Database performance snapshot report reference

This document describes wait events in AlloyDB for PostgreSQL performance snapshot reports. For more information, see Optimize database performance by comparing database snapshots.

Wait events in snapshot reports

The following table describes wait events that are supported by AlloyDB performance snapshot reports. You can use wait events to help you identify performance issues and bottlenecks in your database.

Wait event name	Event class	Description
ColumnarCacheInvalidatedBlockBuild	IPC	Builds the shared invalidated blocks and waits until all workers finish building.
ColumnarCacheInvalidatedBlockCopy	IPC	Copies the shared invalidated blocks into private memory and waits until all workers finish copying.
ColumnarCacheBloomFilterConsolidate	IPC	Consolidates the bloom filter into the bloom filter created in the shared memory, and waits until all workers finish consolidating.
AdaptiveVacuumFreezeXid	IPC	Waits until vacuum advances the oldest frozen XID or until the timeout expires. After the timeout expires, this wait event throws an error telling you to wait for autovacuum or to manually perform the vacuum.
AdaptiveVacuumNewXidDelay	Timeout	New transaction ID delay when the oldest transaction ID age indicates that new transactions are to be throttled.
CommitWaitFlush	I/O	Transaction commits wait for the flush pointer advance before returning to the client.
XlogAdvanceWALBuffer	I/O	Transaction waits for available space in the Write-Ahead Logging (WAL) buffer.
AioWaitSegmentSwitch	I/O	WAL writer waits for WAL segment file switch in asynchronous I/O mode.
WalFlushDueToBufferReplacement	I/O	WAL flush due to buffer replacement.
WalFlushOnBufEvictUnderExtend	I/O	WAL flush on buffer eviction under lock:extend.
WalFlushWhileTruncatingCLOG	I/O	WAL flush while truncating the commit log (CLOG).
WalFlushWhileTruncatingMultiXactData	I/O	WAL flush while truncating MultiXact Data.
WalFlushDueToSLRUPageReplacement	I/O	WAL flush due to simple least-recently used (SLRU) page replacement.
WalFlushByDDL	I/O	WAL flush by Data Definition Language (DDL).
WalFlushDuringCheckPoint	I/O	WAL flush during checkpoint. A checkpoint is the point in the write-ahead log sequence during which all data files are updated to reflect the information in the log. For more information, see Checkpoint.
WalFlushBySmgrOperations	I/O	WAL flush by storage manager operations.
WalFlushWhileReservingReplicationSlot	I/O	WAL flush while reserving a replication slot.
WalFlushWhileSwitchingWALSegment	I/O	WAL flush while switching WAL segments.
WalFlushByRelmapFile	I/O	WAL flush by the relation mapping file.
VacuumDelayRead	I/O	Adaptive vacuum read delay.
VacuumDelayWrite	I/O	Adaptive vacuum write delay.
UltraFastCachePRead	I/O	Latency of a read from the AlloyDB UltraFast Cache.
UltraFastCachePWrite	I/O	Latency of a write to the AlloyDB UltraFast Cache
WalBatchWrite	I/O	Wait time for a minimal batch size to write WAL.
StorageAwaitMayWrite	I/O	Time spent waiting for AlloyDB storage to process logs and allow more writes.
StorageAdvanceFlushPosition	I/O	Time spent waiting to advance the flush position of AlloyDB storage.
StorageDataFilePrefetch	I/O	Latency of prefetch reads from AlloyDB storage.
StorageInProgressDataFileRead	I/O	Appears in pg_stat_activity while a storage read is in progress.
StorageDataFileReadUncached	I/O	Latency of a read from AlloyDB storage.
StorageQueryRelations	I/O	The time it takes to read relation sizes from storage.
StorageLocalFileInit	I/O	The time it takes to read files data from storage and write to local files during startup.
StorageWaitWalLogParsing	I/O	The time it takes to parse logs for AlloyDB storage.
StorageWaitWalBufferSpace	I/O	Time spent waiting for space to free up in the AlloyDB storage buffer.
StorageReadDir	I/O	The time it takes to read a directory's contents from one storage shard.

What's next

• Optimize database performance by comparing database snapshots.