Guidance Required for DSpace 9.1 Horizontal Scaling and Session Management

[Bharat teja Pasupuleti]

Dear DSpace Community Team,

We are currently deploying DSpace 9.1 in a horizontally scaled environment using Kubernetes on GCP.

Our current architecture includes:

* Multiple DSpace 9.1 backend application pods

* Angular/UI frontend deployment

* Kubernetes ingress/load balancer

* Shared database

* Shared file/object storage

While scaling the application horizontally, we are facing login/session-related issues. Since user sessions appear to be maintained in-memory within individual pods, requests routed to different pods are causing session inconsistencies, unexpected logouts, and authentication issues.

We also observed that frontend requests routed through the ingress/load balancer may hit different backend pods during authentication and subsequent API calls, which appears to impact session continuity.

We would like to understand the recommended production-ready approach for session management in DSpace 9.1 when running multiple pods.

We are currently evaluating the following options:

* Sticky Sessions at ingress/load balancer level

* Redis-based distributed sessions

* Database-backed sessions

* JWT/token-based authentication

* SSO/OAuth integration

Could you please provide guidance on:

1. Recommended architecture for horizontal scaling in DSpace 9.1

2. Best practices for frontend-to-backend session handling in Kubernetes environments

3. Best practices for session management across pods

4. Whether Redis/session replication is officially recommended

5. Any known limitations or reference implementations for Kubernetes deployments

6. Suggestions for production-grade HA deployments

Any documentation, community references, or implementation examples would be greatly appreciated.

Thank you for your support and guidance.