2 F.30. pg_prewarm — preload relation data into buffer caches #
5 F.30.2. Configuration Parameters
8 The pg_prewarm module provides a convenient way to load relation data
9 into either the operating system buffer cache or the PostgreSQL buffer
10 cache. Prewarming can be performed manually using the pg_prewarm
11 function, or can be performed automatically by including pg_prewarm in
12 shared_preload_libraries. In the latter case, the system will run a
13 background worker which periodically records the contents of shared
14 buffers in a file called autoprewarm.blocks and will, using 2
15 background workers, reload those same blocks after a restart.
19 pg_prewarm(regclass, mode text default 'buffer', fork text default 'main',
20 first_block int8 default null,
21 last_block int8 default null) RETURNS int8
23 The first argument is the relation to be prewarmed. The second argument
24 is the prewarming method to be used, as further discussed below; the
25 third is the relation fork to be prewarmed, usually main. The fourth
26 argument is the first block number to prewarm (NULL is accepted as a
27 synonym for zero). The fifth argument is the last block number to
28 prewarm (NULL means prewarm through the last block in the relation).
29 The return value is the number of blocks prewarmed.
31 There are three available prewarming methods. prefetch issues
32 asynchronous prefetch requests to the operating system, if this is
33 supported, or throws an error otherwise. read reads the requested range
34 of blocks; unlike prefetch, this is synchronous and supported on all
35 platforms and builds, but may be slower. buffer reads the requested
36 range of blocks into the database buffer cache.
38 Note that with any of these methods, attempting to prewarm more blocks
39 than can be cached — by the OS when using prefetch or read, or by
40 PostgreSQL when using buffer — will likely result in lower-numbered
41 blocks being evicted as higher numbered blocks are read in. Prewarmed
42 data also enjoys no special protection from cache evictions, so it is
43 possible that other system activity may evict the newly prewarmed
44 blocks shortly after they are read; conversely, prewarming may also
45 evict other data from cache. For these reasons, prewarming is typically
46 most useful at startup, when caches are largely empty.
47 autoprewarm_start_worker() RETURNS void
49 Launch the main autoprewarm worker. This will normally happen
50 automatically, but is useful if automatic prewarm was not configured at
51 server startup time and you wish to start up the worker at a later
53 autoprewarm_dump_now() RETURNS int8
55 Update autoprewarm.blocks immediately. This may be useful if the
56 autoprewarm worker is not running but you anticipate running it after
57 the next restart. The return value is the number of records written to
60 F.30.2. Configuration Parameters #
62 pg_prewarm.autoprewarm (boolean)
63 Controls whether the server should run the autoprewarm worker.
64 This is on by default. This parameter can only be set at server
67 pg_prewarm.autoprewarm_interval (integer)
68 This is the interval between updates to autoprewarm.blocks. The
69 default is 300 seconds. If set to 0, the file will not be dumped
70 at regular intervals, but only when the server is shut down.
72 These parameters must be set in postgresql.conf. Typical usage might
75 shared_preload_libraries = 'pg_prewarm'
77 pg_prewarm.autoprewarm = true
78 pg_prewarm.autoprewarm_interval = 300s
83 Robert Haas <rhaas@postgresql.org>