Add cheeztv kids Jellyfin instance and fafflix kids mounts
ci/woodpecker/pr/vector-test Pipeline was successful
ci/woodpecker/pr/pre-commit Pipeline was successful
ci/woodpecker/pr/kubeconform Pipeline was successful

cheeztv is a second, kids-only Jellyfin instance in the media project. Kids
titles need a separate, safe library front (cheeztv.unkin.net) while staying
resumable in the existing adult instance (fafflix).

- apps/base/cheeztv: mirror the jellyfin (fafflix) HA stack 1:1 - same
  jellyfin-ha:v0.1.3 fork image, 2-replica StatefulSet, RWX transcode PVC,
  Intel iGPU transcode, CNPG Postgres + PgBouncer, Valkey transcode store,
  k8up config backup, VMPodScrape. Media mounts subPath kids on the shared
  movies/tv CephFS subvolumes so cheeztv sees only the kids trees.
- Separate state: own namespace, config PVC, cheeztv-postgres cluster,
  cheeztv-valkey, and cnpg-cheeztv / cheeztv-config-backup buckets - nothing
  shared with fafflix.
- Ingress/DNS: cheeztv.unkin.net (internal Traefik + external-dns at
  198.18.200.4, cert-manager cheeztv-tls) plus the cluster hostname variant
  cheeztv.k8s.syd1.au.unkin.net, matching how fafflix and logviewer.unkin.net
  are wired.
- fafflix: add movies/kids and tvshows/kids subPath mounts alongside its
  existing media mounts so kids libraries are browsable/resumable there; its
  existing mounts, hostname and ingress are untouched.
- Register cheeztv in the media ApplicationSet generator and AppProject
  destinations.
This commit is contained in:
2026-08-24 22:08:04 +10:00
parent a064d9a3c4
commit a4253f09cd
26 changed files with 1072 additions and 0 deletions
@@ -0,0 +1,32 @@
---
# Second Ceph RGW (S3) bucket owned by the existing cheeztv backup user
# (cnpg-cheeztv-backup, defined in cnpg_backup.yaml) — one user, two buckets:
# the CNPG barman bucket plus this one, which k8up uses to hold restic backups
# of the cheeztv-config PVC. The BucketAccess emits read-write S3 creds into a
# Secret the k8up Schedule consumes.
apiVersion: ceph.unkin.net/v1alpha1
kind: Bucket
metadata:
name: cheeztv-config-backup
namespace: cheeztv
spec:
placementTarget: ec
bucketName: cheeztv-config-backup
ownerRef: cnpg-cheeztv-backup
versioning: false
tags:
app: cheeztv
purpose: config-backup
retainOnDelete: true
---
apiVersion: ceph.unkin.net/v1alpha1
kind: BucketAccess
metadata:
name: cheeztv-config-backup
namespace: cheeztv
spec:
bucketRef: cheeztv-config-backup
level: read-write
# Operator writes AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY (+ S3_ENDPOINT,
# BUCKET_NAME) into this Secret; the k8up Schedule reads the access keys.
secretName: cheeztv-config-backup-s3
+45
View File
@@ -0,0 +1,45 @@
---
# Ceph RGW (S3) backup target for the cheeztv CNPG cluster, provisioned by the
# in-estate cephrgw-operator: one dedicated bucket + owner user. CNPG reads the
# S3 credential Secret from its own namespace.
apiVersion: ceph.unkin.net/v1alpha1
kind: ObjectStoreUser
metadata:
name: cnpg-cheeztv-backup
namespace: cheeztv
spec:
displayName: "CNPG backup owner (cheeztv)"
uid: cnpg-cheeztv-backup
maxBuckets: 5
secretName: cnpg-cheeztv-backup-s3
retainOnDelete: true
---
apiVersion: ceph.unkin.net/v1alpha1
kind: Bucket
metadata:
name: cnpg-cheeztv
namespace: cheeztv
spec:
placementTarget: ec
bucketName: cnpg-cheeztv
ownerRef: cnpg-cheeztv-backup
versioning: false
tags:
app: cheeztv
purpose: cnpg-backup
retainOnDelete: true
---
# Nightly base backup on top of always-on WAL archiving. Scheduled off-peak and
# staggered from the other CNPG clusters (6-field cron, seconds first).
apiVersion: postgresql.cnpg.io/v1
kind: ScheduledBackup
metadata:
name: cnpg-cheeztv-nightly
namespace: cheeztv
spec:
schedule: "0 35 3 * * *"
immediate: false
backupOwnerReference: self
method: barmanObjectStore
cluster:
name: cheeztv-postgres
+126
View File
@@ -0,0 +1,126 @@
---
# Main Jellyfin database. The cheeztv-ha fork's experimental EF Core provider
# moves the entire Jellyfin DB (incl. library items) off SQLite into PostgreSQL,
# which is what makes a shared-nothing multi-replica deployment possible. No
# bootstrap secret is given, so CNPG generates the cheeztv-postgres-app secret
# (username/password/dbname) that the StatefulSet composes its DSN from.
apiVersion: postgresql.cnpg.io/v1
kind: Cluster
metadata:
name: cheeztv-postgres
namespace: cheeztv
spec:
# Exclude the operator-managed data PVCs (cheeztv-postgres-N) from the
# cheeztv-config k8up Schedule (skipWithoutAnnotation is false cluster-wide,
# so unannotated PVCs are swept in). Postgres has its own barmanObjectStore
# backup below; restic must not touch the raw RWO data volumes.
inheritedMetadata:
annotations:
k8up.io/backup: "false"
affinity:
podAntiAffinityType: preferred
backup:
retentionPolicy: 30d
barmanObjectStore:
# Dedicated per-cluster Ceph RGW bucket (cephrgw-operator provisions it).
destinationPath: s3://cnpg-cheeztv
endpointURL: https://s3.ceph.unkin.net
endpointCA:
name: vault-ca-cert
key: ca.crt
s3Credentials:
accessKeyId:
name: cnpg-cheeztv-backup-s3
key: AWS_ACCESS_KEY_ID
secretAccessKey:
name: cnpg-cheeztv-backup-s3
key: AWS_SECRET_ACCESS_KEY
serverName: cheeztv
data:
compression: bzip2
jobs: 2
wal:
compression: zstd
maxParallel: 2
bootstrap:
initdb:
database: cheeztv
encoding: UTF8
localeCType: C
localeCollate: C
owner: cheeztv
enablePDB: true
enableSuperuserAccess: false
failoverDelay: 0
# PG 17 — accepted by the fork's Npgsql/EF Core provider (needs PG14+); the
# provider generates its own migrations on first start.
imageName: ghcr.io/cloudnative-pg/postgresql:17-system-trixie
instances: 3
logLevel: info
maxSyncReplicas: 0
minSyncReplicas: 0
monitoring:
customQueriesConfigMap:
- key: queries
name: cnpg-default-monitoring
disableDefaultQueries: false
enablePodMonitor: false
postgresql:
parameters:
archive_mode: "on"
archive_timeout: 5min
dynamic_shared_memory_type: posix
effective_cache_size: 256MB
full_page_writes: "on"
log_destination: csvlog
log_directory: /controller/log
log_filename: postgres
log_rotation_age: "0"
log_rotation_size: "0"
log_truncate_on_rotation: "false"
logging_collector: "on"
max_connections: "200"
max_parallel_workers: "16"
max_replication_slots: "16"
max_worker_processes: "16"
shared_buffers: 128MB
shared_memory_type: mmap
ssl_max_protocol_version: TLSv1.3
ssl_min_protocol_version: TLSv1.3
wal_keep_size: 256MB
wal_level: logical
wal_log_hints: "on"
wal_receiver_timeout: 5s
wal_sender_timeout: 5s
syncReplicaElectionConstraint:
enabled: false
primaryUpdateMethod: restart
primaryUpdateStrategy: unsupervised
probes:
liveness:
isolationCheck:
connectionTimeout: 1000
enabled: true
requestTimeout: 1000
replicationSlots:
highAvailability:
enabled: true
slotPrefix: _cnpg_
synchronizeReplicas:
enabled: true
updateInterval: 30
resources:
limits:
cpu: "1"
memory: 1Gi
requests:
cpu: 50m
memory: 512Mi
smartShutdownTimeout: 180
startDelay: 3600
stopDelay: 1800
storage:
resizeInUseVolumes: true
size: 10Gi
storageClass: cephrbd-fast-delete
switchoverDelay: 3600
+36
View File
@@ -0,0 +1,36 @@
---
# PgBouncer pooler in front of the cheeztv-postgres cluster. Jellyfin connects
# here (cheeztv-postgres-pooler:5432) rather than the -rw service so EF Core's
# connection churn is absorbed by the pool.
apiVersion: postgresql.cnpg.io/v1
kind: Pooler
metadata:
name: cheeztv-postgres-pooler
namespace: cheeztv
spec:
cluster:
name: cheeztv-postgres
instances: 2
pgbouncer:
parameters:
default_pool_size: "50"
max_client_conn: "200"
paused: false
poolMode: session
template:
metadata:
labels:
app: cheeztv-pooler
spec:
affinity:
podAntiAffinity:
requiredDuringSchedulingIgnoredDuringExecution:
- labelSelector:
matchExpressions:
- key: app
operator: In
values:
- cheeztv-pooler
topologyKey: kubernetes.io/hostname
containers: []
type: rw
+81
View File
@@ -0,0 +1,81 @@
---
# Friendly public hostname for the kids instance, cheeztv.unkin.net. Same
# internal-Traefik + external-dns pattern as logviewer.unkin.net: external-dns
# publishes the A record at the internal LB VIP (198.18.200.4) and cert-manager
# mints cheeztv-tls (CN cheeztv.unkin.net) off the Vault-PKI cluster issuer.
apiVersion: gateway.networking.k8s.io/v1
kind: Gateway
metadata:
labels:
traefik.io/instance: internal
annotations:
cert-manager.io/cluster-issuer: vault-issuer
cert-manager.io/common-name: cheeztv.unkin.net
cert-manager.io/private-key-size: "4096"
external-dns.alpha.kubernetes.io/hostname: cheeztv.unkin.net
external-dns.alpha.kubernetes.io/target: 198.18.200.4
name: cheeztv
namespace: cheeztv
spec:
gatewayClassName: traefik-internal
listeners:
- allowedRoutes:
namespaces:
from: Same
hostname: cheeztv.unkin.net
name: http
port: 80
protocol: HTTP
- allowedRoutes:
namespaces:
from: Same
hostname: cheeztv.unkin.net
name: https
port: 443
protocol: HTTPS
tls:
certificateRefs:
- group: ""
kind: Secret
name: cheeztv-tls
mode: Terminate
---
# Cluster hostname variant, mirroring fafflix's jellyfin.k8s.syd1.au.unkin.net
# gateway 1:1 (internal Traefik, external-dns at 198.18.200.4). Own cert secret
# so each hostname gets its own leaf.
apiVersion: gateway.networking.k8s.io/v1
kind: Gateway
metadata:
labels:
traefik.io/instance: internal
annotations:
cert-manager.io/cluster-issuer: vault-issuer
cert-manager.io/common-name: cheeztv.k8s.syd1.au.unkin.net
cert-manager.io/private-key-size: "4096"
external-dns.alpha.kubernetes.io/hostname: cheeztv.k8s.syd1.au.unkin.net
external-dns.alpha.kubernetes.io/target: 198.18.200.4
name: cheeztv-internal
namespace: cheeztv
spec:
gatewayClassName: traefik-internal
listeners:
- allowedRoutes:
namespaces:
from: Same
hostname: cheeztv.k8s.syd1.au.unkin.net
name: http
port: 80
protocol: HTTP
- allowedRoutes:
namespaces:
from: Same
hostname: cheeztv.k8s.syd1.au.unkin.net
name: https
port: 443
protocol: HTTPS
tls:
certificateRefs:
- group: ""
kind: Secret
name: cheeztv-internal-tls
mode: Terminate
+98
View File
@@ -0,0 +1,98 @@
---
apiVersion: gateway.networking.k8s.io/v1
kind: HTTPRoute
metadata:
name: http-redirect
namespace: cheeztv
spec:
hostnames:
- cheeztv.unkin.net
parentRefs:
- group: gateway.networking.k8s.io
kind: Gateway
name: cheeztv
sectionName: http
rules:
- filters:
- type: RequestRedirect
requestRedirect:
scheme: https
statusCode: 301
matches:
- path:
type: PathPrefix
value: /
---
apiVersion: gateway.networking.k8s.io/v1
kind: HTTPRoute
metadata:
name: cheeztv-route
namespace: cheeztv
spec:
hostnames:
- cheeztv.unkin.net
parentRefs:
- group: gateway.networking.k8s.io
kind: Gateway
name: cheeztv
sectionName: https
rules:
- backendRefs:
- group: ""
kind: Service
name: cheeztv
port: 8096
weight: 1
matches:
- path:
type: PathPrefix
value: /
---
apiVersion: gateway.networking.k8s.io/v1
kind: HTTPRoute
metadata:
name: http-redirect-internal
namespace: cheeztv
spec:
hostnames:
- cheeztv.k8s.syd1.au.unkin.net
parentRefs:
- group: gateway.networking.k8s.io
kind: Gateway
name: cheeztv-internal
sectionName: http
rules:
- filters:
- type: RequestRedirect
requestRedirect:
scheme: https
statusCode: 301
matches:
- path:
type: PathPrefix
value: /
---
apiVersion: gateway.networking.k8s.io/v1
kind: HTTPRoute
metadata:
name: cheeztv-route-internal
namespace: cheeztv
spec:
hostnames:
- cheeztv.k8s.syd1.au.unkin.net
parentRefs:
- group: gateway.networking.k8s.io
kind: Gateway
name: cheeztv-internal
sectionName: https
rules:
- backendRefs:
- group: ""
kind: Service
name: cheeztv
port: 8096
weight: 1
matches:
- path:
type: PathPrefix
value: /
+26
View File
@@ -0,0 +1,26 @@
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
resources:
- namespace.yaml
- cnpg_cluster.yaml
- cnpg_pooler.yaml
- cnpg_backup.yaml
- cephrgw-config-backup.yaml
- vaultauth.yaml
- vaultstaticsecret.yaml
- schedule.yaml
- pvc-config.yaml
- pvc-transcode.yaml
- pv-media-tv.yaml
- pv-media-movies.yaml
- pvc-media-tv.yaml
- pvc-media-movies.yaml
- statefulset.yaml
- pdb.yaml
- service.yaml
- valkey.yaml
- vmpodscrape.yaml
- gateway.yaml
- httproute.yaml
+5
View File
@@ -0,0 +1,5 @@
---
apiVersion: v1
kind: Namespace
metadata:
name: cheeztv
+13
View File
@@ -0,0 +1,13 @@
---
# Keep at least one Jellyfin replica serving through voluntary disruptions
# (node drains, rollouts) so active streams can fail over rather than drop.
apiVersion: policy/v1
kind: PodDisruptionBudget
metadata:
name: cheeztv
namespace: cheeztv
spec:
minAvailable: 1
selector:
matchLabels:
app: cheeztv
+30
View File
@@ -0,0 +1,30 @@
---
# Static PV for the shared MOVIES CephFS subvolume. Same rootPath as arrstack's
# movies PV so radarr writes and cheeztv reads the identical library tree; each
# namespace gets its own PV (unique name + volumeHandle) pinned by claimRef.
apiVersion: v1
kind: PersistentVolume
metadata:
name: cheeztv-media-movies
spec:
capacity:
storage: 1Ti
accessModes:
- ReadWriteMany
persistentVolumeReclaimPolicy: Retain
storageClassName: ""
volumeMode: Filesystem
claimRef:
namespace: cheeztv
name: cheeztv-media-movies
csi:
driver: cephfs.csi.ceph.com
volumeHandle: cheeztv-media-movies-static
nodeStageSecretRef:
name: csi-cephfs-secret
namespace: csi-cephfs
volumeAttributes:
staticVolume: "true"
clusterID: cephfs_csi_ssd_ec_4_1
fsName: cephfs
rootPath: /volumes/csi_ssd_ec_4_1/media-movies/e95d8ace-c736-465a-acc3-0c3e46dcede9
+30
View File
@@ -0,0 +1,30 @@
---
# Static PV for the shared TV CephFS subvolume. Same rootPath as arrstack's
# TV PV so sonarr writes and cheeztv reads the identical library tree; each
# namespace gets its own PV (unique name + volumeHandle) pinned by claimRef.
apiVersion: v1
kind: PersistentVolume
metadata:
name: cheeztv-media-tv
spec:
capacity:
storage: 1Ti
accessModes:
- ReadWriteMany
persistentVolumeReclaimPolicy: Retain
storageClassName: ""
volumeMode: Filesystem
claimRef:
namespace: cheeztv
name: cheeztv-media-tv
csi:
driver: cephfs.csi.ceph.com
volumeHandle: cheeztv-media-tv-static
nodeStageSecretRef:
name: csi-cephfs-secret
namespace: csi-cephfs
volumeAttributes:
staticVolume: "true"
clusterID: cephfs_csi_ssd_ec_4_1
fsName: cephfs
rootPath: /volumes/csi_ssd_ec_4_1/media-tv/4692957d-f5df-4f72-b9c9-56e4ee6d1333
+17
View File
@@ -0,0 +1,17 @@
---
# Jellyfin config: metadata images, plugins, subtitles and config XML. Shared
# ReadWriteMany across replicas (all pods read/write the same library metadata);
# the main library DB now lives in PostgreSQL, not here. Retain — this is state.
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: cheeztv-config
namespace: cheeztv
spec:
accessModes:
- ReadWriteMany
resources:
requests:
storage: 20Gi
storageClassName: cephfs-raid5-retain
volumeMode: Filesystem
+24
View File
@@ -0,0 +1,24 @@
---
# Movie library, shared read-many across replicas. Statically bound to the
# cheeztv-media-movies PV (shared CephFS subvolume also used by arrstack/radarr).
# storageClassName "" + volumeName disables dynamic provisioning and binds the
# pre-created static PV.
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: cheeztv-media-movies
namespace: cheeztv
annotations:
# Exclude from the cheeztv-config k8up Schedule (skipWithoutAnnotation is
# false cluster-wide, so unannotated PVCs are swept in). Only cheeztv-config
# is backed up; the media library is not restic-backup material.
k8up.io/backup: "false"
spec:
accessModes:
- ReadWriteMany
resources:
requests:
storage: 1Ti
storageClassName: ""
volumeName: cheeztv-media-movies
volumeMode: Filesystem
+24
View File
@@ -0,0 +1,24 @@
---
# TV library, shared read-many across replicas. Statically bound to the
# cheeztv-media-tv PV (shared CephFS subvolume also used by arrstack/sonarr).
# storageClassName "" + volumeName disables dynamic provisioning and binds the
# pre-created static PV.
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: cheeztv-media-tv
namespace: cheeztv
annotations:
# Exclude from the cheeztv-config k8up Schedule (skipWithoutAnnotation is
# false cluster-wide, so unannotated PVCs are swept in). Only cheeztv-config
# is backed up; the media library is not restic-backup material.
k8up.io/backup: "false"
spec:
accessModes:
- ReadWriteMany
resources:
requests:
storage: 1Ti
storageClassName: ""
volumeName: cheeztv-media-tv
volumeMode: Filesystem
+23
View File
@@ -0,0 +1,23 @@
---
# Shared transcode scratch. ReadWriteMany is the hard requirement for the HA
# fork: a taking-over pod must read the in-flight HLS segments written by the
# pod it replaces. Scratch data (delete reclaim); raid5 avoids the raid6
# double-parity write penalty on the many small HLS segment writes.
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: cheeztv-transcode
namespace: cheeztv
annotations:
# Exclude from the cheeztv-config k8up Schedule (skipWithoutAnnotation is
# false cluster-wide, so unannotated PVCs are swept in). Transcode is RWX
# scratch — nothing to back up.
k8up.io/backup: "false"
spec:
accessModes:
- ReadWriteMany
resources:
requests:
storage: 100Gi
storageClassName: cephfs-raid5-delete
volumeMode: Filesystem
+58
View File
@@ -0,0 +1,58 @@
---
# k8up Schedule: restic backups of the cheeztv-config PVC (library metadata,
# plugins, config XML) to the dedicated Ceph RGW config-backup bucket. S3 creds
# come from the cephrgw BucketAccess Secret (cheeztv-config-backup-s3); the
# restic repo password comes from Vault via the cheeztv-k8up-restic Secret.
#
# s3.ceph.unkin.net presents the internal unkin.net CA, which the k8up/restic
# image does not trust by default, so the reflected vault-ca-cert Secret is
# mounted into every job pod and pointed at via backend.tlsOptions.caCert.
apiVersion: k8up.io/v1
kind: Schedule
metadata:
name: cheeztv-config
namespace: cheeztv
spec:
backend:
repoPasswordSecretRef:
name: cheeztv-k8up-restic
key: password
s3:
endpoint: https://s3.ceph.unkin.net
bucket: cheeztv-config-backup
accessKeyIDSecretRef:
name: cheeztv-config-backup-s3
key: AWS_ACCESS_KEY_ID
secretAccessKeySecretRef:
name: cheeztv-config-backup-s3
key: AWS_SECRET_ACCESS_KEY
tlsOptions:
caCert: /etc/k8up/ca/ca.crt
volumeMounts:
- name: vault-ca
mountPath: /etc/k8up/ca
readOnly: true
backup:
schedule: "0 2 * * *"
failedJobsHistoryLimit: 3
successfulJobsHistoryLimit: 3
volumes:
- name: vault-ca
secret:
secretName: vault-ca-cert
prune:
schedule: "0 3 * * 0"
retention:
keepDaily: 14
keepWeekly: 8
keepMonthly: 12
volumes:
- name: vault-ca
secret:
secretName: vault-ca-cert
check:
schedule: "0 4 * * 0"
volumes:
- name: vault-ca
secret:
secretName: vault-ca-cert
+18
View File
@@ -0,0 +1,18 @@
---
apiVersion: v1
kind: Service
metadata:
name: cheeztv
namespace: cheeztv
spec:
internalTrafficPolicy: Cluster
ports:
- name: http
port: 8096
protocol: TCP
targetPort: http
selector:
app: cheeztv
# Pin each client to one replica to reduce transcode-session churn/takeover.
sessionAffinity: ClientIP
type: ClusterIP
+253
View File
@@ -0,0 +1,253 @@
---
apiVersion: apps/v1
kind: StatefulSet
metadata:
name: cheeztv
namespace: cheeztv
spec:
# HA: two replicas coordinate transcode session ownership through Valkey and
# resume each other's HLS segments off the shared RWX transcode PVC. Stable
# pod names (cheeztv-0/1) are the lease owner identity, hence StatefulSet.
replicas: 2
serviceName: cheeztv
podManagementPolicy: Parallel
updateStrategy:
type: RollingUpdate
selector:
matchLabels:
app: cheeztv
template:
metadata:
labels:
app: cheeztv
spec:
securityContext:
# Group-write the shared RWX volumes and grant the render/video groups so
# the runAsUser 1000 process can open the Intel DRI render node injected
# by the device plugin.
fsGroup: 1000
supplementalGroups:
- 44
- 105
- 109
seccompProfile:
type: RuntimeDefault
affinity:
# Spread the two replicas across nodes for node-level HA. Soft so a
# single-GPU-node cluster still schedules both (i915 has 4 shared slots).
podAntiAffinity:
preferredDuringSchedulingIgnoredDuringExecution:
- weight: 100
podAffinityTerm:
labelSelector:
matchLabels:
app: cheeztv
topologyKey: kubernetes.io/hostname
initContainers:
# Seed the fork's PostgreSQL provider (database.xml) and Intel iGPU
# hardware transcode settings (encoding.xml) before Jellyfin starts.
# Runs as root to chown into the shared config volume; mirrors the fork
# Helm chart's inject-db-config. Each file is written only when absent so
# admin changes persisted to the shared RWX /config survive pod restarts.
- name: inject-config
image: busybox:1.37.0
command:
- sh
- -c
- |
mkdir -p /config/config
chown 1000:1000 /config/config
chmod 775 /config/config
if [ ! -e /config/config/database.xml ]; then
cat > /config/config/database.xml << 'DBEOF'
<?xml version="1.0" encoding="utf-8"?>
<DatabaseConfigurationOptions>
<DatabaseType>Jellyfin-PostgreSQL</DatabaseType>
<LockingBehavior>NoLock</LockingBehavior>
</DatabaseConfigurationOptions>
DBEOF
chown 1000:1000 /config/config/database.xml
chmod 664 /config/config/database.xml
fi
# VAAPI on the Intel render node the device plugin injects
# (/dev/dri/renderD128 — ffmpeg's default DRM node, reachable via
# the render/video supplementalGroups). Without this the attached
# iGPU is idle and every transcode runs in software. Omitted
# elements fall back to the fork's EncodingOptions defaults.
if [ ! -e /config/config/encoding.xml ]; then
cat > /config/config/encoding.xml << 'ENCEOF'
<?xml version="1.0" encoding="utf-8"?>
<EncodingOptions xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xsd="http://www.w3.org/2001/XMLSchema">
<EncodingThreadCount>-1</EncodingThreadCount>
<HardwareAccelerationType>vaapi</HardwareAccelerationType>
<VaapiDevice>/dev/dri/renderD128</VaapiDevice>
<EnableHardwareEncoding>true</EnableHardwareEncoding>
<AllowHevcEncoding>true</AllowHevcEncoding>
<AllowAv1Encoding>false</AllowAv1Encoding>
<EnableIntelLowPowerH264HwEncoder>false</EnableIntelLowPowerH264HwEncoder>
<EnableIntelLowPowerHevcHwEncoder>false</EnableIntelLowPowerHevcHwEncoder>
<EnableTonemapping>false</EnableTonemapping>
<EnableVppTonemapping>false</EnableVppTonemapping>
<HardwareDecodingCodecs>
<string>h264</string>
<string>hevc</string>
<string>vc1</string>
<string>vp9</string>
</HardwareDecodingCodecs>
</EncodingOptions>
ENCEOF
chown 1000:1000 /config/config/encoding.xml
chmod 664 /config/config/encoding.xml
fi
resources:
requests:
cpu: 10m
memory: 32Mi
limits:
cpu: 100m
memory: 64Mi
volumeMounts:
- name: config
mountPath: /config
containers:
- name: cheeztv
image: artifactapi.k8s.syd1.au.unkin.net/docker-internal/jellyfin-ha:v0.1.3
imagePullPolicy: IfNotPresent
ports:
- name: http
containerPort: 8096
protocol: TCP
env:
# Pod identity for the Valkey transcode lease owner. The fork reads
# JELLYFIN_INSTANCE_ID (falling back to MachineName); the stable
# StatefulSet pod name gives each replica a unique lease identity so
# takeover can target a dead replica. JELLYFIN_HA_POD_NAME is set for
# parity with the fork Helm chart (nothing currently reads it).
- name: JELLYFIN_INSTANCE_ID
valueFrom:
fieldRef:
fieldPath: metadata.name
- name: JELLYFIN_HA_POD_NAME
valueFrom:
fieldRef:
fieldPath: metadata.name
# Multiple replicas must not each answer UDP auto-discovery.
- name: JELLYFIN_Network__AutoDiscovery
value: "false"
# Config dir must differ from the data root (Jellyfin sanity check).
- name: JELLYFIN_CONFIG_DIR
value: /config/config
# Distributed transcode session store (cheeztv-ha additions).
- name: Jellyfin__TranscodeStore__RedisConnectionString
value: "valkey-cheeztv-valkey:6379,abortConnect=false"
- name: Jellyfin__TranscodeStore__LeaseDurationSeconds
value: "30"
# PostgreSQL main DB via the CNPG-generated app secret, routed through
# the PgBouncer pooler. Composed with $(VAR) expansion so the password
# is never rendered into the manifest; CNPG passwords are URL-safe.
- name: PGUSER
valueFrom:
secretKeyRef:
name: cheeztv-postgres-app
key: username
- name: PGPASSWORD
valueFrom:
secretKeyRef:
name: cheeztv-postgres-app
key: password
- name: PGDB
valueFrom:
secretKeyRef:
name: cheeztv-postgres-app
key: dbname
- name: POSTGRES_CONNECTION_STRING
value: "postgresql://$(PGUSER):$(PGPASSWORD)@cheeztv-postgres-pooler:5432/$(PGDB)"
- name: DATABASE_URL
value: "postgresql://$(PGUSER):$(PGPASSWORD)@cheeztv-postgres-pooler:5432/$(PGDB)"
livenessProbe:
httpGet:
path: /health
port: http
initialDelaySeconds: 30
periodSeconds: 30
timeoutSeconds: 5
failureThreshold: 3
readinessProbe:
httpGet:
path: /health
port: http
initialDelaySeconds: 10
periodSeconds: 10
timeoutSeconds: 5
failureThreshold: 3
resources:
requests:
cpu: "1"
memory: 1Gi
gpu.intel.com/i915: "1"
limits:
cpu: "4"
memory: 6Gi
# Intel iGPU (QSV/VA-API) slot. Requesting it pins the pod to a
# GPU-labelled node and injects /dev/dri/renderD* automatically, so
# no /dev/dri hostPath or privileged container is needed. VA-API is
# pre-enabled via the seeded encoding.xml (see inject-config), so
# transcodes use the iGPU on first boot with no manual UI step.
gpu.intel.com/i915: "1"
securityContext:
runAsUser: 1000
runAsGroup: 1000
volumeMounts:
- name: config
mountPath: /config
- name: transcode
# Fork's real transcode temp path. RWX so a surviving pod reads the
# in-flight .ts/.m3u8 segments of the pod it takes over. A per-pod
# volume here silently breaks HA takeover.
mountPath: /config/transcodes
- name: cache
mountPath: /cache
- name: media-tv
# Kids-only instance: mount just the tvshows/kids subtree of the
# shared TV subvolume (subPath kids) so cheeztv's library never
# exposes the adult tree. Path matches fafflix's /media/tv/kids
# mount so the same episode resolves identically across instances.
mountPath: /media/tv
subPath: kids
readOnly: true
- name: media-movies
# Kids-only instance: mount just the movies/kids subtree of the
# shared movies subvolume (subPath kids).
mountPath: /media/movies
subPath: kids
readOnly: true
volumes:
- name: config
persistentVolumeClaim:
claimName: cheeztv-config
- name: transcode
persistentVolumeClaim:
claimName: cheeztv-transcode
- name: media-tv
persistentVolumeClaim:
claimName: cheeztv-media-tv
- name: media-movies
persistentVolumeClaim:
claimName: cheeztv-media-movies
volumeClaimTemplates:
# Per-pod scratch cache — RWO, disposable, one PVC per replica.
- metadata:
name: cache
annotations:
# Exclude the per-pod cache PVCs from the cheeztv-config k8up Schedule
# (skipWithoutAnnotation is false cluster-wide). Cache is disposable and
# RWO — it would also fail to mount into the backup pod while in use.
k8up.io/backup: "false"
spec:
accessModes:
- ReadWriteOnce
resources:
requests:
storage: 30Gi
storageClassName: cephrbd-fast-delete
volumeMode: Filesystem
+49
View File
@@ -0,0 +1,49 @@
---
# HA Valkey managed by valkey-operator as Jellyfin's distributed transcode-session
# store. shards:1 + replicas:2 is one primary with two replicas in a single shard
# group (three ValkeyNodes total): losing the primary triggers an automatic
# failover to a replica, so a node/pod loss no longer drops the shared transcode
# lease state the two Jellyfin replicas coordinate through. The operator always
# runs Valkey cluster-mode-enabled with protected-mode off and leaves the built-in
# `default` user passwordless, so clients connect with no auth/TLS (parity with the
# old open redis); StackExchange.Redis seeds off the single headless service and
# auto-discovers the topology plus failovers, so the connection string is
# unchanged. scheduling.node.spread.shard:Required renders pod anti-affinity that
# keeps the three nodes on distinct hosts (kubernetes.io/hostname), so one host
# loss removes at most one node. podDisruptionBudget.mode:Cluster lets the operator
# manage a quorum-aware PDB. Persistence stays omitted so /data is an emptyDir:
# lease state is ephemeral (30s TTLs), replication+failover already provide
# redundancy and a replaced node re-syncs from the primary, and an operator-managed
# PVC cannot carry the k8up.io/backup:"false" annotation the namespace's
# cluster-wide k8up Schedule needs to skip in-use RWO volumes.
apiVersion: valkey.io/v1alpha1
kind: ValkeyCluster
metadata:
name: cheeztv-valkey
namespace: cheeztv
spec:
shards: 1
replicas: 2
image: artifactapi.k8s.syd1.au.unkin.net/dockerhub/valkey/valkey:9.0.0
# redis_exporter sidecar (:9121, port name `metrics`) on every ValkeyNode pod;
# the operator manages a dedicated _exporter ACL user for it. Image overridden
# from the operator default (bare dockerhub oliver006/redis_exporter:v1.80.0)
# to the artifactapi-proxied pin. Scraped by the valkey-exporter VMPodScrape
# in vmpodscrape.yaml alongside this file.
exporter:
enabled: true
image: artifactapi.k8s.syd1.au.unkin.net/dockerhub/oliver006/redis_exporter:v1.89.0
scheduling:
node:
spread:
shard:
mode: Required
podDisruptionBudget:
mode: Cluster
resources:
requests:
cpu: 50m
memory: 128Mi
limits:
cpu: 500m
memory: 512Mi
+20
View File
@@ -0,0 +1,20 @@
---
apiVersion: secrets.hashicorp.com/v1beta1
kind: VaultAuth
metadata:
name: default
namespace: cheeztv
annotations:
argocd.argoproj.io/sync-wave: "0"
spec:
allowedNamespaces:
- cheeztv
kubernetes:
audiences:
- vault
role: default
serviceAccount: default
tokenExpirationSeconds: 600
method: kubernetes
mount: k8s/au/syd1
vaultConnectionRef: vso-system/default
+24
View File
@@ -0,0 +1,24 @@
---
# restic repository password for the k8up cheeztv-config backups. Seeded at
# kv/kubernetes/namespace/cheeztv/default/k8up-restic (key: password); the
# default k8s role's templated policy already grants read here, so no
# terraform-vault change is needed. VSO syncs it into the cheeztv-k8up-restic
# Secret that the Schedule references via backend.repoPasswordSecretRef.
apiVersion: secrets.hashicorp.com/v1beta1
kind: VaultStaticSecret
metadata:
name: cheeztv-k8up-restic
namespace: cheeztv
annotations:
argocd.argoproj.io/sync-wave: "0"
spec:
destination:
create: true
name: cheeztv-k8up-restic
overwrite: true
hmacSecretData: true
mount: kv
path: kubernetes/namespace/cheeztv/default/k8up-restic
refreshAfter: 5m
type: kv-v2
vaultAuthRef: default
+19
View File
@@ -0,0 +1,19 @@
---
# Scrape the operator-injected redis_exporter sidecar (:9121, port name
# `metrics`) on the cheeztv-valkey ValkeyNode pods. The valkey-operator gives
# its pods fixed labels only (no pod-label passthrough on the ValkeyCluster CR),
# so select on the operator-managed labels. Picked up by the observability
# VMAgent (selectAllByDefault).
apiVersion: operator.victoriametrics.com/v1beta1
kind: VMPodScrape
metadata:
name: valkey-exporter
namespace: cheeztv
spec:
selector:
matchLabels:
app.kubernetes.io/managed-by: valkey-operator
app.kubernetes.io/component: valkey-node
podMetricsEndpoints:
- port: metrics
path: /metrics
+12
View File
@@ -213,6 +213,18 @@ spec:
- name: media-movies
mountPath: /media/movies
readOnly: true
# Kids subtrees surfaced as their own paths (same media PVCs, subPath
# kids) so a "Kids TV"/"Kids Movies" library can be added here and its
# titles are browsable and resume in this instance's own DB. The full
# /media/{tv,movies} mounts above are unchanged.
- name: media-tv
mountPath: /media/tv-kids
subPath: kids
readOnly: true
- name: media-movies
mountPath: /media/movies-kids
subPath: kids
readOnly: true
volumes:
- name: config
persistentVolumeClaim:
@@ -0,0 +1,6 @@
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
resources:
- ../../../base/cheeztv
+1
View File
@@ -11,6 +11,7 @@ spec:
revision: HEAD
directories:
- path: apps/overlays/*/jellyfin
- path: apps/overlays/*/cheeztv
- path: apps/overlays/*/arrstack
template:
metadata:
+2
View File
@@ -11,6 +11,8 @@ spec:
destinations:
- namespace: 'jellyfin'
server: https://kubernetes.default.svc
- namespace: 'cheeztv'
server: https://kubernetes.default.svc
- namespace: 'arrstack'
server: https://kubernetes.default.svc
clusterResourceWhitelist: