Current-Snapshot Bootstrap

One-page verified rocksdb restore for a returning Materios preprod validator — fetch the current snapshot, sha256-verify, restore, point at the live bootnode, and confirm you're in the canonical GRAND

This is the hand-off recipe for a returning external validator (Draupnir, TrueAiData, Runir, or any operator re-joining after downtime). It does one thing: get your node onto the current preprod state without genesis-replaying into a stall or a divergent fork.

Why you need this. partner-chains-node v1.8.0 has a historical-sync bug — a fresh node genesis-replays and stalls at block 0 with Inherent error: Candidates inherent required. Worse, a node that replays from an old database can finalize into a divergent GRANDPA room that never re-merges. The cure for both is to restore the published rocksdb snapshot, then prove you landed in the canonical room.

1. Restore the current snapshot

The snapshot is published hourly and sha256-verified. Read the manifest, pull the tarball, verify, restore:

mkdir -p ~/materios-preprod/data/chains
cd ~/materios-preprod

# Pull the current pointer
MANIFEST=$(curl -fsSL https://materios.fluxpointstudios.com/operator-snapshots/preprod/latest.json)
TARBALL_URL=$(echo "$MANIFEST" | jq -r .url)
EXPECTED_SHA=$(echo "$MANIFEST" | jq -r .sha256)

echo "$MANIFEST" | jq '{url, sha256, size_bytes, head_block_at_publish, published_at}'

# Download + verify before touching your data dir
curl -fLo snapshot.tar.gz "$TARBALL_URL"
echo "${EXPECTED_SHA}  snapshot.tar.gz" | sha256sum -c -   # MUST print "snapshot.tar.gz: OK"

# Restore only on a clean checksum
tar xzf snapshot.tar.gz && rm snapshot.tar.gz

The tarball contains only data/chains/materios_preprod_v6/db/ — no keystore, no peer-id keys. Your validator identity stays yours. If the sha256sum -c - line does not print OK, stop: the download is corrupt or the manifest moved under you. Re-fetch the manifest and retry.

Re-restoring over an existing DB: if you already have a stalled or divergent database, wipe it first — rm -rf ~/materios-preprod/data/chains/materios_preprod_v6/db — then untar. Leaving the old db/ in place can leave you on the wrong fork.

2. Point at the live bootnode

There is exactly one reliable public bootnode. Use the DNS form — it tracks the node's WAN IP so it survives IP changes:

/dns4/bootnode.materios.fluxpointstudios.com/tcp/30333/p2p/12D3KooWPueKoxRAirTTKH4Y2qQAsJDegWMjS4k89Z7izCbZKgkM

Pass it on the command line:

materios-node-spo \
  --chain ~/materios-preprod/chain-spec-v6-raw.json \
  --base-path ~/materios-preprod/data \
  --validator \
  --name <your-label> \
  --port 30333 \
  --rpc-port 9945 \
  --public-addr /ip4/<YOUR.PUBLIC.IP>/tcp/30333 \
  --bootnodes /dns4/bootnode.materios.fluxpointstudios.com/tcp/30333/p2p/12D3KooWPueKoxRAirTTKH4Y2qQAsJDegWMjS4k89Z7izCbZKgkM

Do not use any /ip4/... bootnode address you may have from an older runbook — those egresses are dead. --public-addr is required for SPOs so peers can dial you back.

3. Post-sync divergence self-check

Restoring the snapshot is necessary but not sufficient — you must confirm you finalized into the canonical room. Compare your node's block hash at the network's finalized height against the network's hash at the same height:

NET=$(curl -fsSL https://materios.fluxpointstudios.com/chain-info)
H=$(echo "$NET" | jq -r .finalized_block)    # network's finalized height
HNET=$(echo "$NET" | jq -r .finalized)        # network's finalized hash at H

# Ask YOUR node for its block hash at the network's finalized height
HLOCAL=$(curl -s -X POST http://127.0.0.1:9945 \
  -H 'content-type: application/json' \
  -d "{\"jsonrpc\":\"2.0\",\"id\":1,\"method\":\"chain_getBlockHash\",\"params\":[$H]}" | jq -r .result)

if [ "$HLOCAL" = "$HNET" ]; then
  echo "CANONICAL room ✓  (local hash at $H == network $HNET)"
elif [ "$HLOCAL" = "null" ]; then
  echo "Not caught up yet — local node hasn't reached $H. Wait and re-run."
else
  echo "DIVERGENT room ✗  local=$HLOCAL  network=$HNET  → re-restore from step 1 and restart"
fi

How to read it:

• Match → you're in the canonical room. Done.

• null → your node simply hasn't synced up to H yet (the network tip moves while you sync). Wait a minute and re-run; this is not a fork.

• Different non-null hash → you finalized a different block at the same height as the network. That is a real divergent fork: wipe data/chains/materios_preprod_v6/db, re-restore from step 1, and restart.

Always compare at the network's finalized height (H from /chain-info), never at your local tip — your local node legitimately lags the network by a few blocks while catching up, and comparing at your own tip would give a false "diverged" reading.

What "good" looks like

After restore + start, within a couple of minutes:

curl -s -X POST http://127.0.0.1:9945 \
  -H 'content-type: application/json' \
  -d '{"jsonrpc":"2.0","id":1,"method":"system_health"}'

Expect peers ≥ 1 and isSyncing flipping to false once you pass the snapshot floor (head_block_at_publish from the manifest) and reach the live tip. Then run the step 3 self-check — a CANONICAL room ✓ is your green light to start producing.

For the full onboarding flow (keys, Cardano registration, Ariadne selection), see SPO Onboarding. For hardware and network requirements, see Node Requirements.