An archive node keeps every historical state (pruning = "nothing"). It needs far more disk than a pruned node, but it can answer queries at any height.

Configure pruning

In ~/.terpd/config/app.toml:

sed -i -e 's/^pruning *=.*/pruning = "nothing"/' $HOME/.terpd/config/app.toml

Restart the node. From here on it retains everything — it does not backfill history it already pruned.

Syncing an archive

Syncing from genesis replays the entire chain and takes a long time. Faster: restore a published archive pack if one is fresh, then let the node follow head:

# same layout as the daily packs, under archive/
curl -s https://s3.terp.network/snapshots/mainnet/morocco-1/archive/snapshot.json | jq -r .latest

Compare the pack height against the live tip first — a stale archive means replaying the gap anyway. Run the binary matching the network's current release (Install); historical heights may need older binaries, see Upgrades history.

Disk planning

Archive data grows monotonically. Give the node dedicated NVMe storage, monitor disk usage with the same monitoring stack as validators, and never run an archive and a validator on the same host competing for I/O.