PhotoPrism on GKE Autopilot — Lab Guide

Overview

Estimated time: 45–90 minutes

PhotoPrism is a self-hosted, AI-powered photo and video management application — it browses, organizes, and shares a personal media library with automatic tagging, facial recognition, and full-text/visual search, all served from a single Go binary with an embedded SQLite database. This lab takes you through the full operational lifecycle of the PhotoPrism on GKE Autopilot module on Google Cloud: deploy it, access and verify it, run it day-to-day, observe it, diagnose common problems, and tear it down.

The lab focuses on operating the GKE module and the Google Cloud platform, not on PhotoPrism product features. For the complete list of provisioned services and every configuration input (organised by group), see the Configuration Guide — this lab deliberately does not duplicate that detail so it stays accurate over time.

Objectives

By the end of this lab you will be able to:

Deploy the module from the RAD platform and locate the resources it provisions.
Connect to the GKE cluster and access the running StatefulSet, including its block Persistent Volume Claim.
Perform day-2 operations — inspect, manage secrets, and back up the media library.
Observe the workload with Cloud Logging and Cloud Monitoring.
Diagnose and resolve the most common deployment and runtime issues.
Tear the deployment down cleanly.

Prerequisites

Services_GCP deployed in the target project (provides the VPC, GKE Autopilot cluster, Artifact Registry, and shared service accounts this module depends on). PhotoPrism itself provisions no Cloud SQL instance — it uses an embedded SQLite database on a block PVC.
A Google Cloud project with billing enabled.
gcloud CLI and kubectl installed; gcloud auth login and gcloud auth application-default login completed.
Project Owner (or equivalent) IAM on the project.
RAD platform access with permission to deploy modules into the project.

Set these shell variables once; every task below reuses them:

export PROJECT="<your-gcp-project-id>"
export REGION="us-central1"           # the region you deploy into

Task 1 — Deploy the module [Automated]

Click Deploy in the RAD platform top navigation, open PhotoPrism (GKE) from the Platform Modules list to start configuration, set project_id, and review the inputs. Configure only what you need — the Configuration Guide documents every input by group, with defaults. Note that stateful_pvc_storage_class defaults to SSD-backed standard-rwo, which draws the tight SSD_TOTAL_GB quota — consider standard (HDD) if you are running this alongside other stateful modules on a quota-constrained project. Review the estimated cost (if credits are enabled) and click Deploy, which opens the deployment status page with real-time logs.
The platform deploys PhotoPrism into the GKE Autopilot cluster as a StatefulSet (pinned to exactly one replica, min=1, max=1) with a 20Gi block Persistent Volume Claim mounted at /photoprism, provisions the auto-generated PHOTOPRISM_ADMIN_PASSWORD Secret Manager secret, and builds and mirrors the container image. A storage Cloud Storage bucket is also created but is not mounted — the block PVC is the durable store here, not gcsfuse. There is no database-init job to wait on — PhotoPrism creates its own SQLite schema on first boot. First deploys typically complete in 10–20 minutes.

Connect to the cluster and discover the namespace with name-agnostic filters:

CLUSTER=$(gcloud container clusters list --project="$PROJECT" --format="value(name)" --limit=1)
gcloud container clusters get-credentials "$CLUSTER" --region="$REGION" --project="$PROJECT"

NS=$(kubectl get ns -o name | grep photoprism | head -1 | cut -d/ -f2)
echo "Cluster: $CLUSTER   Namespace: $NS"
kubectl get all,pvc -n "$NS"

Task 2 — Access & verify [Manual]

Confirm the workload is running and find its external address:
```
kubectl get pods,svc,statefulset,pvc -n "$NS"
EXTERNAL_IP=$(kubectl get svc -n "$NS" \
  -o jsonpath='{.items[?(@.spec.type=="LoadBalancer")].status.loadBalancer.ingress[0].ip}')
echo "External IP: $EXTERNAL_IP"
```
By default the Service is ClusterIP (not LoadBalancer); if EXTERNAL_IP is empty, check whether a custom domain / Gateway route was configured instead (enable_custom_domain = true by default) and use kubectl get gateway,httproute -n "$NS".
Confirm the service is healthy. PhotoPrism exposes an unauthenticated status endpoint that responds once the HTTP server is up and the SQLite index is ready:
```
curl -s "http://${EXTERNAL_IP}/api/v1/status"   # expect a 200 JSON response
```

Retrieve the auto-generated admin password before logging in — no pre-seeded credential is shown anywhere else:

SECRET=$(gcloud secrets list --project="$PROJECT" --filter="name~photoprism-admin-password" \
  --format="value(name)" --limit=1)
gcloud secrets versions access latest --secret="$SECRET" --project="$PROJECT"

Open http://${EXTERNAL_IP} (or your custom domain) in a browser and sign in with username admin (or your configured admin_username) and the password retrieved above. Once you know the external URL, consider setting site_url to it in the RAD platform and applying via Update — this fixes absolute links and thumbnail URLs that otherwise fall back to the request host.

Task 3 — Operate & keep it running (Day-2) [Manual]

Inspect the workload — StatefulSet, pod, and the block PVC (there is no HPA to inspect; PhotoPrism is pinned to one replica):
```
kubectl get statefulset,pods,pvc -n "$NS"
kubectl describe statefulset -n "$NS"
```
Do not scale beyond one replica. min_instance_count and max_instance_count are both fixed at 1 by design — PhotoPrism serves one shared SQLite library from one writable block PVC, and a second concurrent writer risks database and index corruption. This is not a scaling dial to tune.
Tune resources for your library size. cpu_limit/memory_limit default to 1000m/1Gi, but PhotoPrism loads vector indexes into memory for face recognition and thumbnailing, and the application layer's own baseline recommendation is 4Gi for real indexing workloads. If you see pod restarts from OOM in kubectl describe pod events (Task 4) as your library grows, raise memory_limit in the RAD platform and apply via Update. Also confirm stateful_fs_group = 3000 remains set — a mismatched or missing fsGroup leaves the PVC non-writable by PhotoPrism's UID 1000/GID 2000 and blocks startup.
Update the application version by changing the version input in the RAD platform and applying it via Update; a new image builds (pinned to a PHOTOPRISM_VERSION build tag, not the generic version input, when left at latest) and the StatefulSet's pod is replaced.

Manage secrets and the PVC:

kubectl get secrets -n "$NS"
gcloud secrets list --project="$PROJECT" --filter="name~photoprism"
gcloud compute disks list --project="$PROJECT" --filter="name~$NS"

Back up the media library. Because PhotoPrism has no SQL database, a backup is a filesystem archive of the PVC contents (backup_format = tar by default), not a database dump. Review backup_schedule and backup_retention_days in the RAD platform, and raise retention for production libraries.

Task 4 — Observe: Logging & Monitoring [Manual]

Logs — from kubectl or the Logs Explorer:
```
kubectl logs -n "$NS" statefulset/"$(kubectl get statefulset -n "$NS" -o jsonpath='{.items[0].metadata.name}')" --tail=50
```
Logs Explorer filter: resource.type="k8s_container" AND resource.labels.namespace_name="<namespace>".
Monitoring — open the GKE / Kubernetes dashboards and review pod CPU and memory utilisation and restart counts — watch memory closely as your library grows, since indexing and thumbnailing are memory-hungry, and also watch the PVC's disk utilisation against its 20Gi default size. The module can provision an uptime check (disabled by default); enable it under uptime_check_config and confirm it is green under Monitoring → Uptime checks.

Task 5 — Troubleshoot & debug [Manual]

Durable techniques for the failure modes you are most likely to hit. These are platform-level diagnostics and do not change with PhotoPrism releases.

Pod not Ready / CrashLoopBackOff: inspect events and logs. The startup/liveness probes both target /api/v1/status (no auth required); a PVC mount or fsGroup mismatch will keep the pod from becoming Ready.
```
kubectl describe pod -n "$NS" <pod>          # Events section shows scheduling/probe/mount errors
kubectl logs -n "$NS" <pod> --previous       # logs from the crashed container
```
Pod stuck Pending with Quota 'SSD_TOTAL_GB' exceeded: the default standard-rwo StorageClass draws the tight regional SSD quota; override to HDD with -var stateful_pvc_storage_class=standard on quota-constrained projects. Scaling the pod to zero does not free the PVC — only deleting it reclaims quota.
Container OOM-killed: check pod restart count and events; if memory is pinned near the memory_limit ceiling, raise it (see Task 3) — 1Gi is the module default but under-sized for real libraries.
Locked out of the admin account: re-read the password from Secret Manager (Task 2); it is the source of truth and PhotoPrism re-applies it at every boot.
Pending pod / no external IP: check kubectl describe pod events for resource or quota issues, and confirm the Service/Gateway has an assigned IP.
Image pull errors: confirm the image exists in Artifact Registry and the node service account can pull it.

See the Configuration Guide's Configuration Pitfalls section for setting-specific gotchas (including why max_instance_count must never be raised above 1, and why disabling stateful_pvc_enabled falls back to gcsfuse, which cannot safely host SQLite).

Task 6 — Tear down [Automated]

On the Deployments page, open the deployment and click the Trash icon (Delete). Delete runs terraform destroy and is irreversible (the deployment record is retained for history). If a deployment is stuck and the RAD platform can no longer manage it (for example after manual changes that conflict with the Terraform state), use Purge instead — it removes the deployment from RAD's records without destroying the cloud resources (it makes RAD forget the project). This removes everything the module created — the Kubernetes StatefulSet and namespace, the block Persistent Disk (including the embedded SQLite database and all originals it contains), the unused storage GCS bucket, and Secret Manager secrets. Resources owned by Services_GCP (the VPC, GKE cluster, Artifact Registry) are managed separately and are not removed here.

Summary

Task	Type	Outcome
1 — Deploy	Automated	Module deploys a single-replica StatefulSet with a 20Gi block PVC and the admin-password secret — no database init to wait on
2 — Access & verify	Manual	Connect to the cluster; health check passes; retrieve the auto-generated admin password and sign in
3 — Operate	Manual	Inspect the StatefulSet/PVC (never scale beyond 1), tune resources for library size, update version, manage secrets and backups
4 — Observe	Manual	Query Cloud Logging; review Cloud Monitoring metrics, especially memory and PVC utilisation, and uptime check
5 — Troubleshoot	Manual	Diagnose pod, SSD-quota, OOM, admin-credential, scheduling, and image-pull issues
6 — Tear down	Automated	Delete (Trash) removes all module resources, including the block PVC and its data

By Dr Shiyghan Emmanuel Navti · Google Cloud certified: ACE · PCA · PCD · PDE · PSE · CDL

Overview​

Objectives​

Prerequisites​

Task 1 — Deploy the module [Automated]​

Task 2 — Access & verify [Manual]​

Task 3 — Operate & keep it running (Day-2) [Manual]​

Task 4 — Observe: Logging & Monitoring [Manual]​

Task 5 — Troubleshoot & debug [Manual]​

Task 6 — Tear down [Automated]​

Summary​