All pg_deltax functions live in the deltax schema. Names below are shown unqualified for readability — call them as deltax.<fn>(...), e.g.:
SELECT deltax.deltax_create_table('metrics', 'ts', '1 day');
SELECT deltax.time_bucket('1 hour', ts), avg(value) FROM metrics GROUP BY 1;(Or SET search_path TO public, deltax; once if you'd rather call them bare.)
| Function |
Description |
deltax_create_table(relation, time_column, partition_interval DEFAULT '1 day', premake DEFAULT 3) |
Convert a table into a partitioned deltatable. Creates initial partitions around "now". |
deltax_partition_info(relation) |
List all partitions with their range bounds and compression status. |
deltax_deltatable_info(relation) |
Show metadata for a deltatable (time column, interval, partition count). |
deltax_drain_default_partition(relation) |
Move rows that landed in <table>_default (because they fell outside the pre-made partition windows) into proper, time-aligned partitions. Creates the missing partitions on demand. The background worker runs the same step every 60 seconds; call this explicitly after a bulk load to skip the wait. Returns a status string with the row count moved and how many partitions were created. |
| Function |
Description |
deltax_set_retention(relation, drop_after) |
Set a retention policy — partitions older than drop_after are automatically dropped by the background worker. |
deltax_remove_retention(relation) |
Remove the retention policy. |
| Function |
Description |
deltax_enable_compression(relation, segment_by DEFAULT '{}', order_by DEFAULT '{}', segment_size DEFAULT 30000, json_extract DEFAULT NULL) |
Enable compression on a deltatable. Configures how data is segmented and ordered within segments. json_extract is an optional JSONB spec — [{"src","path","name","type"}, ...] — that pulls fields out of a JSONB column into synthetic typed columns at compression time. |
deltax_set_compression_policy(relation, compress_after) |
Set automatic compression — partitions older than compress_after are compressed by the background worker. |
deltax_compress_partition(partition) |
Manually compress a single partition. |
deltax_compress_all_partitions(relation, older_than DEFAULT NULL) |
Compress every sealed uncompressed partition of a deltatable in one call. "Sealed" means range_end <= now() (pg_deltax.mock_now is honoured when set), so the still-open current partition is never touched. With older_than, the threshold becomes now() - older_than. Returns one row (partition_name, result) per partition touched; empty result set means nothing was eligible. |
deltax_decompress_partition(partition) |
Decompress a single partition back to heap storage. |
deltax_analyze_partition(partition) |
Refresh pg_class.reltuples and pg_statistic for a compressed partition from the existing _colstats data. Useful on partitions compressed before the stats-population path shipped, or after an accidental ANALYZE on a compressed partition. |
deltax_analyze_table(relation) |
Run deltax_analyze_partition on every compressed partition of a deltatable. |
deltax_compression_stats(relation) |
Per-partition compression statistics: raw size, compressed size, ratio, row count. |
deltax_table_size(relation) |
Total on-disk size in bytes, accounting for compressed storage. Use with pg_size_pretty() for human-readable output. |
| Function |
Description |
time_bucket(bucket_width, ts) |
Truncate a timestamp to the nearest interval boundary (like date_trunc but for arbitrary intervals). |
time_bucket(bucket_width, ts, origin) |
Same as above but with an offset (e.g., buckets starting at 06:00 instead of 00:00). |
first(value, ts) |
Aggregate: return the value associated with the earliest timestamp. |
last(value, ts) |
Aggregate: return the value associated with the latest timestamp. |
| Function |
Description |
pg_deltax_blob_cache_stats() |
Process-wide blob-cache counters: hits, misses, evictions, current bytes / entries, configured size. |
pg_deltax_blob_cache_shard_stats() |
Same as above but broken down per shard. Used to diagnose hot-shard contention. |