This is a great idea and would significantly improve the resilience of self hosted Ocean integrations, especially for long-running sync jobs.

Today, Ocean runs as a single replica using a recreate strategy. For integrations such as ServiceNow, where a sync can take hours, a routine node restart or upgrade can interrupt the job and force it to restart from scratch. During this time, data in Port can appear stale or unresponsive.

Because only one replica can run, it is not currently possible to use PodDisruptionBudgets, topology spread or anti-affinity rules, or leader election to protect in-flight syncs. There is also no shared or persisted state that would allow a sync to safely resume after a restart, whether via Kubernetes primitives such as CRDs or ConfigMaps, or an external data store.

Adding HA support would reduce unnecessary resyncs, improve reliability for larger integrations, and better align Ocean with how Kubernetes workloads are typically run in production.

Omri Gez Mark Coxcan you please elaborate on this request?