> ## Documentation Index
> Fetch the complete documentation index at: https://seilabs-monty-arm64-prebuilt-binary-docs.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Join the network using Snapshots

> Detailed guide for using snapshots to join the network

<Warning>Before you resync a node using snapshots, make sure that in case of a successful resync that you *under no circumstance* double sign blocks at previous heights with your validator. Failure to do so will cause tombstoning of your validator.</Warning>

Follow this guide to join an existing network through **snapshot sync**. To quickly spin up a fresh full node and join the network, it's recommended to restore from a snapshot instead of replaying all historical blocks.

## Snapshot Sync

Snapshot sync allows a new node to join a network by downloading a recent, compressed copy of the entire application state and extracting it directly into the data directory. This reduces the initial sync time from days to minutes.

<Warning>RocksDB support for the SeiDB state store will be removed. No target release has been published. For new or resynced nodes, use a snapshot only after its provider confirms that the state store uses PebbleDB. Keep `ss-backend = "pebbledb"` in `app.toml`. See [Move off RocksDB](/node/node-operators#move-off-rocksdb) if you are replacing an existing RocksDB node.</Warning>

<Danger>Do not restore a pruned snapshot onto an archive node. It does not contain the earlier state-store versions that the archive node must retain. Use only a provider-confirmed full-history PebbleDB archive snapshot, or follow the archive-node guidance in [Move off RocksDB](/node/node-operators#move-off-rocksdb).</Danger>

## Snapshot Providers

You can select from various providers for downloading snapshots:

* Polkachu: [Mainnet Snapshots](https://www.polkachu.com/tendermint_snapshots/sei) | [Testnet Snapshots](https://www.polkachu.com/testnets/sei/snapshots)
* Imperator.co
* Stakeme
* kjnodes

<iframe src="https://seistream.app/labs?page=snapshots" width="100%" height="600" style={{ border: 'none', borderRadius: '8px' }} title="Sei Snapshot Providers" />

Snapshot providers do not consistently label the state-store backend. Ask the
provider to confirm that the snapshot uses PebbleDB before you restore it. The
post-extraction check below is a second guard against archives that contain an
obvious RocksDB path. It does not replace provider confirmation because some
legacy and custom directory names do not identify their backend.

## Clean Up & Preparation

If you are **not** starting a node from fresh, perform the following backups and clean‑ups first.

<Info>Note: This step is not needed for fresh nodes.</Info>

1. **Stop the service**

   ```bash theme={null}
   sudo systemctl stop seid
   ```

2. **Backup Validator State** (Critical for Validators)
   Assuming your sei home directory is `$HOME/.sei`, back up `priv_validator_key.json` and `priv_validator_state.json`:

   ```bash theme={null}
   cp $HOME/.sei/data/priv_validator_state.json $HOME/priv_validator_state.json
   cp $HOME/.sei/config/priv_validator_key.json $HOME/priv_validator_key.json
   ```

3. **Reset the State**

   ```bash theme={null}
   seid tendermint unsafe-reset-all --home $HOME/.sei
   ```

4. **Check custom state-store paths**
   Print the state-store section from `app.toml`:

   ```bash theme={null}
   sed -n '/^\[state-store\]/,/^\[.*\]/p' $HOME/.sei/config/app.toml
   ```

   If `ss-db-directory` or `evm-ss-db-directory` is not empty, the state store
   may live outside `$HOME/.sei/data`. Back up anything you need, then move or
   remove the old state-store data before you restore the snapshot. The generic
   commands below expect both settings to be empty so the snapshot's default
   paths are used. Follow your provider's placement instructions if you keep
   custom paths. Do not reuse a RocksDB directory with
   `ss-backend = "pebbledb"`.

5. **Remove data and Wasm**
   ```bash theme={null}
   rm -rf $HOME/.sei/data
   rm -rf $HOME/.sei/wasm
   ```

<Warning>All snapshots also include the `wasm` folder. Make sure to copy over the `wasm` folder as well, the node cannot successfully sync without it.</Warning>

## Download & Restore

<Info>The following commands are generic examples. Please verify the `SNAPSHOT_URL` and extraction command from your chosen provider above.</Info>

### Prerequisites

Ensure you have the necessary tools installed (e.g., `lz4`, `aria2`, `pv`, `wget`).

```bash theme={null}
sudo apt update
sudo apt install curl lz4 wget aria2 pv -y
```

### Download and Extract

1. **Set the Snapshot URL**
   Replace `<SNAPSHOT_URL>` with the link from your chosen provider.

   ```bash theme={null}
   SNAPSHOT_URL="<PASTE_SNAPSHOT_URL_HERE>"
   ```

2. **Extract the snapshot**
   Most providers compress the `data` and `wasm` directories directly. Stream
   the archive to avoid storing a second compressed copy:

   ```bash theme={null}
   set -o pipefail
   curl --fail -L "$SNAPSHOT_URL" | lz4 -c -d | tar -x -C "$HOME/.sei"
   ```

   **Alternative: parallel download with aria2**

   `aria2c` supports parallel connections but keeps the compressed archive
   beside the extracted data. Put the archive on a data disk with enough free
   space for both:

   ```bash theme={null}
   set -o pipefail
   SNAPSHOT_FILE="/path/on-data-disk/snapshot.tar.lz4"
   aria2c -x 16 -s 16 \
     -d "$(dirname "$SNAPSHOT_FILE")" \
     -o "$(basename "$SNAPSHOT_FILE")" \
     "$SNAPSHOT_URL" && \
     pv "$SNAPSHOT_FILE" | lz4 -c -d | tar -x -C "$HOME/.sei"
   ```

   <Info>The `-x 16` flag sets the maximum connections per server, and `-s 16` splits the file into 16 segments for parallel downloading. You can adjust these values based on your network conditions.</Info>

   If extraction fails, remove the partially extracted `data` and `wasm`
   directories before you retry. Do not continue to verification.

   <Warning>Some providers wrap the data in another directory or use a different compression format. For a `.tar.gz` snapshot, use the provider's `tar -xzf` command. If the archive contains a root directory such as `sei/data`, adjust the extraction target or `--strip-components`. Run the backend check below after any extraction method.</Warning>

3. **Verify every extracted state-store path**
   The check below scans every backend-labelled default state-store path
   instead of stopping at the first result. It catches archives that contain
   both PebbleDB and RocksDB:

   ```bash theme={null}
   verify_snapshot_backend() {
     local state_store_config backend_dirs backend path

     if [ ! -f "$HOME/.sei/config/app.toml" ]; then
       echo "UNVERIFIED: app.toml not found. Do not start seid." >&2
       return 1
     fi

     state_store_config="$(
       sed -n '/^\[state-store\]/,/^\[.*\]/p' \
         "$HOME/.sei/config/app.toml"
     )"

     if printf '%s\n' "$state_store_config" | \
       grep -Eq "^[[:space:]]*(ss-db-directory|evm-ss-db-directory)[[:space:]]*=[[:space:]]*(\"[^\"]+\"|'[^']+')"; then
       echo "UNVERIFIED: custom state-store directory configured." >&2
       echo "Use the provider-specific placement and verification steps." >&2
       return 1
     fi

     backend_dirs="$(
       for backend in pebbledb rocksdb; do
         for path in \
           "$HOME/.sei/data/$backend" \
           "$HOME/.sei/data/state_store/cosmos/$backend" \
           "$HOME/.sei/data/state_store/evm/$backend"; do
           if [ -d "$path" ]; then
             printf '%s\n' "$path"
           fi
         done
       done
     )"

     printf '%s\n' "$backend_dirs"

     if printf '%s\n' "$backend_dirs" | grep -Eq '(^|/)rocksdb$'; then
       echo "FAIL: RocksDB data found. Do not start seid." >&2
       return 1
     fi

     if ! printf '%s\n' "$backend_dirs" | grep -Eq '(^|/)pebbledb$'; then
       echo "UNVERIFIED: no PebbleDB state-store directory found." >&2
       return 1
     fi

     echo "PASS: PebbleDB found and no RocksDB directory found."
   }

   verify_snapshot_backend
   ```

   The function returns a nonzero status for `FAIL` and `UNVERIFIED`. Continue
   only if the provider confirmed PebbleDB and the command prints `PASS`. A
   legacy EVM store at `data/evm_ss/` does not identify its backend in the
   directory name, which is why provider confirmation is still required. If
   you used `aria2c`, remove the downloaded archive after this check passes:

   ```bash theme={null}
   rm "$SNAPSHOT_FILE"
   ```

4. **Restore validator state**

   ```bash theme={null}
   cp $HOME/priv_validator_state.json $HOME/.sei/data/priv_validator_state.json
   ```

5. **Enable SeiDB and configure PebbleDB**
   Make sure SeiDB is enabled and the state-store backend is PebbleDB:

   ```bash theme={null}
   sed -i.bak -E "/^\[state-commit\]/,/^\[.*\]/ s|^[#[:space:]]*(sc-enable[[:space:]]*=[[:space:]]*).*$|\1true| ; /^\[state-store\]/,/^\[.*\]/ s|^[#[:space:]]*(ss-enable[[:space:]]*=[[:space:]]*).*$|\1true| ; /^\[state-store\]/,/^\[.*\]/ s|^[#[:space:]]*(ss-backend[[:space:]]*=[[:space:]]*).*$|\1\"pebbledb\"|" $HOME/.sei/config/app.toml

   sed -n '/^\[state-store\]/,/^\[.*\]/p' $HOME/.sei/config/app.toml
   ```

   Confirm that the output includes `ss-backend = "pebbledb"`. If the key is
   absent, add it directly below `[state-store]` before you restart the node.

6. **Restart the node**

   ```bash theme={null}
   sudo systemctl start seid
   ```

7. **Monitor logs**
   ```bash theme={null}
   sudo journalctl -fu seid
   ```

## Troubleshooting

**Q: I can't download a snapshot.**
A: Try another time later as these snapshots are refreshed regularly and inform us in the [Sei Tech Chat](https://t.me/+KZdhZ1eE-G01NmZk).

**Q: The snapshot finishes, but I immediately get `AppHash` errors upon regular block syncing.**
A: Make sure that you use the latest version of the node software. This usually means the snapshot version doesn't match your node version, or the snapshot is corrupted. Ensure you are using the correct binary version for the block height of the snapshot.

**Q: "No space left on device"**
A: Snapshots require significant disk space to download and extract. Ensure you have enough free space (check with `df -h`).