Installation

The core installs into Kubernetes from a single signed Helm umbrella chart. This page covers a general install; for a production-shaped AWS walkthrough see Deploy to EKS.

Prerequisites

• Kubernetes 1.26+ with kubectl and Helm 3 pointed at it.
• cluster-admin for the first install — the bundled ClickHouse operator installs CRDs and a ClusterRole/Binding.
• A StorageClass for ClickHouse’s PVC (a cluster default, or set clickhouse.persistence.storageClass).
• Enough disk and CPU for your traffic and retention — see Sizing and capacity. The default CPU and memory suit up to about 1,000 records per second; size the volume for your retention — the default 50 GiB holds about 4 days at 1,000 spans per second.
• Network egress to ghcr.io/ultrvlt for images and the chart — or a private mirror for air-gapped installs.

1. Verify the chart (recommended)

Every release is cosign-signed with an offline-verifiable key. Verify before you install:

export ULTRVLT_VERSION=0.1.29   # use the version your vendor provided
curl -fsSLO https://docs.uvobservability.com/cosign.pub

cosign verify --key cosign.pub --insecure-ignore-tlog=true \
  ghcr.io/ultrvlt/charts/ultrvlt:$ULTRVLT_VERSION

See Verifying signed artifacts for images, SBOMs, and build provenance.

2. Create the license Secret (optional)

The core boots without a license — it runs UI-locked but still ingests, a safe default. To unlock the UI, mount your signed .lic from a Secret:

kubectl create namespace ultrvlt
kubectl -n ultrvlt create secret generic ultrvlt-license \
  --from-file=valid.lic=./customer.lic

The chart’s queryApi.license.existingSecret points at this Secret. (You can also inline the license with queryApi.license.content.)

3. Create the admin credential

The portal ships with auth.mode=local and is fail-closed: the install errors if no admin credential is provided. Prefer a pre-computed argon2id hash so no plaintext is stored at rest:

# Provide ONE of: a plaintext password (hashed at boot) or an argon2id PHC hash.
export ADMIN_HASH='$argon2id$v=19$m=65536,t=3,p=2$...'

If you run more than one query-api replica, also set a stable session-signing secret so replicas accept each other’s sessions:

export SESSION_SECRET=$(openssl rand -hex 32)

See Configuration → Authentication for OIDC single sign-on.

4. Install

helm install ultrvlt oci://ghcr.io/ultrvlt/charts/ultrvlt \
  --version "$ULTRVLT_VERSION" \
  -n ultrvlt --create-namespace \
  --set queryApi.license.existingSecret=ultrvlt-license \
  --set-string queryApi.auth.admin.passwordHash="$ADMIN_HASH" \
  --set queryApi.auth.admin.hashed=true \
  --set queryApi.auth.sessionSecret="$SESSION_SECRET"

The schema-init Job runs automatically on install and each upgrade.

5. Reach the portal

Enable the ingress (or port-forward for a quick look). Production installs should serve the portal over HTTPS — the session cookie is Secure, so login won’t stick over plain HTTP:

helm upgrade ultrvlt oci://ghcr.io/ultrvlt/charts/ultrvlt --version "$ULTRVLT_VERSION" \
  -n ultrvlt --reset-then-reuse-values \
  --set portalUi.ingress.enabled=true \
  --set portalUi.ingress.host=ultrvlt.example.com

--reset-then-reuse-values (Helm 3.14 or later) keeps the settings you’ve already made and also picks up defaults that newer versions add. Plain --reuse-values skips those new defaults, which can leave newly added settings unset after an upgrade.

Caution

Over plain HTTP (or a bare kubectl port-forward), login will appear to succeed but the Secure cookie won’t be stored — you’ll bounce back to the login page. Terminate TLS at your ingress and forward X-Forwarded-Proto: https.

6. Verify the core

# license is valid/unlocked:
kubectl -n ultrvlt exec deploy/ultrvlt-query-api -- \
  wget -qO- localhost:8080/api/v1/license
# clickhouse reachable:
kubectl -n ultrvlt exec deploy/ultrvlt-query-api -- \
  wget -qO- localhost:8080/api/v1/ping

7. Turn on data sources

With no data source the pages are empty until telemetry arrives. Enable one or more:

# eBPF sensor (validate the node kernel first — see Deploy to EKS):
--set sensor.enabled=true --set sensor.siteId=prod --set sensor.clusterId=prod

# pod-log collection (OpenTelemetry Collector DaemonSet):
--set logCollector.enabled=true

# Kubernetes object state for the Infrastructure pages:
--set clusterState.enabled=true

Any OpenTelemetry SDK or Collector can also send OTLP to the gateway on :4317. To take in telemetry from tools that aren’t OTLP/gRPC, turn on the opt-in ingestion integrations: a Prometheus remote-write endpoint and an OTLP/HTTP endpoint that lets Fluent Bit & Fluentd forward logs.

Air-gapped installs

The chart is offline-first. Mirror the signed images to your registry, vendor the subchart, and point registry at your mirror:

helm pull oci://ghcr.io/ultrvlt/charts/ultrvlt --version "$ULTRVLT_VERSION"
# mirror ghcr.io/ultrvlt/* images to your registry, then:
helm install ultrvlt ./ultrvlt-$ULTRVLT_VERSION.tgz -n ultrvlt --create-namespace \
  --set registry=registry.internal.example.com/ultrvlt \
  --set imagePullSecrets[0].name=my-mirror-pull-secret

Nothing in the core reaches the public internet unless you enable connected mode.