Add startupProbe to give Jellyfin sufficient time to start, especially
with large media libraries or slow storage. This prevents Kubernetes
from killing the pod during initial startup phase.
Changes:
- Add startupProbe configuration in values.yaml with sensible defaults
- failureThreshold: 30 (allows up to 5 minutes for startup)
- periodSeconds: 10
- Uses tcpSocket check on http port
- Update deployment.yaml to include startupProbe from values
- Bump chart version 2.5.0 -> 2.5.1
- Add changelog annotation for Artifact Hub
Benefits:
- Prevents pod restarts during slow initial startup (large media libraries)
- Liveness and readiness probes remain responsive after successful startup
- Fully configurable and can be disabled by setting startupProbe: null
- No breaking changes, backward compatible
The startup probe gives Jellyfin up to 5 minutes (30 attempts * 10 seconds)
to complete initialization. After the first success, liveness and readiness
probes take over with their normal timing.
Co-Authored-By: Claude <noreply@anthropic.com>
Signed-off-by: Aleksei Sviridkin <f@lex.la>
Add comprehensive troubleshooting documentation for inotify instance limit
errors that cause Jellyfin crashes in environments with many containers.
Changes:
- Add Troubleshooting section to README with inotify limits explanation
- Document proper solution: increase kernel limits via sysctl
- Provide workaround for managed clusters: DOTNET_USE_POLLING_FILE_WATCHER
- Add commented example in values.yaml for easy copy-paste
- Update Chart version from 2.5.0 to 2.5.1
- Add Artifact Hub changelog annotation
The proper solution is to increase inotify limits on Kubernetes nodes:
```bash
sysctl -w fs.inotify.max_user_instances=512
```
For managed Kubernetes where node access is restricted, users can use the
polling workaround (less efficient but functional):
```yaml
jellyfin:
env:
- name: DOTNET_USE_POLLING_FILE_WATCHER
value: "1"
```
This addresses the crash issue mentioned in #64.
Fixes: #64
Co-Authored-By: Claude <noreply@anthropic.com>
Signed-off-by: Aleksei Sviridkin <f@lex.la>
Add dedicated cache volume configuration to reduce wear on SD cards and improve
performance by allowing cache to be stored on separate storage.
Changes:
- Add persistence.cache configuration in values.yaml with support for:
- PVC (default type when enabled)
- hostPath for direct host mount
- emptyDir (default when disabled)
- Add cache volume mount at /cache in deployment
- Add cache PersistentVolumeClaim template
- Support for existing PVC via existingClaim
- Configurable size (default 10Gi), accessMode, storageClass, and annotations
- Update Chart version from 2.5.0 to 2.5.1
- Add Artifact Hub changelog annotation
The cache volume is disabled by default to maintain backward compatibility.
Users can enable it via:
```yaml
persistence:
cache:
enabled: true
type: pvc
size: 10Gi
```
This addresses the use case mentioned in #62 where users want to avoid SD card
wear by storing cache on separate persistent storage.
Fixes: #62
Co-Authored-By: Claude <noreply@anthropic.com>
Signed-off-by: Aleksei Sviridkin <f@lex.la>
Change default health probes from tcpSocket to httpGet to fix issue #68
where pods fail to start in IPv6-only clusters due to incompatible health checks.
Changes:
- Change default livenessProbe and readinessProbe to use httpGet with /health endpoint
- Add comprehensive IPv6 and dual-stack configuration documentation
- Document ipFamilyPolicy options (SingleStack, PreferDualStack, RequireDualStack)
- Document ipFamilies configuration for IPv4-only, IPv6-only, and dual-stack setups
- Add IPv6 Configuration section to README with examples
- Update Chart version from 2.5.0 to 2.5.1
The httpGet probe works with both IPv4 and IPv6 out of the box, while tcpSocket
may fail in IPv6-only environments. Jellyfin's /health endpoint is specifically
designed for health checks and verifies both HTTP and database connectivity.
Fixes: #68
Co-Authored-By: Claude <noreply@anthropic.com>
Signed-off-by: Aleksei Sviridkin <f@lex.la>
Add support for Kubernetes Gateway API HTTPRoute resource as a modern
alternative to Ingress. The implementation follows the same pattern as
the existing Ingress template.
Changes:
- Add HTTPRoute template with configurable parentRefs, hostnames, and rules
- Add httpRoute section to values.yaml with generic examples
- Update Chart version from 2.5.0 to 2.6.0
- Add Artifact Hub changelog annotation
- Add HTTPRoute documentation to README
Co-Authored-By: Claude <noreply@anthropic.com>
Signed-off-by: Aleksei Sviridkin <f@lex.la>