Snapshots

A snapshot is a tarball of data/ from a known-good Safrochain node. New nodes restore from it instead of replaying every block from genesis, cutting hours-to-days of catch-up to a few minutes.

The foundation publishes snapshots of the live chain from rpc2 once per day.

Live snapshot status

Loading snapshot info…

Endpoint	What it returns
https://snapshots.safrochain.network/	Browser-friendly HTML page — always shows current height, size, time, SHA-256, and a download button
https://snapshots.safrochain.network/latest.json	Machine-readable manifest (height, file, sha256, size_bytes, created)
https://snapshots.safrochain.network/latest.tar.lz4	Symlink to the most recent snapshot tarball
https://snapshots.safrochain.network/<file>.tar.lz4	A specific snapshot by name

Filename pattern: safrochain-1_h<height>_<UTC-timestamp>.tar.lz4 (e.g. safrochain-1_h21815_20260531T080608Z.tar.lz4).

The foundation keeps the last 3 snapshots online; older ones are auto-pruned. Snapshots are LZ4-compressed (faster decode than gzip).

Restoring (from a stopped node)

# 1. Stop the daemon
sudo systemctl stop safrochaind

# 2. Wipe data only (keep config + keys), keep the addr_book
sudo -u safrochain safrochaind comet unsafe-reset-all \
  --home /var/lib/safrochain --keep-addr-book

# 3. Fetch the latest snapshot + its sha256 from the manifest
MANIFEST=$(curl -fsSL https://snapshots.safrochain.network/latest.json)
NAME=$(echo "$MANIFEST" | jq -r .file)
SHA=$(echo  "$MANIFEST" | jq -r .sha256)
SIZE=$(echo "$MANIFEST" | jq -r .size_human)

echo "Downloading $NAME ($SIZE)"
curl -L "https://snapshots.safrochain.network/$NAME" -o /tmp/safro.tar.lz4

# 4. Verify integrity (critical: don't skip)
echo "$SHA  /tmp/safro.tar.lz4" | sha256sum -c -

# 5. Extract over the now-empty data dir
sudo -u safrochain bash -c \
  "lz4 -d /tmp/safro.tar.lz4 | tar -xf - -C /var/lib/safrochain"
rm /tmp/safro.tar.lz4

# 6. (Optional but safe) zero out priv_validator_state if you're a validator
#    that has never signed on this chain. Skip if you already signed and want
#    to keep your double-sign protection.
# echo '{"height":"0","round":0,"step":0}' | sudo -u safrochain tee \
#   /var/lib/safrochain/data/priv_validator_state.json

# 7. Start and watch sync
sudo systemctl start safrochaind
sudo journalctl -u safrochaind -f

The node will boot, replay the few hundred blocks since the snapshot was taken (~3 s per block), and reach chain tip in minutes.

Validating freshness before downloading

SNAP=$(curl -fsSL https://snapshots.safrochain.network/latest.json | jq -r .height)
TIP=$( curl -fsSL https://rpc1.safrochain.network/status \
       | jq -r .result.sync_info.latest_block_height)
echo "snapshot=$SNAP tip=$TIP lag=$((TIP - SNAP)) blocks"

Snapshots are typically taken at 04:00 UTC daily, so worst-case lag is ~28 800 blocks (~24 hours at 3 s/block).

Producing your own snapshot

If you operate a node and want to publish snapshots for your community, run something like this (mirrors what the foundation does on rpc2):

sudo systemctl stop safrochaind
H=$(curl -fsSL http://127.0.0.1:26657/status | jq -r .result.sync_info.latest_block_height)
NAME="safrochain-1_h${H}_$(date -u +%Y%m%dT%H%M%SZ).tar.lz4"
tar -cf - -C /var/lib/safrochain --exclude=data/snapshots --exclude=wasm data \
  | lz4 -3 -z > "/srv/snapshots/$NAME"
sha256sum "/srv/snapshots/$NAME" > "/srv/snapshots/$NAME.sha256"
sudo systemctl start safrochaind

Then expose /srv/snapshots/ via nginx + your domain (and back it up off-host: any S3-compatible bucket — AWS S3, Cloudflare R2, Backblaze B2).

Common pitfalls

• Skipping the SHA-256 check → silent corruption later when the node fails to load a malformed DB. Always verify.
• Mismatched LZ4 version → make sure lz4 -V reports at least 1.9.0.
• Restoring across chain-ids → fails. The data dir embeds chain-id; never mix snapshots from different networks.
• Restoring while still acting as a validator → make sure your validator is jailed / unbonded or that another instance is signing before you restore; otherwise you risk missing blocks long enough to be jailed.
• Forgetting --keep-addr-book → your node restarts with no peer list; you'll be slow to connect. Always pass --keep-addr-book to unsafe-reset-all if you want to retain peer discovery.