Antonio Krsoski, Emanuel Sjövall:
Managing failures in stateful systems,
summary, report, June 2026.

Abstract:

Failures in stateful systems can cause persistent inconsistencies and are difficult to recover from. This is especially relevant in Kubernetes-based environments, where Operators automate the management of stateful applications but also introduce additional complexity and failure modes.

This thesis investigates failures in Operator-managed stateful systems within Neo4j’s cloud architecture. The initiating problem is that failures still occur despite existing testing and rollout strategies, and recovery often requires manual, non-standardized intervention. To address this, we examine two research questions: what root causes lead to failures in managed stateful systems, and how failures caused by Kubernetes Operators can be mitigated.

We identified 74 root causes and grouped them into 36 responsibility domains. Failures often stem from incorrect state evaluation and error-prone deployments. We developed Reconcile Dump to improve debugging and identified key operational pain points.

These findings suggest that improving reliability requires incremental improvements in observability, automation, and deployment design.