Admiral v00041 & v00042: Streamlined ARM64 Pipelines, Native Snapshots, and Hardened Workload Control

We are officially shipping Admiral v00042!

Note: For those tracking our tags, our intermediate v00041 build encountered a CI packaging failure. We have rolled all of those changes forward into this unified v00042 release.

While our preceding releases focused deeply on WAN resilience and kernel-level storage configurations, this release cycles back to foundational OS architecture on low-power platforms, workload snapshots, and expanded hardware security boundaries.

Here is an in-depth breakdown of what is inside this update.

1. Unified ARM64 Updates & OS Hardening

Deploying OS updates to resource-constrained ARM64 edge units (such as single-board gateways and micro-controllers) presents unique challenges compared to standardized x86_64 server architectures. With v00042, we are introducing a heavily reworked architectural framework for ARM64:

• U-Boot Update Engine: We have introduced a native U-Boot upgrade mechanism. The system can now flash and verify low-level bootloader partitions on ARM64 nodes safely without risking corruption in the event of mid-install power losses.
• Architecture Gating & Validation: The host upgrade flow now aggressively verifies both binary-level and rootfs-level architecture descriptors during the download and staging steps, entirely eliminating the risk of cross-architecture “mis-flashing.”
• Uniform Init & Supervisor Architecture: Building directly on the unified multi-arch boot pipeline established in v00039, our custom initialization and supervisor lifecycles have now been fully consolidated. Standardizing these system-level behaviors across all host architectures guarantees that critical process control loops behave identically on an industrial ARM gateway as they do on an x86 server.
• Automated Installations & Selection Mode: Our arm64 provisioning toolset has received a major upgrade, featuring an automated bare-metal installer and dynamic installation-mode selection managed via /app/install_mode.

2. Storage Evolution & Workload Lifecycle Policies

Managing storage at the edge requires high safety margins and minimal write overhead. To achieve this, this release introduces dynamic storage actions and brings cloud-scale file layouts to the edge:

• Migration to Optimized Workload Storage Paths: Custom workload volumes have been transitioned to an advanced storage layout. This arrangement provides strict administrative boundaries, faster cleanup operations, and lower filesystem overhead.
• Automated Volume Restore Points: Leveraging our updated storage layout, Admiral now supports automatic workload volume rollbacks. System engineers can declare snapshot intervals directly within the workload definition, enabling atomic point-in-time recovery points.
• Automated Image and Binary Cleaning: We implemented an automated image pruning manager. Additionally, Admiral now automatically rotates deployment binaries, cleaning up old references in /app while preserving the last 3 working releases.
• AWS ECR Authenticated Image Pulling: You can now natively pull images from private Amazon Elastic Container Registry (ECR) repositories directly. Authentication handshakes are handled automatically at the agent level.
• Cron-Scheduled Workload Rotations: Workloads can now run scheduled execution windows natively. You can now define explicit cron routines inside the deployment specification file to trigger automated, scheduled workload syncs and service updates.

3. Sophisticated Workload & Hardware Routing

As applications running at the edge become more demanding, they require safer, more sophisticated ways to communicate with the host operating system:

• Preloaded Offline Booting (/admrl/preload): [Enterprise Feature] Newly provisioned hardware can now boot up with a preloaded image and functional configuration located at /admrl/preload. This enables true “unconnected” provisioning states for environments that lack DHCP access during the initial staging phase. If you are interested in enabling offline staging for your operations, contact our team for more details.
• Integrated Telemetry Channel Passthrough: Workloads can optionally mount a dedicated telemetry pipeline directly into their execution namespace. This allows legacy applications to relay customized logs directly back to the cloud via the host’s encrypted tunnel without requiring custom VPN configurations.
• Direct Host System Event Access: For workloads requiring immediate system management coordination, workload specifications can now request secure, mediated routing access to host system events.

4. Under-the-Hood Polish

• SquashFS Segmentation: For our x86 platforms, we have segmented our SquashFS layout logic (encompassing system rootfs, modules, and firmware tags) to streamline and safeguard future system update paths.
• Stream Keepalive Protections: Fixed an edge-case in our remote support engine where failed connections stayed alive in a broken state indefinitely, preventing recovery sessions from re-binding to system ports.
• Dependency Cleanups: We ran our structural packages through clean golang module vendoring workflows (go mod vendor).

🚀 Coming Soon: Advanced Spatial Bluetooth Analytics

We are actively testing native Bluetooth Dwell Calculations and real-time iOS Beacon Type parsing directly inside the Admiral telemetry pipeline. These features are designed to enable precise physical site interaction metrics without overloading local container runtimes. Stay tuned for their wide release in an upcoming release cycle!