Bulk import/export for bootstrapping

[tac0turtle]

Summary

Support direct DB import/export to bootstrap new apex instances without syncing millions of heights over RPC.

Use case

A new apex deployment for a rollup with 2M+ DA heights of history. RPC sync at ~100 heights/sec = ~5.5 hours. Direct DB import from a snapshot or another indexer = minutes.

Commands

# Export from a running or stopped apex instance
apex export --namespaces eden,xo --from 1000000 --to 2000000 --output dump.sql

# Import into a new instance
apex import --input dump.sql

Requirements

Export

• Dump headers and blobs tables for specified height range and namespaces
• Output format: SQL INSERT statements (portable, inspectable)
• Optional: compressed output (gzip)
• Can run against a live DB (SQLite WAL mode supports concurrent reads)

Import

• Load SQL dump into a fresh or existing database
• Update sync_state to reflect imported range
• Validate: no duplicate heights, namespace consistency
• Transaction-wrapped for atomicity (rollback on error)

Snapshot (alternative)

• Since it's SQLite, a raw file copy of the .db file is also valid
• Document the "stop, copy, start" approach as the simplest option
• apex snapshot --output ./backup/ could do a safe SQLite backup via the backup API

References

• Design doc: test_plan.md — Bulk Import section
• SQLite online backup API: sqlite3_backup_init