v0.7.26.15Current · stable
- Block-template integrity: before building work for miners, the pool now independently double-checks the difficulty target it's about to hand out against the network's own. Before, a mismatch here would only surface after the fact, once a finished block was rejected. Now the pool catches it up front and refuses that piece of work outright rather than build on it, automatically serving the last known-good work while it retries — a miner never sees a gap. No effect on mining, share accounting, or how a won block is found and kept.
- Accounting durability: the pool's internal record-keeping — shares, blocks, and other accounting — now survives a sustained storage hiccup without losing anything. Writes that can't land immediately are queued to a recovery file and replayed automatically the moment storage recovers, with nothing ever counted twice. No effect on mining or payouts; this is entirely about the pool's own bookkeeping staying accurate under fault conditions that are otherwise invisible day to day.
- Block-broadcast robustness: when a found block is broadcast to multiple nodes at once, one slow or unresponsive node can no longer hold back the others from hearing about it quickly — each node gets its own bounded window to respond, so a healthy node's acceptance is never delayed by a stuck one. No effect on mining, share accounting, or payouts.
- Reconnection efficiency: a brief node-connection blip no longer causes the pool to redo work it already had in hand — a repeated notification from a reconnecting node is now recognized and skipped instead of triggering a redundant rebuild. No effect on mining, share accounting, or payouts.
- Latency telemetry: the pool now measures, in microseconds, how long it takes from seeing a new block to handing a miner fresh work — tracked separately for classic (V1) and modern (V2) connections. Diagnostic only; no effect on mining, share accounting, or payouts.