Sometimes a
Claymore Ethereum mining rig can become unstable or unresponsive, particularly when optimizing GPU performance by overclocking and undervolting. If the inactivity goes unnoticed, or if a person can’t get immediate physical access to cycle the power, the resulting idleness wastes potential mining capacity.
I wrote some software called
watchclay that watches Claymore via its remote management port. It watches for several kinds of issue:
- Claymore and/or the rig become unresponsive, and return no data.
- The hash rate falls below some expected performance.
- The mining pool rejects submitted shares.
- The temperature of one or more GPUs exceeds some maximum amount.
For most of these, if the issues persists, watchclay power cycles the designated outlet on a
Ubiquiti mPower strip https://www.ubnt.com/mfi/mpower/ to reset the rig and return Claymore to mining. In case of GPU overheating, watchclay immediately powers down the designated outlet to prevent permanent damage.
Email updates are sent whenever an issue persists and the rig is power cycled, and again when the rig returns to normal. An email update is also sent periodically to indicate normal operation.
A high-level summary of Claymore rig health can be monitored using tail -f on a logfile of watchclay output. The summary includes overall hashrate, slowest GPU, shares accepted and rejected by the mining pool, the temperature of the hottest GPU, and total amps drawn.
I’m publishing this Python software on github in the hope that other miners might find it useful.
https://github.com/llang629/watchclayFeedback (and tips) welcome. Happy mining!
