknowledgment of writes from cache before backing media. Advanced option; requires durable cache, power-loss safeguards and recovery plan.

Write-through is the more conservative disk-cache starting point. Do not choose advanced caching solely for performance without validating durability and recovery.

STANDARD ACTIVATION FLOW

  1. In the StratiSTOR provisioning menu, open Storage Management → SDCache Management → Configure SDCache.
  2. Choose RAM-based or disk-based acceleration; confirm node, backing drives, target cache device and mode.
  3. Validate all devices in any wizard-expanded range. Drive paths can shift between boots; cross-check serial/WWN inventory.
  4. For protected disk caching, create and activate the approved device group before assigning it to SDCache.
  5. Run SDCache validation. If existing signatures or partitions appear, stop and independently confirm device ownership.
  6. Activate approved storage devices, monitor provisioning logs, and compare online data-device/OSD counts with the storage map.

USEFUL NON-DESTRUCTIVE CHECKS

lsblk -o NAME,SIZE,TYPE,FSTYPE,MOUNTPOINTS,MODEL,SERIAL
nvme list                          # if NVMe devices are present
wipefs -n /dev/<verified-device>   # read-only signature scan

DO NOT ERASE DISKS TO CLEAR VALIDATION WARNINGS: The standard partner workflow deliberately uses non-destructive inspection. Sanitation requires separate approval, verified disk serial/WWN and a release-qualified process.

STORAGE READINESS CHECKS

GATE 5 — STORAGE HEALTHY: Do not enable SMB, NFS, S3, block access, or HCI compute workloads until storage activation has completed with the expected health.

8 | ACTIVATE THE REQUIRED SERVICES

After the cluster foundation is ready, activate only those services included in the approved customer design. Do not enable all available protocols by default.

SERVICE ACTIVATION BY DEPLOYMENT TYPE

Product Activate Verify with a simple test
StratiSTOR Required block, file and/or object endpoints; reserve VIPs and DNS. Connect a test client to each enabled service and verify access.
StratiSERV Required VM, container or Kubernetes services. Create or access a representative workload; verify its network and storage.
HyperSERV Storage first; then VM and container services on converged nodes. Check storage health first, then create a test workload and verify mobility where applicable.

COMMON SERVICE PLANNING NOTES

ACCESS THE MANAGEMENT INTERFACE

https://<node-management-or-front-end-IP>:3199

Initial administrative account names in the source deployment guide are stratistor-admin for StratiSTOR/HyperSERV storage administration and stratiserv-admin for StratiSERV. Obtain bootstrap passwords from the controlled installation handoff; do not publish them in a general guide.

ROTATE BOOTSTRAP CREDENTIALS AT FIRST ACCESS: Replace root, product-admin, cluster and service bootstrap passwords. Store secrets in the customer-approved vault, document access ownership and restrict SSH and TCP/3199 to trusted management networks.

9 | VALIDATE, SECURE AND HAND OFF

Deployment is complete only when the infrastructure works, management access is secured, and an accurate as-built record is delivered. Capture evidence from the running environment, not only wizard success messages.

ACCEPTANCE MATRIX

Area Acceptance test Evidence
Nodes All nodes visible, names/addresses/time sync correct. Cluster/node health view.
Licenses Each node license valid and bound to its machine ID. License status.
Storage (if used) Expected devices/OSDs online with no unexpected critical health state. Storage health and device map.
Workloads (if used) Test VM/container starts and reaches approved networking and storage. Workload test record.
Services (if used) Every enabled SMB/NFS/S3/block endpoint works. Client connectivity results.
Management Approved administrators can access UI and console on approved networks. Login and network verification.
Protection HA, redundancy and recovery assumptions documented. Design-to-as-built review.

MINIMUM SECURITY CLOSEOUT

CUSTOMER HANDOFF PACKAGE

Deliverable Include
As-built worksheet Node names, serials, NICs, IPs, VLANs, MTUs, VIPs and service URLs.
Storage map OS/data/cache roles, serials, RAID/cache design and expected disk inventory.
Configuration archive Approved deployment definition and device-group files, with secrets protected or redacted.
Health and access evidence Success views, service client tests and administrative access ownership.
Support/lifecycle record Installed release, support contact, licensing and escalation route.

GATE 6 — READY FOR ACCEPTANCE: All applicable checks must pass and the as-built package must be delivered before customer signoff. Use Appendix D for the final acceptance record.

10 | TROUBLESHOOT SAFELY

Most first-deployment problems trace back to networking, mismatched definitions, licensing, incomplete preparation or wrong phase ordering. Repair the first failed checkpoint instead of skipping forward.

Symptom First checks and recommended next action
Node unreachable Use BMC console; check IP/prefix, link, gateway, VLAN and address uniqueness.
JSON rejected Use python3 -m json.tool, fix syntax and redistribute the approved file.
Definition hashes differ Redistribute one approved master definition and verify SHA-256 on all nodes.
License rejected Check machine ID and license binding; obtain correction if necessary.
/.prepcomplete missing Read /.preplog and resolve the first failure. Do not provision yet.
Peer fails Phase 1 Check private network, DNS, time, credentials and definition consistency. Do not start Node 1 Phase 2.
SDCache finds signatures Stop; identify serial/WWN and ownership. Use read-only inspection, not wipes.
Data devices missing Compare inventory to activation logs and storage health.
Management UI unavailable Verify route and TCP/3199 connectivity from approved network; inspect service using supported tools.

RECOVERY RULES

EVIDENCE FOR STEELDOME SUPPORT

Record the installed release, affected node and machine ID, failing phase and timestamp, redacted deployment definition, relevant preparation/provisioner logs, and network/disk inventory when useful. Protect credentials and private keys.

APPENDIX A | DEPLOYMENT WORKSHEETS

Complete these fields during planning and update them with as-built values. Add rows as needed. Never record plaintext passwords.

A.1 | NODE IDENTITY AND CONNECTIVITY

Node Hostname Front-end IP Front-end NIC Private IP Private NIC
Node 1
Node 2
Node 3
Node 4

A.2 | SHARED SITE VALUES

Site setting Approved value Site setting Approved value
Front-end prefix Private subnet/prefix
Default gateway Front-end MTU
DNS servers Private MTU
NTP servers Management/BMC subnet
Service VIP(s) Service DNS name(s)
Cluster name/use StratiSYSTEM release

A.3 | DISK OWNERSHIP AND DATA SAFETY

Node Device serial/WWN Capacity Purpose Wipe approved?
No / Yes
No / Yes
No / Yes
No / Yes
No / Yes
No / Yes

DISK INVENTORY RULE: Mark shared LUNs and devices carrying required data No. Never authorize destructive operations solely from Linux device letters.

APPENDIX B | CONFIGURATION FIELD MAP

These field definitions are a reading aid, not a complete or executable JSON configuration. Keep keys and defaults delivered with the installed release. Approved release definitions, protection policies and service settings take precedence.

B.1 | CLUSTERCONFIG — CORE VALUES

Field in current_deployment.json Purpose
ClusterNodeUserId / ClusterNodePwd Administrator identity and protected bootstrap secret.
ClusterPcsPwd / ClusterPwd Cluster service and bootstrap secrets.
FrontendAccessSubnet Front-end prefix, for example /24.
ClusterPrivateNetwork / ClusterPrivateSubnet Private address range and prefix.
AlertEmailsCommaDelimited Operations alert recipients.
SmbServicesVip SMB HA endpoint where enabled.
ClusterDefaultPoolType Default pool protection method; use approved design.
ClusterECPoolDefinition Advanced EC and placement settings; do not improvise.
EnableNfsHaServices / EnableS3HaServices Enable only when the approved design requires them.

B.2 | NODECONFIG — PER-NODE VALUES

Field Purpose
Name Must match installed node hostname.
FrontendAccessInterface / FrontendAccessBridge Service/client NIC or bond and bridge, normally br0.
FrontendAccessIpAddr / FrontendAccessGwAddr / FrontendAccessDnsAddrs Front-end IPv4 and routed network parameters.
FrontendAccessMtu End-to-end front-end MTU.
ClusterPrivateInterface / ClusterPrivateIpAddr Private NIC and unique inter-node IP.
ClusterPrivateMtu / ClusterPrivateBridge Private MTU; bridge usually blank for standard deployments.
ClusterPrivateGwAddr / ClusterPrivateDnsAddrs Normally empty for an isolated private network.
SmbPrimaryInterface Usually br0 where SMB is enabled.

USE THE ACTUAL STARTING FILE: Use the current_deployment.json supplied with the installed StratiSYSTEM preparation project. An arbitrary other-site copy can carry wrong addressing, credentials, protection rules or versions.

APPENDIX C | COMMAND QUICK REFERENCE

Run these from the appropriate node or authorized administration session. Replace placeholders such as <node-ip> and <verified-device> with approved values.

Purpose Command or location
SSH ssh root@<node-ip>
System identity hostnamectl
Addresses ip -4 address show
Routes ip route
Interfaces ip link show
Device inventory lsblk -o NAME,SIZE,TYPE,FSTYPE,MOUNTPOINTS,MODEL,SERIAL
Preparation cd /stratisystem-prep ; source start.sh
Preparation marker ls -l /.prepcomplete
Preparation log tail -n 200 /.preplog
StratiSTOR provisioner cd /stratistor && ./start.sh
StratiSERV provisioner cd /stratiserv && ./start.sh
List NVMe devices nvme list
Read-only signature check wipefs -n /dev/<verified-device>
Follow provisioner log tail -f <provisioner-log-path>
Web console https://<node-ip>:3199

FURTHER READING

For specialized capabilities and release-specific procedures, use SteelDome documentation. Advanced features are intentionally outside this standard deployment sequence.

APPENDIX D | FINAL ACCEPTANCE RECORD

Mark each applicable item complete, attach evidence and record customer signoff.

SIGNOFF RECORD

Acceptance Name / organization Date
Deployment engineer
Customer representative
Final approval / notes

DEPLOYMENT READY: Customer acceptance is appropriate only when all applicable checklist entries are complete, health is stable and handoff materials have been delivered.