ODIN.fun Ships Major Indexer and Staking Updates in Week 32

ODIN.fun has reported 187 commits and 45 pull requests in its Week 32 development update, with a series of infrastructure, staking and user-facing changes moving into production.

Several weeks of work on indexer monitoring were promoted to production during the latest development cycle. Health endpoints, a watchdog system, telemetry, alerting and read-replica routing are now live, giving both indexers full production observability.

The team also introduced a Kafka-based change data capture pipeline after building it from the ground up in two days. The pipeline includes health probes, latency tracking, structured logging and PagerDuty routing, alongside an 11-part operations runbook covering its management and response procedures.

Changes to the staking canister focused on tightening safeguards around lock operations. BTC is now rejected across every lock path rather than relying on checks at individual call sites. The admin force-release endpoint has also been changed to use both the user and token ID as identifiers, linking the action to a specific position.

ODIN.fun has also replaced its blacklist mirror for staking with local management, reducing reliance on the previous approach.

The staking data model has been rebuilt around an explicit contract for event indexing, including changes to deduplication and roll-up logic. The update is designed to prevent incorrect counts when new operation types are introduced and to ensure re-emitted events are handled rather than silently discarded.

Some of the latest changes are visible directly to users. Staking durations are now displayed in minutes, while the Locked tab shows lock progress rather than a ticker. The unstake confirmation window also displays the staked amount and lock duration before a user completes the action.

The Week 32 update reflects ODIN.fun’s ongoing practice of publishing development work as it moves from internal branches into production, covering both backend infrastructure and changes users can see in the platform.