mirror of
https://github.com/timescale/timescaledb.git
synced 2025-05-14 09:37:00 +08:00
08050fb9d6
New procedures to `merge_chunks` are introduced that can merge an arbitrary number of chunks if the right conditions apply. Basic checks are done to ensure that the chunks can be merged from a partitioning perspective. Some more advanced cases that are potentially mergeable are not supported at this time (e.g., chunks with non-adjacent partitioning) and merging of compressed chunks. Merging compressed chunks requires additional work, although the same basic rewrite approach should work also on the internal compressed relations. Still, one needs to handle merges of a compressed chunk and a non-compressed chunk, or two compressed chunks with different compression settings, partially compressed chunks, and so forth. This is left for a future enhancement. Currently, the merge defaults to taking an AccessExclusive lock on the merged chunks to prevent deadlocks and concurrent modifications. Weaker locking is supported via an anonymous settings variable, and it is used in tests to illustrate various deadlock scenarios. Alternative locking approaches, including multi-transactional merges, can be considered in the future. The actual merging is done by rewriting all the data from multiple chunks into a (temporary) merged heap using the same approach as that implemented to support VACUUM FULL and CLUSTER. Then this new heap is swapped into one of the original relations while the rest are dropped. This approach is MVCC compliant and implements correct visibility under higher isolation levels, while also cleaning up garbage tuples.
2 lines
49 B
Plaintext
2 lines
49 B
Plaintext
Implements: #7433 Add support for merging chunks
|