Miner options

tenrai-miner --help prints the same list. Every option has a default except your address: tenrai-miner -a <address> mines solo on this machine's node with every GPU and two CPU threads, and shows its dashboard on http://127.0.0.1:28770/.

Connection

While the node catches up

The node hands out work only while it is synchronized with its peers; a dip of up to 30 seconds, such as peers reconnecting, keeps the work coming. While the node catches up or cannot be reached, the miner keeps its job for a few seconds and then pauses. It says so once, reminds you once a minute with the node's height and starts again by itself. Stratum bridges and pools do the same for their miners.

For a private development chain or a lone test node, set mine_when_not_synced: true in node.json or start the node with tenraid run --mine-when-not-synced. That is a node setting, not a miner option.

Devices

GPU backends

On NVIDIA cards auto runs CUDA when NVRTC is installed. Without it auto runs OpenCL 1.2 there, because the drivers report OpenCL 3.0 without the needed features. CUDA needs NVIDIA driver R550 or newer and NVRTC 12 (nvrtc64_120_0.dll on Windows, libnvrtc.so.12 on Linux) from the CUDA toolkit or next to the miner.

Every GPU is tested against the CPU at the start of each epoch, and every block or share it finds is hashed again on the CPU before it is sent. A card that fails, or returns three wrong hashes in a row, moves to its next backend. With no backend left it retries after 30 seconds, then at longer intervals up to 5 minutes.

TenQ

In full mode the miner falls back to light while there is no room for the dataset. The next epoch's dataset is built 1,024 cuts ahead when 2.5 GiB are free.

Statistics and logs

Keep --api-bind on 127.0.0.1 unless a firewall protects the port; the dashboard and the statistics have no password.

Device states in /v1/stats: mining, waiting (no work), paused (temperature limit), failed (retrying), stalled (no hashes for 90 seconds).

Other

Config file

Without --config the miner reads tenrai-miner.json next to the program and creates it with the default settings on the first start. The file uses the long option names. true turns a switch on. A value on the command line replaces the file's; lists from both are added together. With your address in the file, tenrai-miner starts without arguments.

{
  "mining-address": "trai1...",
  "stratum": ["stratum+tls://pool-a.example.org:28741", "stratum2+tls://pool-b.example.org:28742"],
  "worker": "rig1",
  "pool": true,
  "backend": "cuda",
  "temp-limit": 80,
  "threads": "all"
}

HiveOS

Build the package on Linux and put the archive where the rig can download it:

cargo build --release -p tenrai-miner
bins/miner/hiveos/package.sh

In the flight sheet, choose Custom as miner:

HiveOS reads hashrate, temperatures and shares from /v1/stats.