Strimzi Validation Results¶
Test Platform¶
- Platform: OpenShift 4.20
- Strimzi Version: OLM-managed (AMQ Streams)
- Test Date: 2026-05
Results¶
| Experiment | Component | Injection | Verdict | Notes |
|---|---|---|---|---|
| cluster-operator/pod-kill | cluster-operator | PodKill | Resilient | |
| cluster-operator/network-partition | cluster-operator | NetworkPartition | Resilient | |
| cluster-operator/label-stomping | cluster-operator | LabelStomping | Resilient | |
| cluster-operator/quota-exhaustion | cluster-operator | QuotaExhaustion | Resilient | |
| cluster-operator/rbac-revoke | cluster-operator | RBACRevoke | Resilient | |
| cluster-operator/deployment-scale-zero | cluster-operator | DeploymentScaleZero | Degraded | OLM does not restore replicas |
| cluster-operator/leader-election-disrupt | cluster-operator | LeaderElectionDisrupt | Resilient | |
| cluster-operator/config-drift | cluster-operator | ConfigDrift | Resilient |
Key Findings¶
DeploymentScaleZero: OLM Gap¶
The only non-Resilient result. When the cluster-operator Deployment is scaled to zero replicas, OLM does not automatically restore the replica count. This is a known OLM limitation affecting all OLM-managed operators. Manual intervention (or an external controller) is needed to restore replicas.
Strong Core Resilience¶
All other failure modes (PodKill, NetworkPartition, LabelStomping, QuotaExhaustion, RBACRevoke, LeaderElectionDisrupt, ConfigDrift) recover automatically. The Strimzi operator handles disruptions well across the board.