Optionally prefetch referenced data in recovery.

Introduce a new GUC recovery_prefetch, disabled by default.  When
enabled, look ahead in the WAL and try to initiate asynchronous reading
of referenced data blocks that are not yet cached in our buffer pool.
For now, this is done with posix_fadvise(), which has several caveats.
Better mechanisms will follow in later work on the I/O subsystem.

The GUC maintenance_io_concurrency is used to limit the number of
concurrent I/Os we allow ourselves to initiate, based on pessimistic
heuristics used to infer that I/Os have begun and completed.

The GUC wal_decode_buffer_size is used to limit the maximum distance we
are prepared to read ahead in the WAL to find uncached blocks.

Reviewed-by: Alvaro Herrera <[email protected]> (parts)
Reviewed-by: Andres Freund <[email protected]> (parts)
Reviewed-by: Tomas Vondra <[email protected]> (parts)
Tested-by: Tomas Vondra <[email protected]>
Tested-by: Jakub Wartak <[email protected]>
Tested-by: Dmitry Dolgov <[email protected]>
Tested-by: Sait Talha Nisanci <[email protected]>
Discussion: https://postgr.es/m/CA%2BhUKGJ4VJN8ttxScUFM8dOKX0BrBiboo5uz1cq%3DAovOddfHpA%40mail.gmail.com

1 files changed, 2 insertions, 1 deletions

diff --git a/src/backend/storage/freespace/freespace.c b/src/backend/storage/freespace/freespace.c
index 8c12dda2380..cfa0414e5ab 100644
--- a/src/backend/storage/freespace/freespace.c
+++ b/src/backend/storage/freespace/freespace.c
@@ -210,7 +210,8 @@ XLogRecordPageWithFreeSpace(RelFileNode rnode, BlockNumber heapBlk,
 	blkno = fsm_logical_to_physical(addr);
 
 	/* If the page doesn't exist already, extend */
-	buf = XLogReadBufferExtended(rnode, FSM_FORKNUM, blkno, RBM_ZERO_ON_ERROR);
+	buf = XLogReadBufferExtended(rnode, FSM_FORKNUM, blkno, RBM_ZERO_ON_ERROR,
+								 InvalidBuffer);
 	LockBuffer(buf, BUFFER_LOCK_EXCLUSIVE);
 
 	page = BufferGetPage(buf);