proxy.conf.example

listen_addr = 0.0.0.0
listen_port = 3334

# Maximum concurrent stratum connections, across every listener. A connection
# past this is closed immediately at accept. Each one costs a thread and an fd.
max_conns = 500

bitcoind_url = http://127.0.0.1:18443
# bitcoind_user / bitcoind_pass are optional. Omit BOTH for a block-template
# backend that doesn't require authentication; the RPC call is then made
# without a basic-auth header. Set them for a stock bitcoind using rpcuser/
# rpcpassword (cookie auth is not supported).
bitcoind_user = drivepool
bitcoind_pass = drivepool
# How often to re-fetch the block template. On a drivechain pool this is also
# the worst-case delay before a sidechain's BMM request reaches a stratum job:
# a sidechain publishes its BMM transaction only after it sees the new tip, so
# any block found within one poll of that cannot carry the commitment. Lower it
# (5000-10000) if sidechains need to merge-mine reliably; the cost is one extra
# getblocktemplate call per interval.
bitcoind_poll_interval_ms = 30000

# Miners are paid directly: the stratum username must be the miner's
# bitcoin address (`<address>` or `<address>.<rig_label>`). The configured
# operator_address below receives a fee_bps cut of every block reward.
operator_address = bc1qREPLACEME   # REPLACE with a real address for your network (bc1 mainnet / tb1 testnet / bcrt1 regtest)
fee_bps = 100              # 100 = 1%; valid range 0..1000 (max 10%)
coinbase_tag = /simplepool/

# vardiff — auto-adjust each connection's difficulty to keep the share
# rate near `target_spm` shares/minute. Set vardiff_enabled = 0 to pin
# every connection to initial_diff (legacy behaviour).
#
# These are the settings for listen_port, which is the port home miners are
# pointed at. Rented hashrate does not belong here — give it its own listener
# below, with its own floor.
initial_diff       = 1
vardiff_enabled    = 1
vardiff_target_spm = 12       # ~1 share every 5 seconds per connection
vardiff_min        = 1
vardiff_max        = 1e12
vardiff_window_sec = 30       # how often to retarget

# Extra stratum ports, each with its own difficulty policy. Repeatable —
# unlike every other key here, a second `listener` line adds a port rather
# than replacing the first. listen_port above is always served too, on the
# settings above.
#
#   port           the port to bind (required)
#   min_diff       vardiff floor for this port, and its starting difficulty
#   initial_diff   starting difficulty, if it should differ from min_diff
#   max_diff       vardiff ceiling for this port
#   label          shown in the log and on the dashboard; [A-Za-z0-9_-]
#
# Why a second port at all: one difficulty cannot serve both a home ASIC and
# rented hashrate. Rented hashrate does not arrive as many small miners — a
# marketplace aggregates a whole fleet behind one connection, and 1 PH/s at
# difficulty 1024 is ~227 shares per SECOND down that single socket. The
# marketplaces know this and refuse to deliver to a pool below their floor:
# Braiins wants at least 1024 and recommends 65536, NiceHash requires 500000.
#
# Vardiff cannot bridge the gap. It moves by at most 4x per window, so at the
# 30s default it takes eight windows — four minutes — to climb from 1 to
# 65536, and the reject flood on the way there is what gets an order
# cancelled. The miner has to arrive at the right difficulty, which is what a
# dedicated port is for.
#
# listener = port=3335 min_diff=65536 label=braiins
# listener = port=3336 min_diff=500000 label=nicehash
#
# IMPORTANT — min_diff is KEPT even when the chain is easier than it, and
# that costs blocks. Share difficulty is otherwise capped at the network
# difficulty, because a miner filters locally against the stratum target and
# a harder share target makes it discard valid blocks before the pool ever
# sees them. min_diff overrides that cap, on purpose: a marketplace measures
# the difficulty on the wire, so a 500000 port that quietly served 1200 would
# have its order cancelled and the pool would never learn why.
#
# The trade is therefore explicit and confined to the ports that ask for it.
# On a 500000 port over a chain at 1200, miners discard roughly 416 of every
# 417 blocks they solve. simplepool warns at startup for every listener in
# that position, and the dashboard reports it too (the "Stratum ports can
# hold their difficulty" health check). If keeping every block matters more
# to you than serving rented hashrate on that port, drop its min_diff — the
# cap then applies as it always has.
#
# listen_port and any listener without a min_diff are untouched by all of
# this: no floor, cap as before.

# Ceiling on mining.submit per second, per connection. 0 disables it.
#
# A connection's share rate is its hashrate divided by the difficulty it was
# assigned, and nothing stops those from being badly mismatched — an
# aggregated fleet pointed at a home-miner port is 1 PH/s against difficulty
# 1, which is ~232,000 submits per second down one socket. Validating a submit
# costs about 9 microseconds, so that one connection asks for more than two
# cores, and every share it lands passes through the store's ring, which drops
# events once full: work a miner was told was accepted, never credited.
#
# The default is far above anything a correctly configured miner reaches. With
# vardiff on, the steady state is vardiff_target_spm — 0.2/s at the default —
# for a connection of ANY size, because that is what vardiff converges on. The
# only rates that come near this ceiling are transients while vardiff climbs,
# and they only get near it when the difficulty is already badly wrong. For
# scale, the largest plausible legitimate burst — 1 EH/s at difficulty
# 65536, before vardiff has raised it — is about 3,500/s.
#
# Raise it if you pin a low difficulty for a very large miner
# (vardiff_enabled = 0), since then a high sustained rate is the correct
# behaviour rather than a symptom.
max_submits_per_sec = 20000

# Idle-connection reaper, in seconds. Two budgets, because the two states are
# not the same risk:
#
#   idle_timeout_sec            a socket that has not authorized. It has told
#                               us nothing and costs an fd for nothing.
#   idle_timeout_authorized_sec a miner that HAS authorized. Having nothing to
#                               say is its normal resting state — the pool
#                               asks it nothing between shares — so a small
#                               rig that has not cleared its assigned
#                               difficulty yet is silent but perfectly
#                               healthy. Reaping it on the short budget
#                               disconnects working hashrate, which is one of
#                               the behaviours marketplaces delist pools for.
#
# TCP keepalive (2 min idle + 3x30s probes) already reaps a socket whose peer
# is genuinely gone, so the long budget only has to catch a peer answering
# keepalives while doing no work. Negative disables either one.
idle_timeout_sec            = 600
idle_timeout_authorized_sec = 7200

db_path = ./data/shares.db
commit_window_ms = 100
commit_max_shares = 100

# Days of template history to keep on the /templates page. One row is kept per
# materially distinct template, so this is roughly one row per block plus one
# per source/commitment change — set 0 to keep everything. Only the dashboard
# reads this table; the ledger lives in shares and rate_history.
templates_retention_days = 30

# Redis broadcast (optional). When set, accepted shares / rejects / blocks /
# tip changes are PUBLISHed on channels pool:shares / pool:rejects /
# pool:blocks / pool:tip as JSON. SQLite remains the source of truth — the
# broadcast is fire-and-forget. Leave empty to disable.
# redis_url = redis://127.0.0.1:6379
redis_url =
redis_publish_timeout_ms = 200
redis_reconnect_backoff_ms = 2000

# pool_mode controls the coinbase shape and the username validation:
#   solo        default. Coinbase pays the miner (stratum username is a
#               BTC address) minus the operator fee. No PPS accrual.
#   pplns-thunder / pplns-btc
#               Pay Per Last N Shares. The coinbase pays pool_btc_address, as
#               in pps-classic, but nothing is credited when a share arrives.
#               When a block matures, the reward and its transaction fees are
#               split across everyone whose shares fall inside the last-N
#               window, in proportion to the difficulty each contributed.
#
#               The difference that matters to an operator is who carries the
#               variance. PPS guarantees a price per share, so a run of bad
#               luck is the operator's problem and has to be absorbed by a
#               reserve measured in block rewards. PPLNS never owes more than
#               it has just been paid, so there is no reserve to size and
#               operator ruin is not a failure mode. The miners carry the
#               variance instead, which is why the fee is normally set lower
#               than under PPS -- there is no risk premium to charge.
#
#               The suffix picks the payout rail, and it is one knob rather
#               than two because a pool runs one or the other: the rail
#               decides what a stratum username IS. pplns-thunder takes
#               Thunder addresses and pays over Thunder, exactly as
#               pps-classic does. pplns-btc takes Bitcoin addresses and pays
#               on L1.
#
#               pplns-btc needs the bip300301_enforcer running with
#               --enable-wallet, and pool_btc_address set to an address from
#               that wallet (WalletService/CreateNewAddress). The pool holds
#               no keys and builds no transactions: the coinbase pays into
#               the enforcer's wallet, and paying miners is one RPC to
#               WalletService/SendTransaction, which selects the inputs,
#               signs and broadcasts. The payout worker needs PAYOUT_RAIL=btc
#               and ENFORCER_RPC_ADDR to match — see payout/README.md.
#
#   pps-classic Coinbase pays a pool BTC address (pool_btc_address below)
#               as a normal P2WPKH/P2PKH output. Miners authorize with a
#               Thunder address (base58 of a 20-byte hash) and accrue in
#               pps_credits. The operator batches accumulated BTC into
#               Thunder from the admin dashboard, and the payout worker
#               drains that reserve to miners.
pool_mode = solo

# pps-classic and both pplns modes — required
# pool_btc_address = REPLACE_WITH_POOL_BTC_ADDRESS

# pplns — the window, as a multiple of the CURRENT network difficulty.
# 2.0 means "the last two blocks' worth of expected work".
#
# A multiple rather than an absolute share count or difficulty sum, because it
# self-scales across retargets. An absolute window silently changes meaning
# every time the chain retargets: on a forknet moving 4x it becomes four times
# longer or shorter than the operator chose, with nothing in the config having
# changed to say so.
#
# Below 1.0 the window covers less work than a block is expected to take, so a
# block pays out across less work than it took to find. That rewards whoever
# happened to be connected at the moment over the work that actually produced
# the block -- which is the pool-hopping incentive PPLNS exists to remove. The
# proxy warns if you set it there.
# pplns_window_diff_multiple = 2.0

# pps-classic — OPTIONAL rate override, sats credited per unit of share
# difficulty. Leave it commented out and the proxy derives the rate from each
# block template as (coinbasevalue / network_difficulty) * (1 - fee_bps/1e4).
# That is the recommended setup: fee_bps becomes the single knob controlling
# the fee, and the rate follows difficulty instead of going stale.
#
# If you do set it, the value is used verbatim and treated as ALREADY NET of
# fee — fee_bps is NOT applied on top. A fixed rate drifts as difficulty
# moves and can invert into paying miners more than each share earns, so the
# proxy logs the fee your value actually implies and warns when it disagrees
# with fee_bps. Prefer leaving it unset.
# pps_sats_per_diff = 1000

# pps-classic — READ THIS BEFORE RUNNING PPS ON A NEW CHAIN.
#
# The derived rate is block_value / network_difficulty, which is a share's
# expected value. That is correct only while every share the pool produces has
# a real chance of becoming a block — which needs difficulty to be calibrated
# to hashrate. On a young chain it is not: difficulty starts at 1 and climbs,
# and until it catches up the pool produces solutions far faster than the chain
# can accept blocks. The formula does not know that, and prices every share as
# though it were worth a whole block.
#
# This is not theoretical. A 40 TH/s pool on a forknet that began at difficulty
# 1 accrued 15,561,471 BTC of liability in under four hours, against 943.60 BTC
# it had actually mined. Nothing was paid out only because the payout worker
# had not run yet.
#
# Set the floor to the difficulty at which YOUR pool alone would find one block
# per block interval:
#
#     pps_min_network_difficulty = hashrate_in_H/s * block_interval_sec / 2^32
#
#   10 TH/s  ->  ~1,400,000        100 TH/s ->  ~14,000,000
#   40 TH/s  ->  ~5,600,000          1 PH/s ->  ~140,000,000
#
# That is a FLOOR, not a target — you share the chain with other miners, so the
# genuinely safe difficulty is higher. Below it the proxy credits nothing and,
# by default, refuses new miners rather than taking work it will not pay for.
# It resumes on its own once the chain retargets, and logs the difficulty it
# observes to be necessary either way.
#
# 0 (the default) disables the check. That is only safe on a chain whose
# difficulty is already calibrated: mainnet, testnet, signet. The proxy also
# applies an automatic ceiling that caps accrual at what the chain can mint,
# but that needs a minute of hashrate history — it cannot protect the first
# shares after a restart. This floor is what does.
# pps_min_network_difficulty = 5600000

# Target seconds between blocks. 600 for Bitcoin and every chain forked from
# it. Only change this for a chain that genuinely retargets to a different
# interval — it feeds both the floor above and the issuance ceiling.
# block_interval_sec = 600

# While accrual is suspended by the floor, refuse mining.authorize and reject
# submits, so miners point elsewhere instead of hashing uncredited. Default on.
# Turn it off only if the miners are yours and you know they are working for
# nothing.
# pps_refuse_shares_below_min = 1

log_level = info