<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>19.4. Resource Consumption</title><link rel="stylesheet" type="text/css" href="stylesheet.css" /><link rev="made" href="pgsql-docs@lists.postgresql.org" /><meta name="generator" content="DocBook XSL Stylesheets Vsnapshot" /><link rel="prev" href="runtime-config-connection.html" title="19.3. Connections and Authentication" /><link rel="next" href="runtime-config-wal.html" title="19.5. Write Ahead Log" /></head><body><div xmlns="http://www.w3.org/TR/xhtml1/transitional" class="navheader"><table width="100%" summary="Navigation header"><tr><th colspan="5" align="center">19.4. Resource Consumption</th></tr><tr><td width="10%" align="left"><a accesskey="p" href="runtime-config-connection.html" title="19.3. Connections and Authentication">Prev</a> </td><td width="10%" align="left"><a accesskey="u" href="runtime-config.html" title="Chapter 19. Server Configuration">Up</a></td><th width="60%" align="center">Chapter 19. Server Configuration</th><td width="10%" align="right"><a accesskey="h" href="index.html" title="PostgreSQL 11.5 Documentation">Home</a></td><td width="10%" align="right"> <a accesskey="n" href="runtime-config-wal.html" title="19.5. Write Ahead Log">Next</a></td></tr></table><hr></hr></div><div class="sect1" id="RUNTIME-CONFIG-RESOURCE"><div class="titlepage"><div><div><h2 class="title" style="clear: both">19.4. Resource Consumption</h2></div></div></div><div class="toc"><dl class="toc"><dt><span class="sect2"><a href="runtime-config-resource.html#RUNTIME-CONFIG-RESOURCE-MEMORY">19.4.1. Memory</a></span></dt><dt><span class="sect2"><a href="runtime-config-resource.html#RUNTIME-CONFIG-RESOURCE-DISK">19.4.2. Disk</a></span></dt><dt><span class="sect2"><a href="runtime-config-resource.html#RUNTIME-CONFIG-RESOURCE-KERNEL">19.4.3. Kernel Resource Usage</a></span></dt><dt><span class="sect2"><a href="runtime-config-resource.html#RUNTIME-CONFIG-RESOURCE-VACUUM-COST">19.4.4. Cost-based Vacuum Delay</a></span></dt><dt><span class="sect2"><a href="runtime-config-resource.html#RUNTIME-CONFIG-RESOURCE-BACKGROUND-WRITER">19.4.5. Background Writer</a></span></dt><dt><span class="sect2"><a href="runtime-config-resource.html#RUNTIME-CONFIG-RESOURCE-ASYNC-BEHAVIOR">19.4.6. Asynchronous Behavior</a></span></dt></dl></div><div class="sect2" id="RUNTIME-CONFIG-RESOURCE-MEMORY"><div class="titlepage"><div><div><h3 class="title">19.4.1. Memory</h3></div></div></div><div class="variablelist"><dl class="variablelist"><dt id="GUC-SHARED-BUFFERS"><span class="term"><code class="varname">shared_buffers</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.2.2.1.1.3" class="indexterm"></a>
</span></dt><dd><p>
Sets the amount of memory the database server uses for shared
memory buffers. The default is typically 128 megabytes
(<code class="literal">128MB</code>), but might be less if your kernel settings will
not support it (as determined during <span class="application">initdb</span>).
This setting must be at least 128 kilobytes. (Non-default
values of <code class="symbol">BLCKSZ</code> change the minimum.) However,
settings significantly higher than the minimum are usually needed
for good performance. This parameter can only be set at server start.
</p><p>
If you have a dedicated database server with 1GB or more of RAM, a
reasonable starting value for <code class="varname">shared_buffers</code> is 25%
of the memory in your system. There are some workloads where even
larger settings for <code class="varname">shared_buffers</code> are effective, but
because <span class="productname">PostgreSQL</span> also relies on the
operating system cache, it is unlikely that an allocation of more than
40% of RAM to <code class="varname">shared_buffers</code> will work better than a
smaller amount. Larger settings for <code class="varname">shared_buffers</code>
usually require a corresponding increase in
<code class="varname">max_wal_size</code>, in order to spread out the
process of writing large quantities of new or changed data over a
longer period of time.
</p><p>
On systems with less than 1GB of RAM, a smaller percentage of RAM is
appropriate, so as to leave adequate space for the operating system.
</p></dd><dt id="GUC-HUGE-PAGES"><span class="term"><code class="varname">huge_pages</code> (<code class="type">enum</code>)
<a id="id-1.6.6.7.2.2.2.1.3" class="indexterm"></a>
</span></dt><dd><p>
Controls whether huge pages are requested for the main shared memory
area. Valid values are <code class="literal">try</code> (the default),
<code class="literal">on</code>, and <code class="literal">off</code>. With
<code class="varname">huge_pages</code> set to <code class="literal">try</code>, the
server will try to request huge pages, but fall back to the default if
that fails. With <code class="literal">on</code>, failure to request huge pages
will prevent the server from starting up. With <code class="literal">off</code>,
huge pages will not be requested.
</p><p>
At present, this setting is supported only on Linux and Windows. The
setting is ignored on other systems when set to
<code class="literal">try</code>.
</p><p>
The use of huge pages results in smaller page tables and less CPU time
spent on memory management, increasing performance. For more details about
using huge pages on Linux, see <a class="xref" href="kernel-resources.html#LINUX-HUGE-PAGES" title="18.4.5. Linux Huge Pages">Section 18.4.5</a>.
</p><p>
Huge pages are known as large pages on Windows. To use them, you need to
assign the user right Lock Pages in Memory to the Windows user account
that runs <span class="productname">PostgreSQL</span>.
You can use Windows Group Policy tool (gpedit.msc) to assign the user right
Lock Pages in Memory.
To start the database server on the command prompt as a standalone process,
not as a Windows service, the command prompt must be run as an administrator or
User Access Control (UAC) must be disabled. When the UAC is enabled, the normal
command prompt revokes the user right Lock Pages in Memory when started.
</p><p>
Note that this setting only affects the main shared memory area.
Operating systems such as Linux, FreeBSD, and Illumos can also use
huge pages (also known as <span class="quote">“<span class="quote">super</span>”</span> pages or
<span class="quote">“<span class="quote">large</span>”</span> pages) automatically for normal memory
allocation, without an explicit request from
<span class="productname">PostgreSQL</span>. On Linux, this is called
<span class="quote">“<span class="quote">transparent huge pages</span>”</span><a id="id-1.6.6.7.2.2.2.2.5.5" class="indexterm"></a> (THP). That feature has been known to
cause performance degradation with
<span class="productname">PostgreSQL</span> for some users on some Linux
versions, so its use is currently discouraged (unlike explicit use of
<code class="varname">huge_pages</code>).
</p></dd><dt id="GUC-TEMP-BUFFERS"><span class="term"><code class="varname">temp_buffers</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.2.2.3.1.3" class="indexterm"></a>
</span></dt><dd><p>
Sets the maximum number of temporary buffers used by each database
session. These are session-local buffers used only for access to
temporary tables. The default is eight megabytes
(<code class="literal">8MB</code>). The setting can be changed within individual
sessions, but only before the first use of temporary tables
within the session; subsequent attempts to change the value will
have no effect on that session.
</p><p>
A session will allocate temporary buffers as needed up to the limit
given by <code class="varname">temp_buffers</code>. The cost of setting a large
value in sessions that do not actually need many temporary
buffers is only a buffer descriptor, or about 64 bytes, per
increment in <code class="varname">temp_buffers</code>. However if a buffer is
actually used an additional 8192 bytes will be consumed for it
(or in general, <code class="symbol">BLCKSZ</code> bytes).
</p></dd><dt id="GUC-MAX-PREPARED-TRANSACTIONS"><span class="term"><code class="varname">max_prepared_transactions</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.2.2.4.1.3" class="indexterm"></a>
</span></dt><dd><p>
Sets the maximum number of transactions that can be in the
<span class="quote">“<span class="quote">prepared</span>”</span> state simultaneously (see <a class="xref" href="sql-prepare-transaction.html" title="PREPARE TRANSACTION"><span class="refentrytitle">PREPARE TRANSACTION</span></a>).
Setting this parameter to zero (which is the default)
disables the prepared-transaction feature.
This parameter can only be set at server start.
</p><p>
If you are not planning to use prepared transactions, this parameter
should be set to zero to prevent accidental creation of prepared
transactions. If you are using prepared transactions, you will
probably want <code class="varname">max_prepared_transactions</code> to be at
least as large as <a class="xref" href="runtime-config-connection.html#GUC-MAX-CONNECTIONS">max_connections</a>, so that every
session can have a prepared transaction pending.
</p><p>
When running a standby server, you must set this parameter to the
same or higher value than on the master server. Otherwise, queries
will not be allowed in the standby server.
</p></dd><dt id="GUC-WORK-MEM"><span class="term"><code class="varname">work_mem</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.2.2.5.1.3" class="indexterm"></a>
</span></dt><dd><p>
Specifies the amount of memory to be used by internal sort operations
and hash tables before writing to temporary disk files. The value
defaults to four megabytes (<code class="literal">4MB</code>).
Note that for a complex query, several sort or hash operations might be
running in parallel; each operation will be allowed to use as much memory
as this value specifies before it starts to write data into temporary
files. Also, several running sessions could be doing such operations
concurrently. Therefore, the total memory used could be many
times the value of <code class="varname">work_mem</code>; it is necessary to
keep this fact in mind when choosing the value. Sort operations are
used for <code class="literal">ORDER BY</code>, <code class="literal">DISTINCT</code>, and
merge joins.
Hash tables are used in hash joins, hash-based aggregation, and
hash-based processing of <code class="literal">IN</code> subqueries.
</p></dd><dt id="GUC-MAINTENANCE-WORK-MEM"><span class="term"><code class="varname">maintenance_work_mem</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.2.2.6.1.3" class="indexterm"></a>
</span></dt><dd><p>
Specifies the maximum amount of memory to be used by maintenance
operations, such as <code class="command">VACUUM</code>, <code class="command">CREATE
INDEX</code>, and <code class="command">ALTER TABLE ADD FOREIGN KEY</code>. It defaults
to 64 megabytes (<code class="literal">64MB</code>). Since only one of these
operations can be executed at a time by a database session, and
an installation normally doesn't have many of them running
concurrently, it's safe to set this value significantly larger
than <code class="varname">work_mem</code>. Larger settings might improve
performance for vacuuming and for restoring database dumps.
</p><p>
Note that when autovacuum runs, up to
<a class="xref" href="runtime-config-autovacuum.html#GUC-AUTOVACUUM-MAX-WORKERS">autovacuum_max_workers</a> times this memory
may be allocated, so be careful not to set the default value
too high. It may be useful to control for this by separately
setting <a class="xref" href="runtime-config-resource.html#GUC-AUTOVACUUM-WORK-MEM">autovacuum_work_mem</a>.
</p></dd><dt id="GUC-AUTOVACUUM-WORK-MEM"><span class="term"><code class="varname">autovacuum_work_mem</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.2.2.7.1.3" class="indexterm"></a>
</span></dt><dd><p>
Specifies the maximum amount of memory to be used by each
autovacuum worker process. It defaults to -1, indicating that
the value of <a class="xref" href="runtime-config-resource.html#GUC-MAINTENANCE-WORK-MEM">maintenance_work_mem</a> should
be used instead. The setting has no effect on the behavior of
<code class="command">VACUUM</code> when run in other contexts.
</p></dd><dt id="GUC-MAX-STACK-DEPTH"><span class="term"><code class="varname">max_stack_depth</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.2.2.8.1.3" class="indexterm"></a>
</span></dt><dd><p>
Specifies the maximum safe depth of the server's execution stack.
The ideal setting for this parameter is the actual stack size limit
enforced by the kernel (as set by <code class="literal">ulimit -s</code> or local
equivalent), less a safety margin of a megabyte or so. The safety
margin is needed because the stack depth is not checked in every
routine in the server, but only in key potentially-recursive routines
such as expression evaluation. The default setting is two
megabytes (<code class="literal">2MB</code>), which is conservatively small and
unlikely to risk crashes. However, it might be too small to allow
execution of complex functions. Only superusers can change this
setting.
</p><p>
Setting <code class="varname">max_stack_depth</code> higher than
the actual kernel limit will mean that a runaway recursive function
can crash an individual backend process. On platforms where
<span class="productname">PostgreSQL</span> can determine the kernel limit,
the server will not allow this variable to be set to an unsafe
value. However, not all platforms provide the information,
so caution is recommended in selecting a value.
</p></dd><dt id="GUC-DYNAMIC-SHARED-MEMORY-TYPE"><span class="term"><code class="varname">dynamic_shared_memory_type</code> (<code class="type">enum</code>)
<a id="id-1.6.6.7.2.2.9.1.3" class="indexterm"></a>
</span></dt><dd><p>
Specifies the dynamic shared memory implementation that the server
should use. Possible values are <code class="literal">posix</code> (for POSIX shared
memory allocated using <code class="literal">shm_open</code>), <code class="literal">sysv</code>
(for System V shared memory allocated via <code class="literal">shmget</code>),
<code class="literal">windows</code> (for Windows shared memory), <code class="literal">mmap</code>
(to simulate shared memory using memory-mapped files stored in the
data directory), and <code class="literal">none</code> (to disable this feature).
Not all values are supported on all platforms; the first supported
option is the default for that platform. The use of the
<code class="literal">mmap</code> option, which is not the default on any platform,
is generally discouraged because the operating system may write
modified pages back to disk repeatedly, increasing system I/O load;
however, it may be useful for debugging, when the
<code class="literal">pg_dynshmem</code> directory is stored on a RAM disk, or when
other shared memory facilities are not available.
</p></dd></dl></div></div><div class="sect2" id="RUNTIME-CONFIG-RESOURCE-DISK"><div class="titlepage"><div><div><h3 class="title">19.4.2. Disk</h3></div></div></div><div class="variablelist"><dl class="variablelist"><dt id="GUC-TEMP-FILE-LIMIT"><span class="term"><code class="varname">temp_file_limit</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.3.2.1.1.3" class="indexterm"></a>
</span></dt><dd><p>
Specifies the maximum amount of disk space that a process can use
for temporary files, such as sort and hash temporary files, or the
storage file for a held cursor. A transaction attempting to exceed
this limit will be canceled.
The value is specified in kilobytes, and <code class="literal">-1</code> (the
default) means no limit.
Only superusers can change this setting.
</p><p>
This setting constrains the total space used at any instant by all
temporary files used by a given <span class="productname">PostgreSQL</span> process.
It should be noted that disk space used for explicit temporary
tables, as opposed to temporary files used behind-the-scenes in query
execution, does <span class="emphasis"><em>not</em></span> count against this limit.
</p></dd></dl></div></div><div class="sect2" id="RUNTIME-CONFIG-RESOURCE-KERNEL"><div class="titlepage"><div><div><h3 class="title">19.4.3. Kernel Resource Usage</h3></div></div></div><div class="variablelist"><dl class="variablelist"><dt id="GUC-MAX-FILES-PER-PROCESS"><span class="term"><code class="varname">max_files_per_process</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.4.2.1.1.3" class="indexterm"></a>
</span></dt><dd><p>
Sets the maximum number of simultaneously open files allowed to each
server subprocess. The default is one thousand files. If the kernel is enforcing
a safe per-process limit, you don't need to worry about this setting.
But on some platforms (notably, most BSD systems), the kernel will
allow individual processes to open many more files than the system
can actually support if many processes all try to open
that many files. If you find yourself seeing <span class="quote">“<span class="quote">Too many open
files</span>”</span> failures, try reducing this setting.
This parameter can only be set at server start.
</p></dd></dl></div></div><div class="sect2" id="RUNTIME-CONFIG-RESOURCE-VACUUM-COST"><div class="titlepage"><div><div><h3 class="title">19.4.4. Cost-based Vacuum Delay</h3></div></div></div><p>
During the execution of <a class="xref" href="sql-vacuum.html" title="VACUUM"><span class="refentrytitle">VACUUM</span></a>
and <a class="xref" href="sql-analyze.html" title="ANALYZE"><span class="refentrytitle">ANALYZE</span></a>
commands, the system maintains an
internal counter that keeps track of the estimated cost of the
various I/O operations that are performed. When the accumulated
cost reaches a limit (specified by
<code class="varname">vacuum_cost_limit</code>), the process performing
the operation will sleep for a short period of time, as specified by
<code class="varname">vacuum_cost_delay</code>. Then it will reset the
counter and continue execution.
</p><p>
The intent of this feature is to allow administrators to reduce
the I/O impact of these commands on concurrent database
activity. There are many situations where it is not
important that maintenance commands like
<code class="command">VACUUM</code> and <code class="command">ANALYZE</code> finish
quickly; however, it is usually very important that these
commands do not significantly interfere with the ability of the
system to perform other database operations. Cost-based vacuum
delay provides a way for administrators to achieve this.
</p><p>
This feature is disabled by default for manually issued
<code class="command">VACUUM</code> commands. To enable it, set the
<code class="varname">vacuum_cost_delay</code> variable to a nonzero
value.
</p><div class="variablelist"><dl class="variablelist"><dt id="GUC-VACUUM-COST-DELAY"><span class="term"><code class="varname">vacuum_cost_delay</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.5.5.1.1.3" class="indexterm"></a>
</span></dt><dd><p>
The length of time, in milliseconds, that the process will sleep
when the cost limit has been exceeded.
The default value is zero, which disables the cost-based vacuum
delay feature. Positive values enable cost-based vacuuming.
Note that on many systems, the effective resolution
of sleep delays is 10 milliseconds; setting
<code class="varname">vacuum_cost_delay</code> to a value that is
not a multiple of 10 might have the same results as setting it
to the next higher multiple of 10.
</p><p>
When using cost-based vacuuming, appropriate values for
<code class="varname">vacuum_cost_delay</code> are usually quite small, perhaps
10 or 20 milliseconds. Adjusting vacuum's resource consumption
is best done by changing the other vacuum cost parameters.
</p></dd><dt id="GUC-VACUUM-COST-PAGE-HIT"><span class="term"><code class="varname">vacuum_cost_page_hit</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.5.5.2.1.3" class="indexterm"></a>
</span></dt><dd><p>
The estimated cost for vacuuming a buffer found in the shared buffer
cache. It represents the cost to lock the buffer pool, lookup
the shared hash table and scan the content of the page. The
default value is one.
</p></dd><dt id="GUC-VACUUM-COST-PAGE-MISS"><span class="term"><code class="varname">vacuum_cost_page_miss</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.5.5.3.1.3" class="indexterm"></a>
</span></dt><dd><p>
The estimated cost for vacuuming a buffer that has to be read from
disk. This represents the effort to lock the buffer pool,
lookup the shared hash table, read the desired block in from
the disk and scan its content. The default value is 10.
</p></dd><dt id="GUC-VACUUM-COST-PAGE-DIRTY"><span class="term"><code class="varname">vacuum_cost_page_dirty</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.5.5.4.1.3" class="indexterm"></a>
</span></dt><dd><p>
The estimated cost charged when vacuum modifies a block that was
previously clean. It represents the extra I/O required to
flush the dirty block out to disk again. The default value is
20.
</p></dd><dt id="GUC-VACUUM-COST-LIMIT"><span class="term"><code class="varname">vacuum_cost_limit</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.5.5.5.1.3" class="indexterm"></a>
</span></dt><dd><p>
The accumulated cost that will cause the vacuuming process to sleep.
The default value is 200.
</p></dd></dl></div><div class="note"><h3 class="title">Note</h3><p>
There are certain operations that hold critical locks and should
therefore complete as quickly as possible. Cost-based vacuum
delays do not occur during such operations. Therefore it is
possible that the cost accumulates far higher than the specified
limit. To avoid uselessly long delays in such cases, the actual
delay is calculated as <code class="varname">vacuum_cost_delay</code> *
<code class="varname">accumulated_balance</code> /
<code class="varname">vacuum_cost_limit</code> with a maximum of
<code class="varname">vacuum_cost_delay</code> * 4.
</p></div></div><div class="sect2" id="RUNTIME-CONFIG-RESOURCE-BACKGROUND-WRITER"><div class="titlepage"><div><div><h3 class="title">19.4.5. Background Writer</h3></div></div></div><p>
There is a separate server
process called the <em class="firstterm">background writer</em>, whose function
is to issue writes of <span class="quote">“<span class="quote">dirty</span>”</span> (new or modified) shared
buffers. It writes shared buffers so server processes handling
user queries seldom or never need to wait for a write to occur.
However, the background writer does cause a net overall
increase in I/O load, because while a repeatedly-dirtied page might
otherwise be written only once per checkpoint interval, the
background writer might write it several times as it is dirtied
in the same interval. The parameters discussed in this subsection
can be used to tune the behavior for local needs.
</p><div class="variablelist"><dl class="variablelist"><dt id="GUC-BGWRITER-DELAY"><span class="term"><code class="varname">bgwriter_delay</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.6.3.1.1.3" class="indexterm"></a>
</span></dt><dd><p>
Specifies the delay between activity rounds for the
background writer. In each round the writer issues writes
for some number of dirty buffers (controllable by the
following parameters). It then sleeps for <code class="varname">bgwriter_delay</code>
milliseconds, and repeats. When there are no dirty buffers in the
buffer pool, though, it goes into a longer sleep regardless of
<code class="varname">bgwriter_delay</code>. The default value is 200
milliseconds (<code class="literal">200ms</code>). Note that on many systems, the
effective resolution of sleep delays is 10 milliseconds; setting
<code class="varname">bgwriter_delay</code> to a value that is not a multiple of 10
might have the same results as setting it to the next higher multiple
of 10. This parameter can only be set in the
<code class="filename">postgresql.conf</code> file or on the server command line.
</p></dd><dt id="GUC-BGWRITER-LRU-MAXPAGES"><span class="term"><code class="varname">bgwriter_lru_maxpages</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.6.3.2.1.3" class="indexterm"></a>
</span></dt><dd><p>
In each round, no more than this many buffers will be written
by the background writer. Setting this to zero disables
background writing. (Note that checkpoints, which are managed by
a separate, dedicated auxiliary process, are unaffected.)
The default value is 100 buffers.
This parameter can only be set in the <code class="filename">postgresql.conf</code>
file or on the server command line.
</p></dd><dt id="GUC-BGWRITER-LRU-MULTIPLIER"><span class="term"><code class="varname">bgwriter_lru_multiplier</code> (<code class="type">floating point</code>)
<a id="id-1.6.6.7.6.3.3.1.3" class="indexterm"></a>
</span></dt><dd><p>
The number of dirty buffers written in each round is based on the
number of new buffers that have been needed by server processes
during recent rounds. The average recent need is multiplied by
<code class="varname">bgwriter_lru_multiplier</code> to arrive at an estimate of the
number of buffers that will be needed during the next round. Dirty
buffers are written until there are that many clean, reusable buffers
available. (However, no more than <code class="varname">bgwriter_lru_maxpages</code>
buffers will be written per round.)
Thus, a setting of 1.0 represents a <span class="quote">“<span class="quote">just in time</span>”</span> policy
of writing exactly the number of buffers predicted to be needed.
Larger values provide some cushion against spikes in demand,
while smaller values intentionally leave writes to be done by
server processes.
The default is 2.0.
This parameter can only be set in the <code class="filename">postgresql.conf</code>
file or on the server command line.
</p></dd><dt id="GUC-BGWRITER-FLUSH-AFTER"><span class="term"><code class="varname">bgwriter_flush_after</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.6.3.4.1.3" class="indexterm"></a>
</span></dt><dd><p>
Whenever more than <code class="varname">bgwriter_flush_after</code> bytes have
been written by the background writer, attempt to force the OS to issue these
writes to the underlying storage. Doing so will limit the amount of
dirty data in the kernel's page cache, reducing the likelihood of
stalls when an <code class="function">fsync</code> is issued at the end of a checkpoint, or when
the OS writes data back in larger batches in the background. Often
that will result in greatly reduced transaction latency, but there
also are some cases, especially with workloads that are bigger than
<a class="xref" href="runtime-config-resource.html#GUC-SHARED-BUFFERS">shared_buffers</a>, but smaller than the OS's page
cache, where performance might degrade. This setting may have no
effect on some platforms. The valid range is between
<code class="literal">0</code>, which disables forced writeback, and
<code class="literal">2MB</code>. The default is <code class="literal">512kB</code> on Linux,
<code class="literal">0</code> elsewhere. (If <code class="symbol">BLCKSZ</code> is not 8kB,
the default and maximum values scale proportionally to it.)
This parameter can only be set in the <code class="filename">postgresql.conf</code>
file or on the server command line.
</p></dd></dl></div><p>
Smaller values of <code class="varname">bgwriter_lru_maxpages</code> and
<code class="varname">bgwriter_lru_multiplier</code> reduce the extra I/O load
caused by the background writer, but make it more likely that server
processes will have to issue writes for themselves, delaying interactive
queries.
</p></div><div class="sect2" id="RUNTIME-CONFIG-RESOURCE-ASYNC-BEHAVIOR"><div class="titlepage"><div><div><h3 class="title">19.4.6. Asynchronous Behavior</h3></div></div></div><div class="variablelist"><dl class="variablelist"><dt id="GUC-EFFECTIVE-IO-CONCURRENCY"><span class="term"><code class="varname">effective_io_concurrency</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.7.2.1.1.3" class="indexterm"></a>
</span></dt><dd><p>
Sets the number of concurrent disk I/O operations that
<span class="productname">PostgreSQL</span> expects can be executed
simultaneously. Raising this value will increase the number of I/O
operations that any individual <span class="productname">PostgreSQL</span> session
attempts to initiate in parallel. The allowed range is 1 to 1000,
or zero to disable issuance of asynchronous I/O requests. Currently,
this setting only affects bitmap heap scans.
</p><p>
For magnetic drives, a good starting point for this setting is the
number of separate
drives comprising a RAID 0 stripe or RAID 1 mirror being used for the
database. (For RAID 5 the parity drive should not be counted.)
However, if the database is often busy with multiple queries issued in
concurrent sessions, lower values may be sufficient to keep the disk
array busy. A value higher than needed to keep the disks busy will
only result in extra CPU overhead.
SSDs and other memory-based storage can often process many
concurrent requests, so the best value might be in the hundreds.
</p><p>
Asynchronous I/O depends on an effective <code class="function">posix_fadvise</code>
function, which some operating systems lack. If the function is not
present then setting this parameter to anything but zero will result
in an error. On some operating systems (e.g., Solaris), the function
is present but does not actually do anything.
</p><p>
The default is 1 on supported systems, otherwise 0. This value can
be overridden for tables in a particular tablespace by setting the
tablespace parameter of the same name (see
<a class="xref" href="sql-altertablespace.html" title="ALTER TABLESPACE"><span class="refentrytitle">ALTER TABLESPACE</span></a>).
</p></dd><dt id="GUC-MAX-WORKER-PROCESSES"><span class="term"><code class="varname">max_worker_processes</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.7.2.2.1.3" class="indexterm"></a>
</span></dt><dd><p>
Sets the maximum number of background processes that the system
can support. This parameter can only be set at server start. The
default is 8.
</p><p>
When running a standby server, you must set this parameter to the
same or higher value than on the master server. Otherwise, queries
will not be allowed in the standby server.
</p><p>
When changing this value, consider also adjusting
<a class="xref" href="runtime-config-resource.html#GUC-MAX-PARALLEL-WORKERS">max_parallel_workers</a>,
<a class="xref" href="runtime-config-resource.html#GUC-MAX-PARALLEL-WORKERS-MAINTENANCE">max_parallel_maintenance_workers</a>, and
<a class="xref" href="runtime-config-resource.html#GUC-MAX-PARALLEL-WORKERS-PER-GATHER">max_parallel_workers_per_gather</a>.
</p></dd><dt id="GUC-MAX-PARALLEL-WORKERS-PER-GATHER"><span class="term"><code class="varname">max_parallel_workers_per_gather</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.7.2.3.1.3" class="indexterm"></a>
</span></dt><dd><p>
Sets the maximum number of workers that can be started by a single
<code class="literal">Gather</code> or <code class="literal">Gather Merge</code> node.
Parallel workers are taken from the pool of processes established by
<a class="xref" href="runtime-config-resource.html#GUC-MAX-WORKER-PROCESSES">max_worker_processes</a>, limited by
<a class="xref" href="runtime-config-resource.html#GUC-MAX-PARALLEL-WORKERS">max_parallel_workers</a>. Note that the requested
number of workers may not actually be available at run time. If this
occurs, the plan will run with fewer workers than expected, which may
be inefficient. The default value is 2. Setting this value to 0
disables parallel query execution.
</p><p>
Note that parallel queries may consume very substantially more
resources than non-parallel queries, because each worker process is
a completely separate process which has roughly the same impact on the
system as an additional user session. This should be taken into
account when choosing a value for this setting, as well as when
configuring other settings that control resource utilization, such
as <a class="xref" href="runtime-config-resource.html#GUC-WORK-MEM">work_mem</a>. Resource limits such as
<code class="varname">work_mem</code> are applied individually to each worker,
which means the total utilization may be much higher across all
processes than it would normally be for any single process.
For example, a parallel query using 4 workers may use up to 5 times
as much CPU time, memory, I/O bandwidth, and so forth as a query which
uses no workers at all.
</p><p>
For more information on parallel query, see
<a class="xref" href="parallel-query.html" title="Chapter 15. Parallel Query">Chapter 15</a>.
</p></dd><dt id="GUC-MAX-PARALLEL-WORKERS-MAINTENANCE"><span class="term"><code class="varname">max_parallel_maintenance_workers</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.7.2.4.1.3" class="indexterm"></a>
</span></dt><dd><p>
Sets the maximum number of parallel workers that can be
started by a single utility command. Currently, the only
parallel utility command that supports the use of parallel
workers is <code class="command">CREATE INDEX</code>, and only when
building a B-tree index. Parallel workers are taken from the
pool of processes established by <a class="xref" href="runtime-config-resource.html#GUC-MAX-WORKER-PROCESSES">max_worker_processes</a>, limited by <a class="xref" href="runtime-config-resource.html#GUC-MAX-PARALLEL-WORKERS">max_parallel_workers</a>. Note that the requested
number of workers may not actually be available at run time.
If this occurs, the utility operation will run with fewer
workers than expected. The default value is 2. Setting this
value to 0 disables the use of parallel workers by utility
commands.
</p><p>
Note that parallel utility commands should not consume
substantially more memory than equivalent non-parallel
operations. This strategy differs from that of parallel
query, where resource limits generally apply per worker
process. Parallel utility commands treat the resource limit
<code class="varname">maintenance_work_mem</code> as a limit to be applied to
the entire utility command, regardless of the number of
parallel worker processes. However, parallel utility
commands may still consume substantially more CPU resources
and I/O bandwidth.
</p></dd><dt id="GUC-MAX-PARALLEL-WORKERS"><span class="term"><code class="varname">max_parallel_workers</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.7.2.5.1.3" class="indexterm"></a>
</span></dt><dd><p>
Sets the maximum number of workers that the system can support for
parallel operations. The default value is 8. When increasing or
decreasing this value, consider also adjusting
<a class="xref" href="runtime-config-resource.html#GUC-MAX-PARALLEL-WORKERS-MAINTENANCE">max_parallel_maintenance_workers</a> and
<a class="xref" href="runtime-config-resource.html#GUC-MAX-PARALLEL-WORKERS-PER-GATHER">max_parallel_workers_per_gather</a>.
Also, note that a setting for this value which is higher than
<a class="xref" href="runtime-config-resource.html#GUC-MAX-WORKER-PROCESSES">max_worker_processes</a> will have no effect,
since parallel workers are taken from the pool of worker processes
established by that setting.
</p></dd><dt id="GUC-BACKEND-FLUSH-AFTER"><span class="term"><code class="varname">backend_flush_after</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.7.2.6.1.3" class="indexterm"></a>
</span></dt><dd><p>
Whenever more than <code class="varname">backend_flush_after</code> bytes have
been written by a single backend, attempt to force the OS to issue
these writes to the underlying storage. Doing so will limit the
amount of dirty data in the kernel's page cache, reducing the
likelihood of stalls when an <code class="function">fsync</code> is issued at the end of a
checkpoint, or when the OS writes data back in larger batches in the
background. Often that will result in greatly reduced transaction
latency, but there also are some cases, especially with workloads
that are bigger than <a class="xref" href="runtime-config-resource.html#GUC-SHARED-BUFFERS">shared_buffers</a>, but smaller
than the OS's page cache, where performance might degrade. This
setting may have no effect on some platforms. The valid range is
between <code class="literal">0</code>, which disables forced writeback,
and <code class="literal">2MB</code>. The default is <code class="literal">0</code>, i.e., no
forced writeback. (If <code class="symbol">BLCKSZ</code> is not 8kB,
the maximum value scales proportionally to it.)
</p></dd><dt id="GUC-OLD-SNAPSHOT-THRESHOLD"><span class="term"><code class="varname">old_snapshot_threshold</code> (<code class="type">integer</code>)
<a id="id-1.6.6.7.7.2.7.1.3" class="indexterm"></a>
</span></dt><dd><p>
Sets the minimum time that a snapshot can be used without risk of a
<code class="literal">snapshot too old</code> error occurring when using the snapshot.
This parameter can only be set at server start.
</p><p>
Beyond the threshold, old data may be vacuumed away. This can help
prevent bloat in the face of snapshots which remain in use for a
long time. To prevent incorrect results due to cleanup of data which
would otherwise be visible to the snapshot, an error is generated
when the snapshot is older than this threshold and the snapshot is
used to read a page which has been modified since the snapshot was
built.
</p><p>
A value of <code class="literal">-1</code> disables this feature, and is the default.
Useful values for production work probably range from a small number
of hours to a few days. The setting will be coerced to a granularity
of minutes, and small numbers (such as <code class="literal">0</code> or
<code class="literal">1min</code>) are only allowed because they may sometimes be
useful for testing. While a setting as high as <code class="literal">60d</code> is
allowed, please note that in many workloads extreme bloat or
transaction ID wraparound may occur in much shorter time frames.
</p><p>
When this feature is enabled, freed space at the end of a relation
cannot be released to the operating system, since that could remove
information needed to detect the <code class="literal">snapshot too old</code>
condition. All space allocated to a relation remains associated with
that relation for reuse only within that relation unless explicitly
freed (for example, with <code class="command">VACUUM FULL</code>).
</p><p>
This setting does not attempt to guarantee that an error will be
generated under any particular circumstances. In fact, if the
correct results can be generated from (for example) a cursor which
has materialized a result set, no error will be generated even if the
underlying rows in the referenced table have been vacuumed away.
Some tables cannot safely be vacuumed early, and so will not be
affected by this setting, such as system catalogs. For such tables
this setting will neither reduce bloat nor create a possibility
of a <code class="literal">snapshot too old</code> error on scanning.
</p></dd></dl></div></div></div><div class="navfooter"><hr /><table width="100%" summary="Navigation footer"><tr><td width="40%" align="left"><a accesskey="p" href="runtime-config-connection.html">Prev</a> </td><td width="20%" align="center"><a accesskey="u" href="runtime-config.html">Up</a></td><td width="40%" align="right"> <a accesskey="n" href="runtime-config-wal.html">Next</a></td></tr><tr><td width="40%" align="left" valign="top">19.3. Connections and Authentication </td><td width="20%" align="center"><a accesskey="h" href="index.html">Home</a></td><td width="40%" align="right" valign="top"> 19.5. Write Ahead Log</td></tr></table></div></body></html>
