Bitcoin Infrastructure Electrs

Electrs

Rust implementation of Electrum protocol server. Reads from Bitcoin Core, indexes per-address history, serves Electrum clients.

Repo: github.com/romanz/electrs. Current release v0.12.0 (13 September 2026).

Why

Electrum protocol is the de-facto standard for SPV-style clients querying history per address. Public Electrum servers exist but for privacy you should run your own.

Setup

Since v0.12.0 (13 September 2026) electrs indexes through bindex and reads blocks over bitcoind's REST interface instead of the P2P protocol, so it requires Bitcoin Core 31.0+ (released April 2026). Requires Bitcoin Core with:

server=1
rest=1
prune=0

txindex is not required - electrs maintains its own index (it is still allowed, and other services such as eclair may need it). Set blockfilterindex=1 / peerblockfilters=1 only when BIP157 clients query Core directly; electrs exposes no compact-filter RPC of its own.

Then:

electrs --network bitcoin \
  --daemon-rpc-addr=127.0.0.1:8332 \
  --electrum-rpc-addr=127.0.0.1:50001

Upstream measured the initial index at ~2 h for ~800 GB of blocks/*.dat on a 6-core / 32 GB / NVMe host, and ~18 h on an ODROID-HC1 (electrs docs, July 2026). The RocksDB index settles at ~7% of blocks/*.dat - ~56 GB on that ~800 GB chain - peaking near ~14% just before the final compaction.

Compatibility

• Sparrow: connect to electrs.
• Electrum (desktop): connect.
• BlueWallet (Watchtower / privacy mode): supports.
• BTCPay Server: optional Electrum backend.
• LDK / btcwallet: scan BIP157 filters served by Core (blockfilterindex / peerblockfilters), not by electrs.

Compared to Fulcrum

Common bugs

• Failing to set rest=1 on Core → electrs v0.12.0+ cannot fetch blocks at all, since it no longer falls back to the P2P protocol.
• Upgrading to v0.12.0 without planning downtime → the bindex format is new and forces a full reindex; upstream asks for 120 GB free before starting (electrs doc/upgrading.md, September 2026).
• electrum-rpc-addr exposed publicly without TLS proxy → DoS risk.
• Reorg handling lag during deep reorgs (rare on mainnet).

See also

• fulcrum/SKILL.md
• esplora/SKILL.md
• ../core/operations/SKILL.md
• ../core/indexes/SKILL.md