Component
core/ai-model/ollama
Ollama running in-cluster as a self-hosted inference server. A Recipe synthesizes the runtime directly: a weights-pull initContainer resolves the model pointer into a model-cache volume the serving container mounts, and a ClusterIP Service fronts the OpenAI-compatible API. The model value is a deterministic pointer resolved at pod start, never fetched at compile time. Exports its host, port, endpoint, model, and protocol to consumers via the uses topology, so an agent reaches the model without hardcoding an endpoint. On Kubernetes there are two realizations and the Manifest chooses between them. `ollama` synthesizes the runtime as a workload and fetches nothing, and it is the one to reach for where the build must complete with no chart repository reachable, or where more than one serving replica is wanted — it gives each replica its own model cache. `ollama-chart` renders the community-maintained Ollama Helm chart, which additionally renders a horizontal autoscaler from the replica bounds below, so reach for it where a chart repository is reachable at build time. That realization backs its single serving replica with one model cache and refuses a higher replica count rather than rendering one that cannot schedule. Both realizations run liveness and readiness probes against the serving endpoint, so a rollout reports the server ready only once it answers.
Provenance
- Content digest
- sha256:f4ba1749f86821df8fb448502daac7b19b489aaa7e7d05f6c367a8310b43c9cd
- Published
- 2026-11-10T10:09:28Z
- Signing identity
- https://token.actions.githubusercontent.com/https://github.com/aggrega-ai/aggrega-hub/.github/workflows/catalog-publish-sign.yml@refs/heads/main
Signature
- Officialapplication/vnd.dev.sigstore.bundle.v0.3+json
https://token.actions.githubusercontent.com/https://github.com/aggrega-ai/aggrega-hub/.github/workflows/catalog-publish-sign.yml@refs/heads/main
sha256:ec8388b4514dfcab227cd99381037b0806bc32599b85519d8af166924086b4ba
Reference at this version
oci://hub.aggrega.ai/component/core/ai-model/ollama:1.0.0Developer parameters
Values the developer sets in the Aggregate.
Extends BaseComponents#/schemas/BaseModelParameter
Operator sizing
Infrastructure sizing the operator tunes per environment.
| Name | Type | Default | Required | Description |
|---|---|---|---|---|
| port | integer (min 1024, max 65535) | 11434 | no | Server listen port for the model API. Most operators should omit this and let the Recipe assign the default (11434). |
| engine_version | string | 0.15.6 | no | Ollama runtime image tag. Selects the image tag on the synthesized serving and weights-pull containers; the chart-backed realization pins its own image tag. |
| source_version | string | — | no | Version of the Helm chart the chart-backed realization pulls. Read only by that realization; the workload-synthesis realization pulls no source artifact and ignores it. Changing it away from the Recipe's default does not guarantee value-mapping compatibility with the new chart. |
| replicas | integer (min 1) | 1 | no | Number of serving replicas. On the workload-synthesis realization each replica holds its own model-cache volume and pulls the weights independently; the chart-backed realization backs every replica with one shared cache and therefore serves one replica only, halting rather than rendering a count it cannot schedule. |
| min_replicas | integer (min 1) | — | no | Lower bound for the replica count when horizontal autoscaling is configured. Read together with max_replicas by the chart-backed realization, which renders a horizontal autoscaler from the pair; the workload-synthesis realization serves a fixed replica count and records the bound without acting on it. |
| max_replicas | integer (min 1) | — | no | Upper bound for the replica count when horizontal autoscaling is configured. Read together with min_replicas by the chart-backed realization, which renders a horizontal autoscaler from the pair; the workload-synthesis realization serves a fixed replica count and records the bound without acting on it. |
| compute | object | — | no | — |
| model_cache | object | — | no | Per-replica model-weights cache volume. Sized for the pulled model; each serving replica gets its own claim. |
| model_artifact | string | — | no | Optional OCI reference to the model weights, an alternative weights source to the Hugging Face model id in values.model. A deterministic pointer the runtime resolves at pod start; the compiler never fetches it. Declared ahead of rendering — the synth recipes resolve weights from values.model and do not yet consume this pointer. |
| overrides | object (open) | — | no | Provider-specific overrides. Accepts any nested structure. Each Recipe documents which keys it reads. |
| gpu | object | — | no | Optional GPU scheduling for the serving container. When set, the Recipe requests GPUs as the extended resource nvidia.com/gpu and, when a runtime class is named, schedules the pod under it. Requires a GPU-capable cluster and, for runtime_class, a matching RuntimeClass configured in the cluster. Omit the whole object to serve on CPU. |
Output contract
Values this component publishes to consumers over the uses topology.
| Name | Type | Default | Required | Description |
|---|---|---|---|---|
| host | string | — | yes | In-cluster DNS name of the Service fronting the inference server. A consumer that composes its own address dials this name; one that does not reads `endpoint` instead. |
| port | integer | — | yes | TCP port that Service accepts model requests on. It follows the port an operator set in the Release, so a consumer reads it rather than assuming a default. |
| endpoint | string | — | yes | Full base URL of the OpenAI-compatible API the server serves, with the host, the port and the API path already composed. A consumer hands it to a client library as its base URL. |
| model | string | — | yes | Name a request must ask this server for, which is the model it pulled at start rather than a provider catalogue name. |
| protocol | string | — | no | URL scheme to reach `host` and `port` over, for a consumer composing its own address rather than reading `endpoint`. |
Uses sockets
The dependency sockets this component connects through. The target of every connection wired through a socket must satisfy its label selector, and each connection may carry an attachment in the slots the socket declares. A required slot is owed by each connection the Aggregate or the Release wires through the socket.
No uses sockets declared.
Attachment slots
Named binding points at which a Feature attaches to this component. A slot's selector constrains which Feature may fill it and its inject list names how the Feature's material may be delivered. A required slot must be bound on every instance, by an attachment on the instance or at a scope above it, and a sensitive slot is declared to handle secret material. A Feature may also attach under a name no slot declares.
| Slot | Selector | Required | Inject | Sensitive | Description |
|---|---|---|---|---|---|
| model_pull_token | feature.aggrega.ai/kind=secret | no | env | yes | Optional model-registry pull credential for gated model weights. When attached, the weights-pull initContainer receives it as an environment variable; unattached, weights are pulled anonymously. |
Self-inject allowlist
env
Relationships
Typed edges to and from other published entities. A reference with no published target renders as a broken edge rather than a link.
Uses
- $refextends schema (values)
BaseComponents#/schemas/BaseModelParameter
Used by
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «model»
uses socket «target_model»
uses socket «extraction_model»
uses socket «embedding_model»
uses socket «models»
uses socket «models»
uses socket «models»
uses socket «models»
uses socket «models»
uses socket «models»
uses socket «models»
uses socket «upstream»
uses socket «models»
uses socket «forward_target»
uses socket «scrape_target»
uses socket «store»
uses socket «content_safety»
applies to
applies to
applies to
applies to
applies to
applies to
applies to
implements
implements
implements