<?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>Chapter 48. Background Worker Processes</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="spi-spi-start-transaction.html" title="SPI_start_transaction" /><link rel="next" href="logicaldecoding.html" title="Chapter 49. Logical Decoding" /></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">Chapter 48. Background Worker Processes</th></tr><tr><td width="10%" align="left"><a accesskey="p" href="spi-spi-start-transaction.html" title="SPI_start_transaction">Prev</a> </td><td width="10%" align="left"><a accesskey="u" href="server-programming.html" title="Part V. Server Programming">Up</a></td><th width="60%" align="center">Part V. Server Programming</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="logicaldecoding.html" title="Chapter 49. Logical Decoding">Next</a></td></tr></table><hr></hr></div><div class="chapter" id="BGWORKER"><div class="titlepage"><div><div><h2 class="title">Chapter 48. Background Worker Processes</h2></div></div></div><a id="id-1.8.13.2" class="indexterm"></a><p>
PostgreSQL can be extended to run user-supplied code in separate processes.
Such processes are started, stopped and monitored by <code class="command">postgres</code>,
which permits them to have a lifetime closely linked to the server's status.
These processes have the option to attach to <span class="productname">PostgreSQL</span>'s
shared memory area and to connect to databases internally; they can also run
multiple transactions serially, just like a regular client-connected server
process. Also, by linking to <span class="application">libpq</span> they can connect to the
server and behave like a regular client application.
</p><div class="warning"><h3 class="title">Warning</h3><p>
There are considerable robustness and security risks in using background
worker processes because, being written in the <code class="literal">C</code> language,
they have unrestricted access to data. Administrators wishing to enable
modules that include background worker process should exercise extreme
caution. Only carefully audited modules should be permitted to run
background worker processes.
</p></div><p>
Background workers can be initialized at the time that
<span class="productname">PostgreSQL</span> is started by including the module name in
<code class="varname">shared_preload_libraries</code>. A module wishing to run a background
worker can register it by calling
<code class="function">RegisterBackgroundWorker(<code class="type">BackgroundWorker *worker</code>)</code>
from its <code class="function">_PG_init()</code>. Background workers can also be started
after the system is up and running by calling the function
<code class="function">RegisterDynamicBackgroundWorker(<code class="type">BackgroundWorker
*worker, BackgroundWorkerHandle **handle</code>)</code>. Unlike
<code class="function">RegisterBackgroundWorker</code>, which can only be called from within
the postmaster, <code class="function">RegisterDynamicBackgroundWorker</code> must be
called from a regular backend or another background worker.
</p><p>
The structure <code class="structname">BackgroundWorker</code> is defined thus:
</p><pre class="programlisting">
typedef void (*bgworker_main_type)(Datum main_arg);
typedef struct BackgroundWorker
{
char bgw_name[BGW_MAXLEN];
char bgw_type[BGW_MAXLEN];
int bgw_flags;
BgWorkerStartTime bgw_start_time;
int bgw_restart_time; /* in seconds, or BGW_NEVER_RESTART */
char bgw_library_name[BGW_MAXLEN];
char bgw_function_name[BGW_MAXLEN];
Datum bgw_main_arg;
char bgw_extra[BGW_EXTRALEN];
int bgw_notify_pid;
} BackgroundWorker;
</pre><p>
</p><p>
<code class="structfield">bgw_name</code> and <code class="structfield">bgw_type</code> are
strings to be used in log messages, process listings and similar contexts.
<code class="structfield">bgw_type</code> should be the same for all background
workers of the same type, so that it is possible to group such workers in a
process listing, for example. <code class="structfield">bgw_name</code> on the
other hand can contain additional information about the specific process.
(Typically, the string for <code class="structfield">bgw_name</code> will contain
the type somehow, but that is not strictly required.)
</p><p>
<code class="structfield">bgw_flags</code> is a bitwise-or'd bit mask indicating the
capabilities that the module wants. Possible values are:
</p><div class="variablelist"><dl class="variablelist"><dt><span class="term"><code class="literal">BGWORKER_SHMEM_ACCESS</code></span></dt><dd><p>
<a id="id-1.8.13.8.2.1.2.1.1" class="indexterm"></a>
Requests shared memory access. Workers without shared memory access
cannot access any of <span class="productname">PostgreSQL's</span> shared
data structures, such as heavyweight or lightweight locks, shared
buffers, or any custom data structures which the worker itself may
wish to create and use.
</p></dd><dt><span class="term"><code class="literal">BGWORKER_BACKEND_DATABASE_CONNECTION</code></span></dt><dd><p>
<a id="id-1.8.13.8.2.2.2.1.1" class="indexterm"></a>
Requests the ability to establish a database connection through which it
can later run transactions and queries. A background worker using
<code class="literal">BGWORKER_BACKEND_DATABASE_CONNECTION</code> to connect to a
database must also attach shared memory using
<code class="literal">BGWORKER_SHMEM_ACCESS</code>, or worker start-up will fail.
</p></dd></dl></div><p>
</p><p>
<code class="structfield">bgw_start_time</code> is the server state during which
<code class="command">postgres</code> should start the process; it can be one of
<code class="literal">BgWorkerStart_PostmasterStart</code> (start as soon as
<code class="command">postgres</code> itself has finished its own initialization; processes
requesting this are not eligible for database connections),
<code class="literal">BgWorkerStart_ConsistentState</code> (start as soon as a consistent state
has been reached in a hot standby, allowing processes to connect to
databases and run read-only queries), and
<code class="literal">BgWorkerStart_RecoveryFinished</code> (start as soon as the system has
entered normal read-write state). Note the last two values are equivalent
in a server that's not a hot standby. Note that this setting only indicates
when the processes are to be started; they do not stop when a different state
is reached.
</p><p>
<code class="structfield">bgw_restart_time</code> is the interval, in seconds, that
<code class="command">postgres</code> should wait before restarting the process, in
case it crashes. It can be any positive value,
or <code class="literal">BGW_NEVER_RESTART</code>, indicating not to restart the
process in case of a crash.
</p><p>
<code class="structfield">bgw_library_name</code> is the name of a library in
which the initial entry point for the background worker should be sought.
The named library will be dynamically loaded by the worker process and
<code class="structfield">bgw_function_name</code> will be used to identify the
function to be called. If loading a function from the core code, this must
be set to "postgres".
</p><p>
<code class="structfield">bgw_function_name</code> is the name of a function in
a dynamically loaded library which should be used as the initial entry point
for a new background worker.
</p><p>
<code class="structfield">bgw_main_arg</code> is the <code class="type">Datum</code> argument
to the background worker main function. This main function should take a
single argument of type <code class="type">Datum</code> and return <code class="type">void</code>.
<code class="structfield">bgw_main_arg</code> will be passed as the argument.
In addition, the global variable <code class="literal">MyBgworkerEntry</code>
points to a copy of the <code class="structname">BackgroundWorker</code> structure
passed at registration time; the worker may find it helpful to examine
this structure.
</p><p>
On Windows (and anywhere else where <code class="literal">EXEC_BACKEND</code> is
defined) or in dynamic background workers it is not safe to pass a
<code class="type">Datum</code> by reference, only by value. If an argument is required, it
is safest to pass an int32 or other small value and use that as an index
into an array allocated in shared memory. If a value like a <code class="type">cstring</code>
or <code class="type">text</code> is passed then the pointer won't be valid from the
new background worker process.
</p><p>
<code class="structfield">bgw_extra</code> can contain extra data to be passed
to the background worker. Unlike <code class="structfield">bgw_main_arg</code>, this data
is not passed as an argument to the worker's main function, but it can be
accessed via <code class="literal">MyBgworkerEntry</code>, as discussed above.
</p><p>
<code class="structfield">bgw_notify_pid</code> is the PID of a PostgreSQL
backend process to which the postmaster should send <code class="literal">SIGUSR1</code>
when the process is started or exits. It should be 0 for workers registered
at postmaster startup time, or when the backend registering the worker does
not wish to wait for the worker to start up. Otherwise, it should be
initialized to <code class="literal">MyProcPid</code>.
</p><p>Once running, the process can connect to a database by calling
<code class="function">BackgroundWorkerInitializeConnection(<em class="parameter"><code>char *dbname</code></em>, <em class="parameter"><code>char *username</code></em>, <em class="parameter"><code>uint32 flags</code></em>)</code> or
<code class="function">BackgroundWorkerInitializeConnectionByOid(<em class="parameter"><code>Oid dboid</code></em>, <em class="parameter"><code>Oid useroid</code></em>, <em class="parameter"><code>uint32 flags</code></em>)</code>.
This allows the process to run transactions and queries using the
<code class="literal">SPI</code> interface. If <code class="varname">dbname</code> is NULL or
<code class="varname">dboid</code> is <code class="literal">InvalidOid</code>, the session is not connected
to any particular database, but shared catalogs can be accessed.
If <code class="varname">username</code> is NULL or <code class="varname">useroid</code> is
<code class="literal">InvalidOid</code>, the process will run as the superuser created
during <code class="command">initdb</code>. If <code class="literal">BGWORKER_BYPASS_ALLOWCONN</code>
is specified as <code class="varname">flags</code> it is possible to bypass the restriction
to connect to databases not allowing user connections.
A background worker can only call one of these two functions, and only
once. It is not possible to switch databases.
</p><p>
Signals are initially blocked when control reaches the
background worker's main function, and must be unblocked by it; this is to
allow the process to customize its signal handlers, if necessary.
Signals can be unblocked in the new process by calling
<code class="function">BackgroundWorkerUnblockSignals</code> and blocked by calling
<code class="function">BackgroundWorkerBlockSignals</code>.
</p><p>
If <code class="structfield">bgw_restart_time</code> for a background worker is
configured as <code class="literal">BGW_NEVER_RESTART</code>, or if it exits with an exit
code of 0 or is terminated by <code class="function">TerminateBackgroundWorker</code>,
it will be automatically unregistered by the postmaster on exit.
Otherwise, it will be restarted after the time period configured via
<code class="structfield">bgw_restart_time</code>, or immediately if the postmaster
reinitializes the cluster due to a backend failure. Backends which need
to suspend execution only temporarily should use an interruptible sleep
rather than exiting; this can be achieved by calling
<code class="function">WaitLatch()</code>. Make sure the
<code class="literal">WL_POSTMASTER_DEATH</code> flag is set when calling that function, and
verify the return code for a prompt exit in the emergency case that
<code class="command">postgres</code> itself has terminated.
</p><p>
When a background worker is registered using the
<code class="function">RegisterDynamicBackgroundWorker</code> function, it is
possible for the backend performing the registration to obtain information
regarding the status of the worker. Backends wishing to do this should
pass the address of a <code class="type">BackgroundWorkerHandle *</code> as the second
argument to <code class="function">RegisterDynamicBackgroundWorker</code>. If the
worker is successfully registered, this pointer will be initialized with an
opaque handle that can subsequently be passed to
<code class="function">GetBackgroundWorkerPid(<em class="parameter"><code>BackgroundWorkerHandle *</code></em>, <em class="parameter"><code>pid_t *</code></em>)</code> or
<code class="function">TerminateBackgroundWorker(<em class="parameter"><code>BackgroundWorkerHandle *</code></em>)</code>.
<code class="function">GetBackgroundWorkerPid</code> can be used to poll the status of the
worker: a return value of <code class="literal">BGWH_NOT_YET_STARTED</code> indicates that
the worker has not yet been started by the postmaster;
<code class="literal">BGWH_STOPPED</code> indicates that it has been started but is
no longer running; and <code class="literal">BGWH_STARTED</code> indicates that it is
currently running. In this last case, the PID will also be returned via the
second argument.
<code class="function">TerminateBackgroundWorker</code> causes the postmaster to send
<code class="literal">SIGTERM</code> to the worker if it is running, and to unregister it
as soon as it is not.
</p><p>
In some cases, a process which registers a background worker may wish to
wait for the worker to start up. This can be accomplished by initializing
<code class="structfield">bgw_notify_pid</code> to <code class="literal">MyProcPid</code> and
then passing the <code class="type">BackgroundWorkerHandle *</code> obtained at
registration time to
<code class="function">WaitForBackgroundWorkerStartup(<em class="parameter"><code>BackgroundWorkerHandle
*handle</code></em>, <em class="parameter"><code>pid_t *</code></em>)</code> function.
This function will block until the postmaster has attempted to start the
background worker, or until the postmaster dies. If the background worker
is running, the return value will be <code class="literal">BGWH_STARTED</code>, and
the PID will be written to the provided address. Otherwise, the return
value will be <code class="literal">BGWH_STOPPED</code> or
<code class="literal">BGWH_POSTMASTER_DIED</code>.
</p><p>
A process can also wait for a background worker to shut down, by using the
<code class="function">WaitForBackgroundWorkerShutdown(<em class="parameter"><code>BackgroundWorkerHandle
*handle</code></em>)</code> function and passing the
<code class="type">BackgroundWorkerHandle *</code> obtained at registration. This
function will block until the background worker exits, or postmaster dies.
When the background worker exits, the return value is
<code class="literal">BGWH_STOPPED</code>, if postmaster dies it will return
<code class="literal">BGWH_POSTMASTER_DIED</code>.
</p><p>
If a background worker sends asynchronous notifications with the
<code class="command">NOTIFY</code> command via the Server Programming Interface
(<acronym class="acronym">SPI</acronym>), it should call
<code class="function">ProcessCompletedNotifies</code> explicitly after committing
the enclosing transaction so that any notifications can be delivered. If a
background worker registers to receive asynchronous notifications with
the <code class="command">LISTEN</code> through <acronym class="acronym">SPI</acronym>, the worker
will log those notifications, but there is no programmatic way for the
worker to intercept and respond to those notifications.
</p><p>
The <code class="filename">src/test/modules/worker_spi</code> module
contains a working example,
which demonstrates some useful techniques.
</p><p>
The maximum number of registered background workers is limited by
<a class="xref" href="runtime-config-resource.html#GUC-MAX-WORKER-PROCESSES">max_worker_processes</a>.
</p></div><div class="navfooter"><hr /><table width="100%" summary="Navigation footer"><tr><td width="40%" align="left"><a accesskey="p" href="spi-spi-start-transaction.html">Prev</a> </td><td width="20%" align="center"><a accesskey="u" href="server-programming.html">Up</a></td><td width="40%" align="right"> <a accesskey="n" href="logicaldecoding.html">Next</a></td></tr><tr><td width="40%" align="left" valign="top">SPI_start_transaction </td><td width="20%" align="center"><a accesskey="h" href="index.html">Home</a></td><td width="40%" align="right" valign="top"> Chapter 49. Logical Decoding</td></tr></table></div></body></html>
