How do you take and restore snapshots in Qdrant?

A snapshot is a point-in-time backup of a collection (or the entire node), which Qdrant can create on demand and later use to restore that collection's exact state — useful for backups, migrating data between clusters, or cloning a collection for testing.

# Create a snapshot of a collection
snapshot_info = client.create_snapshot(collection_name="documents")

# List available snapshots
client.list_snapshots(collection_name="documents")

# Restore is typically done via the REST API on a fresh/target node
# PUT /collections/{collection_name}/snapshots/recover

Snapshots capture the collection's segments, payload data, and configuration as a single downloadable archive, and restoring one loads that archive back into a collection — either the same collection to roll back to a known-good state, or a new collection/node entirely, which is how snapshots double as a migration mechanism between separate Qdrant deployments.

In a distributed cluster, snapshots can be taken per-shard as well as per-collection, and Qdrant's production guidance recommends regular scheduled snapshots as a standard part of an operational backup strategy, separate from the durability the WAL provides against process crashes — the WAL protects against losing recent writes, while snapshots protect against needing to recover an entire collection's history at a specific point in time.