Neo4j Graph Database Transaction Log Files (neostore.transaction.db.*)

Bottom Line: It is safe to prune old committed Neo4j transaction logs using Neo4j’s configuration retention settings or the dbms.checkpoint() procedure.

Why Do Neo4j Transaction Logs Exist?

Neo4j graph database writes write-ahead transaction logs (neostore.transaction.db.0, neostore.transaction.db.1) inside data/databases/neo4j/transactions/.

• Primary Purpose: Transaction recovery, ACID compliance, and cluster synchronization across Causal Clustering nodes.
• Storage Growth: High-volume graph updates (node insertions, relationship updates) generate gigabytes of log files.
• Storage Impact: Retaining default transaction logs consumes 5 GB to 50 GB+.

What Happens If You Prune Old Neo4j Transaction Logs?

• System Safety: ✅ Safe to Prune. Old logs that have already been checkpointed to main store files are no longer needed for crash recovery.
• Automatic Retention: Configuring dbms.tx_log.rotation.retention_policy keeps transaction logs bounded automatically.
• Reclaimed Storage: Pruning old logs frees 5 GB to 50 GB+ of disk space.

How to Prune Neo4j Transaction Logs Safely

Method 1: Configure Retention Policy in neo4j.conf

Edit conf/neo4j.conf to prune logs older than 2 days or 10 GB:

# Retain logs for 2 days maximum
dbms.tx_log.rotation.retention_policy=2 days

# Or retain a maximum size of 10 GB
# dbms.tx_log.rotation.retention_policy=10 GB

Method 2: Trigger Checkpoint via Cypher Shell

CALL dbms.checkpoint();

Frequently Asked Questions (FAQ)

Can I manually delete active neostore.transaction.db files while Neo4j is running?

No. Deleting active transaction log files while the Neo4j service is running causes database store corruption and transaction failure.