EXPLAIN

Query Plan

EXPLAIN SELECT * FROM users WHERE age > 30;

Shows the logical and physical query plan. Useful for understanding how the query planner routes to engines and which indexes are used.

Cross-shard dispatch

For writes that span multiple vShards, EXPLAIN reports whether the query routes through the Calvin sequencer:

EXPLAIN INSERT INTO orders (id, region, total) VALUES ('o1', 'eu', 99.99);
-- Calvin static dispatch | vshards: [3, 7] | mode: strict | epoch: <assigned by sequencer at admission> | position: <assigned by sequencer>

EXPLAIN INSERT INTO local_cache (id, val) VALUES ('k1', 'v');
-- (single-shard plans emit no preamble — the plan rows that follow describe the local execution)

The cross-shard preamble appears only for sequenced multi-vShard writes. The epoch and position fields show placeholder text at EXPLAIN time — they are assigned by the sequencer when the transaction is actually submitted. Other dispatch modes (Calvin dependent-read dispatch (OLLP), Best-effort multi-shard dispatch [NON-ATOMIC]) replace the leading label.

Session Variables

SET nodedb.consistency = 'eventual';
SHOW nodedb.consistency;
SHOW ALL;
RESET nodedb.consistency;

Change Tracking

SHOW CHANGES FOR users SINCE '2025-01-01' LIMIT 100;

Introspection Commands

SHOW COLLECTIONS;
DESCRIBE users;
SHOW INDEXES;
SHOW TRIGGERS;
SHOW FUNCTIONS;
SHOW PROCEDURES;
SHOW CHANGE STREAMS;
SHOW TOPICS;
SHOW SCHEDULES;
SHOW CONTINUOUS AGGREGATES;
SHOW MATERIALIZED VIEWS;
SHOW CONNECTIONS;
SHOW USERS;
SHOW CLUSTER;
SHOW NODES;
SHOW RAFT GROUPS;
SHOW AUDIT LOG LIMIT 100;

Limitations

EXPLAIN ANALYZE with actual per-operator runtime stats is not yet supported — it requires instrumentation across the SPSC bridge to collect per-core execution stats from the Data Plane. The cross-shard preamble (epoch, position, participating vshards) does appear for sequenced queries.