Skip to content

Starting Node from Global State Backup DB

Jump to bottom
Joe Sacher edited this page Apr 2, 2026 · 2 revisions

We will be automating the backup practice for tip only global state DB of TestNet and MainNet. Documentation will update as this is completed.

Validate Node Operation

  • Setup a node normally for the environment.
  • Get trusted hash and setup keys.
  • Start with empty DB.

You should see the node link up and get into "CatchUp" mode with casper-node-util watch. This validates that all on the node should be good and you are syncing historical blocks from the network. This will work, but since syncing is not prioritized over operation, the process will take some time.

Stop the node sudo casper-node-util stop.

Prerequisites

Insure that zstandard package is installed: sudo apt install zstd

Storage Organization

Links to versions of the global-state archive will be stored in: [https://genesis.casper.network/network name]/backup/versions

We can use the follow to get the latest version with no new line for downloading: curl -s https://genesis.casper.network/[network name]/backup/versions | head -n 1 | tr -d '\n\r'

TestNet Tip DB

Process to pull down TestNet DB tip archive:

# Make sure you are in this directory before deleting everything below
cd /var/lib/casper/casper-node/casper-test
rm -rf *

# Download and decompress the global-state DB
DOWNLOAD_URL=$(curl -s https://genesis.casper.network/casper-test/backup/versions | head -n 1 | tr -d '\n\r')
# Print out the location to verify it was pulled correctly if the next step fails.
echo "Downloading from $DOWNLOAD_URL"

# Download and decompress DB, this will take some time
curl -s --output - "$DOWNLOAD_URL" | zstd -d --long=31 | sudo -u casper tar -xv

MainNet Tip DB

Process to pull down MainNet DB tip archive:

cd /var/lib/casper/casper-node/casper
rm -rf *

# Download and decompress the global-state DB
DOWNLOAD_URL=$(curl -s https://genesis.casper.network/casper/backup/versions | head -n 1 | tr -d '\n\r')
# Print out the location to verify it was pulled correctly if the next step fails.
echo "Downloading from $DOWNLOAD_URL"

# Download and decompress DB, this will take some time
curl -s --output - "$DOWNLOAD_URL" | zstd -d --long=31 | sudo -u casper tar -xv

Starting Node

Update your trusted_hash to have current one before starting.

When you start the node and pull up watch: casper-node-util watch, you will see it move into CatchUp mode and see the block the db archive was captured at:

Reactor State: CatchUp
Available Block Range - Low: 5427143  High: 5427143

You will not sync all blocks back to this, just catch up global-state and the TTL window if using default settings.

You can see the sync of data tries with metrics:

# HELP trie_or_chunk_fetch_total number of trie_or_chunk all fetch requests made
# TYPE trie_or_chunk_fetch_total counter
trie_or_chunk_fetch_total 10831

Rerunning this command should so more totals fetched.

After some time getting the missing global state and needed block data at the tip, your node will switch to KeepUp. This makes it ready to follow the network and add a bid if you wish this node to be a validator.

Clone this wiki locally