[martinsander00]

Summary of Changes

• This PR makes the worker retry a failed merkle root posting 3 more times, one minute apart, before it reports the failure.
• The scheduler and calculate-rewards both post the root through the same call, so both now retry.
• Before, one transient RPC error ended the run. The worker no longer fails the run at the first error.

Testing Verification

• The backfill of DZ epoch 241 failed once with RPC response error -32004: Block not available for slot 462537200. This retry covers that error.
• No test was added, because the change wraps one existing call in the retry helper that the crate already uses.

[Copilot]

🟢 Approval recommended

No active review issues block the merkle root posting retry change.

No open issues 🎉

What changed in this PR

This PR adds retries for merkle root posting in the contributor rewards worker, reducing failures from transient RPC errors.

Changes:

• The worker retries a failed posting three more times, with a one-minute delay and a warning before each retry.
• The scheduler and calculate-rewards both use the updated posting path.
• The changelog records the retry behavior.

File	Description
offchain/​crates/​contributor-rewards/​src/​calculator/​orchestrator.rs	Adds bounded retries and warning logs around merkle root posting.
offchain/​crates/​contributor-rewards/​CHANGELOG.md	Documents the three additional posting attempts and one-minute delays.

_{Review effort: Balanced}

[nikw9944]

The retry covers the Block not available for slot error from epoch 241. That error comes from the get_block_time call inside wait_for_grace_period, and the new retry wraps that call. One non-blocking note: the retry also covers the grace-period wait, so a timeout can now stretch to about four times grace_period_max_wait_seconds.

[nikw9944]

The earlier note about retries stretching the grace-period wait is fixed. orchestrator.rs now computes grace_period_deadline once, before the retry loop, and wait_for_grace_period checks against that deadline. A retry after a timeout now fails within one poll, and grace_period_max_wait_seconds again limits the total wait. The retry still covers the get_block_time error from epoch 241.