Statesync

Statesync uses CometBFT light-client verification to fetch a recent state snapshot directly from peers. Unlike a tarball, the result is cryptographically verified against the validator set itself.

For Safrochain we recommend statesync to rpc1 + rpc2 as trust anchors.

Prerequisites

• A fresh safrochaind init home (no data/).
• The chain's genesis.json already in place.
• Two trusted RPC URLs that have statesync = true in their app.toml (rpc1 and rpc2 do).

Configure

Edit ~/.safrochain/config/config.toml:

[statesync]
enable = true

# pin a recent trusted height
rpc_servers = "https://rpc1.safrochain.network:443,https://rpc2.safrochain.network:443"
trust_height = <RECENT_HEIGHT>
trust_hash   = "<HEX_BLOCK_HASH_AT_TRUST_HEIGHT>"
trust_period = "168h"   # 7 days; must be <= chain unbonding period

Pick <RECENT_HEIGHT> and <HEX_BLOCK_HASH_AT_TRUST_HEIGHT> with this one-liner:

LATEST=$(curl -s https://rpc.safrochain.network/status \
           | jq -r '.result.sync_info.latest_block_height')
TRUST=$((LATEST - 1000))
HASH=$(curl -s "https://rpc.safrochain.network/block?height=$TRUST" \
         | jq -r '.result.block_id.hash')
echo "trust_height = $TRUST"
echo "trust_hash   = \"$HASH\""

(1 000 blocks of safety margin is plenty.)

Configure the snapshot endpoint

In ~/.safrochain/config/app.toml make sure:

[state-sync]
snapshot-interval = 1000
snapshot-keep-recent = 2

These are read by the node serving the snapshot, not the joiner. rpc1/rpc2 already have these set.

Start

safrochaind start

You should see log lines like:

INFO Discovering snapshots ...
INFO Discovered snapshot height=2499000 chunks=42
INFO Fetching snapshot chunks ...
INFO Verified snapshot height=2499000 hash=...
INFO Snapshot restored
INFO Catching up from snapshot height to chain tip

After a few minutes, a curl /status should show catching_up: false.

Why this is safer than a tarball

• The trust_hash you paste is verified against the actual chain header at trust_height.
• The state snapshot's Merkle commitments are checked against the verified header.
• A malicious RPC can serve you a fake snapshot, but it cannot forge a matching header without forging signatures from > 1/3 of the validator set within the trust period.

When statesync fails

Symptom	Likely cause	Fix
no snapshots discovered	the RPC has no snapshots yet	switch to a different RPC, or wait an hour
light client error: trusted height invalid	trust_height > chain tip, or stale	refresh trust_height + trust_hash
client expired mid-restore	trust_period elapsed	increase trust_period (≤ unbonding)
stuck after restore	bad peers	clear addrbook.json, restart

If statesync fails repeatedly, fall back to a Snapshot tarball.

After you've synced

You can keep statesync enabled. It has no effect once you are at chain tip. Or set:

[statesync]
enable = false

…to make subsequent restarts faster (skips the discovery handshake).