Kubernetes
Deploy into your cluster
Kubernetes manifests provide a starting point for service configuration, storage, networking, and secrets. Build or supply your own image, then review the topology and storage settings for your workload.
Enterprise · On-premise
Your agents. Your infrastructure.
Keep your agent’s memory in an environment your team operates. Akasha, the engine behind Minds, can run on your own infrastructure, with source and deployment resources for teams that need control over data location, networking, and operations.
Enterprise · deployment planning
Your environment. Your operating model.
Choose a starting point, then make the infrastructure decisions explicit.
Deployment worksheet
YOUR INFRASTRUCTUREDeploy into your cluster.
Adapt the supplied manifests to the topology, storage, and network policy your team operates.
- Artifact
- Kubernetes manifests
- Data
- /var/lib/akasha · PVC
- Transport
- TLS certificates you mount
- Review
- Instance count and storage together
Make the responsibilities visible.
Decide who handles updates, backups, recovery exercises, and monitoring before your team starts using the service.
Deployment planning example. Talk with us about the configuration and support your environment needs.
Deployment options
Start with the way your team works.
Evaluate locally, build your own container, or adapt the Kubernetes manifests to your environment. Your team configures and operates the infrastructure. We’ll help define the deployment and support requirements with you.
Docker Compose
Evaluate with a local stack
Docker Compose brings the engine, UI, and supporting services together for local evaluation. Included monitoring tools help you inspect the stack as you test.
Source installer
Build directly from source
The source installer builds the engine and command-line tools into a location you choose. The build requires network access to obtain its dependencies.
Dockerfile
Use a container you control
A multi-stage Dockerfile packages the engine in a Debian-based runtime running as a non-root user. Your team owns the image build, registry, updates, and deployment policy.
Before deployment
Configure for the environment you run.
Review the image, secrets, storage, TLS certificates, and monitoring endpoints before applying the manifests. Choose an instance and storage topology suited to the deployment; the supplied replica settings need review before production use.
Deployment checklist
- Topology
- Review instance count and persistent storage together
- Ports
- 8080 HTTP · 50051 gRPC · 8815 Arrow Flight
- Process
- UID 1000 · read-only root · capabilities dropped
- Data
- PVC at /var/lib/akasha
- TLS
- require_tls in daemon.toml · certs you mount
- Tracing
- Configure an OTLP destination such as in-cluster Tempo
kubectl applybash
kubectl apply -f akasha-daemon/deployment/k8s/
kubectl -n akasha-system get podsDisconnected environments
Plan for every dependency.
The published source installer and container build require network access. Minds does not currently provide a packaged offline distribution. If your environment must be disconnected, discuss dependency preparation, model access, updates, and support with us before committing to that deployment.