Rejected shares: the troubleshooting ladder

Your miner's log distinguishes reject causes: stale (job changed before your share arrived — latency), invalid (bad hash or duplicate — usually overclocks or a dying board), low difficulty (submitted below the assigned target — vardiff racing). The mix tells you which ladder to climb.

• 1. Latency — ping the pool; over ~150 ms consider the nearest server region or another pool. Stales dominate? Start here.
• 2. Connection quality — swap Wi-Fi for ethernet; check packet loss, not just ping. One lost job update = a stale burst.
• 3. Thermals — check temps against your baseline; hot boards make math errors (invalid shares) before they crash.
• 4. Clocks — back off memory/core overclocks to stock, verify, then re-tune one step at a time.
• 5. Cables & PSU — under-powered rigs miscompute under load; a marginal cable produces 'random' invalids.
• 6. Pool side — compare reject rate across two pools for an hour; if only one rejects, the problem is theirs — check their status page or move.

Pool cards list each pool's location/endpoints — the latency rung of the ladder — and the stale-shares lesson covers the underlying mechanics in depth.