Verify the Deployment
Pods that are Running prove that the containers started, not that OLIVE works. A deployment can be fully Ready while storage credentials are wrong, a network policy blocks a path that is only used at runtime, or an admission policy is silently evaluating nothing.
Run through this after importing the foundational OLIVE Connectivity Bundle (OCB).
1. The workloads are up
kubectl -n <tenant-namespace> get pods
kubectl -n <tenant-namespace> get events --sort-by=.lastTimestamp | tail -30
Every pod Running and Ready, and no repeated restarts. A pod that is Running but never Ready is a probe; a pod in CreateContainerConfigError is a missing Secret and produces no logs.
Remember that kubectl get all omits custom resources — check the ones OLIVE depends on explicitly:
kubectl -n <tenant-namespace> get certificate,vaultstaticsecret
2. The platform answers
Open https://<tenant>.<cluster>.<domain> and sign in. In a production deployment this exercises the whole authentication path at once: DNS, ingress, certificate, OAuth2 Proxy, Keycloak realm, and the JWKS call that every service makes. If sign-in works, a large part of the deployment is proven.
3. The liveness check is running
The foundational OCB contains a scheduled task, olive-platform-liveness-check-scheduler, which runs a connector every five minutes. The connector uses CouchDB, a file write, a file read, and data orchestration — so a successful run proves that the per-tenant data store, the object storage and the orchestrator are all working together, not merely that their pods are up.
Each run logs a security event with the result:
{"timestamp":"2026-09-15 09:00:14.566","level":"INFO","loglabel":"oli-security","messages":["event: OLIALIVE1","message: Liveness check result: ","service: data-orchestrator","userID: c6afca88-7c6a-49e1-a718-c0f8d545b8aa","ipAddress: ::ffff:172.31.19.193","remoteIpAddress: ::ffff:172.31.19.193","callingService: data-orchestrator","requestID: 2b2e2a52bb59","tenantID: multi","args: null","security","multi","b8669939"]}
The event code is OLIALIVE1, on the oli-security log label, emitted by data-orchestrator and carrying the tenantID.
kubectl -n <tenant-namespace> logs deploy/data-orchestrator --since=15m | grep OLIALIVE1
Within fifteen minutes you should see at least two entries. Nothing at all means either that the foundational bundle was not imported, or that the scheduler is not running.
The liveness check is not only an installation test. It keeps running, so it is the one signal that tells you the tenant still works end to end — and the useful alert is on its absence: no OLIALIVE1 for a tenant within, say, fifteen minutes means something in the chain broke. A failing result tells you what broke; silence tells you that something did, which is the harder thing to notice. See Log Events and Automated Reactions.
4. The security controls are actually in effect
Several of them fail silently when a prerequisite is missing, so a values file that says a control is on is not evidence that it is.
# Policies exist — and confirm with your operator that the CNI enforces them
kubectl -n <tenant-namespace> get networkpolicy
# The namespace is actually being evaluated; an empty report means it is not
kubectl get policyreport -n <tenant-namespace>
Whether the Kyverno policies are present and active is a cluster-scoped read — ask your cluster operator to confirm the policies are present and active if the report above is empty when it should not be.
If you enabled the service mesh, confirm that the namespace carries the enrolment label, that probes still pass, and that traffic is tunnelled. MFA for administrative roles is a Keycloak realm setting, not a chart value, so verify it in Keycloak.
Security Configuration has the sequence for narrowing down which control is responsible when something is wrong.
5. Connectivity reaches what it needs to
1.3 blocks calls to internal address ranges and Kubernetes-internal service names, logging OLIFETCH1. Public destinations are unaffected. Run one datapoint against each external system your content depends on rather than discovering the boundary later.
Related
- Production Readiness Checklist — the fuller sign-off list, for before an environment carries real data
- Troubleshooting — failure modes by symptom
- Security Configuration
- Log Events and Automated Reactions