<!-- doc/src/sgml/config.sgml -->
<sect1 id="runtime-config-connection">
- <title>Connections and Authentication</title>
+ <title>Connections and Authentication</title>
- <sect2 id="runtime-config-connection-settings">
- <title>Connection Settings</title>
+ <sect2 id="runtime-config-connection-settings">
+ <title>Connection Settings</title>
- <variablelist>
+ <variablelist>
- <varlistentry id="guc-listen-addresses" xreflabel="listen_addresses">
+ <varlistentry id="guc-listen-addresses" xreflabel="listen_addresses">
<term><varname>listen_addresses</varname> (<type>string</type>)
- <indexterm>
-
<primary><varname>listen_addresses</varname> configuration parameter</primary>
- </indexterm>
+ <indexterm>
+ <primary><varname>listen_addresses</varname> configuration parameter</primary>
+ </indexterm>
</term>
<listitem>
- <para>
- Specifies the hostname or IP address, on which <productname>Pgpool-II</>
- will accept TCP/IP connections. <literal>'*'</literal>
- accepts all incoming connections. <literal>''</literal>
- disables TCP/IP connections. Default
- is <literal>'localhost'</literal>. Connections via UNIX
- domain socket are always accepted.
- </para>
- <para>
- This parameter can only be set at server start.
- </para>
+ <para>
+ Specifies the hostname or IP address, on
+ which <productname>Pgpool-II</productname> will accept
+ TCP/IP connections. <literal>'*'</literal> accepts all
+ incoming connections. <literal>''</literal> disables
+ TCP/IP connections. Default
+ is <literal>'localhost'</literal>. Connections via UNIX
+ domain socket are always accepted.
+ </para>
+ <para>
+ This parameter can only be set at server start.
+ </para>
</listitem>
- </varlistentry>
+ </varlistentry>
- <varlistentry id="guc-port" xreflabel="port">
+ <varlistentry id="guc-port" xreflabel="port">
<term><varname>port</varname> (<type>integer</type>)
- <indexterm>
-
<primary><varname>port</varname> configuration parameter</primary>
- </indexterm>
+ <indexterm>
+ <primary><varname>port</varname> configuration parameter</primary>
+ </indexterm>
</term>
<listitem>
- The port number used
-
by <productname>Pgpool-II</productname> to listen for
- connections. Default is 9999.
- </para>
- This parameter can only be set at server start.
- </para>
+ <para>
+ The port number used
+ by <productname>Pgpool-II</productname> to listen for
+ connections. Default is 9999.
+ </para>
+ <para>
+ This parameter can only be set at server start.
+ </para>
</listitem>
- </varlistentry>
+ </varlistentry>
- <varlistentry id="guc-socket-dir" xreflabel="socket_dir">
+ <varlistentry id="guc-socket-dir" xreflabel="socket_dir">
<term><varname>socket_dir</varname> (<type>string</type>)
- <indexterm>
-
<primary><varname>socket_dir</varname> configuration parameter</primary>
- </indexterm>
+ <indexterm>
+ <primary><varname>socket_dir</varname> configuration parameter</primary>
+ </indexterm>
</term>
<listitem>
- <para>
- The directory where the UNIX domain socket accepting connections for
- <productname>Pgpool-II</productname> will be
- created. Default is <literal>/tmp</literal>. Be aware that this
- socket might be deleted by a cron job. We recommend to set this
- value to <literal>/var/run</literal> or such directory.
- </para>
- <para>
- This parameter can only be set at server start.
- </para>
+ <para>
+ The directory where the UNIX domain socket accepting connections for
+ <productname>Pgpool-II</productname> will be
+ created. Default is <literal>/tmp</literal>. Be aware that
+ this socket might be deleted by a cron job. We recommend
+ to set this value to <literal>/var/run</literal> or such
+ directory.
+ </para>
+ <para>
+ This parameter can only be set at server start.
+ </para>
</listitem>
- </varlistentry>
+ </varlistentry>
- <varlistentry id="guc-pcp-listen-addresses" xreflabel="pcp_listen_addresses">
+ <varlistentry id="guc-pcp-listen-addresses" xreflabel="pcp_listen_addresses">
<term><varname>pcp_listen_addresses</varname> (<type>string</type>)
- <indexterm>
-
<primary><varname>pcp_listen_addresses</varname> configuration parameter</primary>
- </indexterm>
+ <indexterm>
+ <primary><varname>pcp_listen_addresses</varname> configuration parameter</primary>
+ </indexterm>
</term>
<listitem>
- Specifies the hostname or IP address, on which pcp process
- will accept TCP/IP connections. <literal>*</literal>
- accepts all incoming connections. <literal>"" </literal>
- disables TCP/IP connections. Default
- is <literal>*</literal>. Connections via UNIX domain
- socket are always accepted.
- </para>
- This parameter can only be set at server start.
- </para>
+ <para>
+ Specifies the hostname or IP address, on which pcp process
+ will accept TCP/IP connections. <literal>*</literal>
+ accepts all incoming connections. <literal>"" </literal>
+ disables TCP/IP connections. Default
+ is <literal>*</literal>. Connections via UNIX domain
+ socket are always accepted.
+ </para>
+ <para>
+ This parameter can only be set at server start.
+ </para>
</listitem>
- </varlistentry>
+ </varlistentry>
- <varlistentry id="guc-pcp-port" xreflabel="pcp_port">
+ <varlistentry id="guc-pcp-port" xreflabel="pcp_port">
<term><varname>pcp_port</varname> (<type>integer</type>)
- <indexterm>
-
<primary><varname>pcp_port</varname> configuration parameter</primary>
- </indexterm>
+ <indexterm>
+ <primary><varname>pcp_port</varname> configuration parameter</primary>
+ </indexterm>
</term>
<listitem>
- The port number used by PCP
- process to listen for connections. Default is 9898.
- </para>
- This parameter can only be set at server start.
- </para>
+ <para>
+ The port number used by PCP
+ process to listen for connections. Default is 9898.
+ </para>
+ <para>
+ This parameter can only be set at server start.
+ </para>
</listitem>
- </varlistentry>
+ </varlistentry>
- <varlistentry id="guc-pcp-socket-dir" xreflabel="pcp_socket_dir">
+ <varlistentry id="guc-pcp-socket-dir" xreflabel="pcp_socket_dir">
<term><varname>pcp_socket_dir</varname> (<type>string</type>)
- <indexterm>
-
<primary><varname>pcp_socket_dir</varname> configuration parameter</primary>
- </indexterm>
+ <indexterm>
+ <primary><varname>pcp_socket_dir</varname> configuration parameter</primary>
+ </indexterm>
</term>
<listitem>
- <para>
- The directory where the UNIX domain socket accepting connections for
- PCP process will be
- created. Default is <literal>/tmp</literal>. Be aware that
- this socket might be deleted by a cron job. We recommend
- to set this value to <literal>/var/run</literal> or such
- directory.
- </para>
- <para>
- This parameter can only be set at server start.
- </para>
+ <para>
+ The directory where the UNIX domain socket accepting
+ connections for PCP process will be created. Default
+ is <literal>/tmp</literal>. Be aware that this socket
+ might be deleted by a cron job. We recommend to set this
+ value to <literal>/var/run</literal> or such directory.
+ </para>
+ <para>
+ This parameter can only be set at server start.
+ </para>
</listitem>
- </varlistentry>
+ </varlistentry>
- <varlistentry id="guc-num-init-children" xreflabel="num_init_children">
+ <varlistentry id="guc-num-init-children" xreflabel="num_init_children">
<term><varname>num_init_children</varname> (<type>integer</type>)
- <indexterm>
-
<primary><varname>num_init_children</varname> configuration parameter</primary>
- </indexterm>
+ <indexterm>
+ <primary><varname>num_init_children</varname> configuration parameter</primary>
+ </indexterm>
</term>
<listitem>
- <para>
- The number of
- preforked <productname>Pgpool-II</productname> server
- processes. Default is 32. num_init_children is also the
- concurrent connections limit
- to <productname>Pgpool-II</productname> from clients. If
- more than num_init_children clients try to connect to
- <productname>Pgpool-II</productname>, <emphasis>they are
- blocked (not rejected with an error, like <productname>PostgreSQL</>)
- until a connection to any <productname>Pgpool-II</productname>
- process is closed</emphasis>. Up to
- <xref linkend="guc-listen-backlog-multiplier">*
- num_init_children can be queued.
- </para>
- <para>
- The queue is inside the kernel called "listen queue". The
- length of the listen queue is called "backlog". There is
- an upper limit of the backlog in some systems, and if
- num_init_children*<xref linkend="guc-listen-backlog-multiplier">
- exceeds the number, you need to set the backlog higher.
- Otherwise, following problems may occur in heavy loaded systems:
- 1) connecting to <productname>Pgpool-II</productname> fails
- 2) connecting to <productname>Pgpool-II</productname> is
- getting slow because of retries in the kernel. You can
- check if the listen queue is actually overflowed by using
- "netstat -s" command. If you find something like:
- <programlisting>
-535 times the listen queue of a socket overflowed
- </programlisting>
- then the listen queue is definitely overflowed.
- You should increase the backlog in this case (you will be required a super user privilege).
- <programlisting>
-# sysctl net.core.somaxconn
-net.core.somaxconn = 128
-# sysctl -w net.core.somaxconn = 256
- </programlisting>
-You could add following to /etc/sysctl.conf instead.
- <programlisting>
-net.core.somaxconn = 256
- </programlisting>
- </para>
- <para>
- Number of connections to each <productname>PostgreSQL</> is roughly max_pool*num_init_children.
- </para>
-
- <para>
- However, canceling a query creates another
- connection to the backend; thus, a query cannot be canceled if
- all the connections are in use. If you want to ensure that queries can
- be canceled, set this value to twice the expected connections.
- </para>
- <para>
- In addition, <productname>PostgreSQL</productname> allows concurrent
- connections for non superusers up to max_connections -
- superuser_reserved_connections.
- </para>
- <para>
- In summary, max_pool, num_init_children, max_connections,
- superuser_reserved_connections must satisfy the following formula:
- <programlisting>
-max_pool*num_init_children <= (max_connections - superuser_reserved_connections) (no query canceling needed)
-max_pool*num_init_children*2 <= (max_connections - superuser_reserved_connections) (query canceling needed)
- </programlisting>
- </para>
- <para>
- This parameter can only be set at server start.
- </para>
+ <para>
+ The number of
+ preforked <productname>Pgpool-II</productname> server
+ processes. Default is 32. num_init_children is also the
+ concurrent connections limit
+ to <productname>Pgpool-II</productname> from clients. If
+ more than num_init_children clients try to connect to
+ <productname>Pgpool-II</productname>, <emphasis>they are
+ blocked (not rejected with an error,
+ like <productname>PostgreSQL</productname>) until a
+ connection to any <productname>Pgpool-II</productname>
+ process is closed</emphasis>. Up to
+ <xref linkend="guc-listen-backlog-multiplier">*
+ num_init_children can be queued.
+ </para>
+ <para>
+ The queue is inside the kernel called "listen queue". The
+ length of the listen queue is called "backlog". There is
+ an upper limit of the backlog in some systems, and if
+ num_init_children*<xref linkend="guc-listen-backlog-multiplier">
+ exceeds the number, you need to set the backlog higher.
+ Otherwise, following problems may occur in heavy loaded systems:
+ 1) connecting to <productname>Pgpool-II</productname> fails
+ 2) connecting to <productname>Pgpool-II</productname> is
+ getting slow because of retries in the kernel. You can
+ check if the listen queue is actually overflowed by using
+ "netstat -s" command. If you find something like:
+ <programlisting>
+ 535 times the listen queue of a socket overflowed
+ </programlisting>
+ then the listen queue is definitely overflowed. You
+ should increase the backlog in this case (you will be
+ required a super user privilege).
+ <programlisting>
+ # sysctl net.core.somaxconn
+ net.core.somaxconn = 128
+ # sysctl -w net.core.somaxconn = 256
+ </programlisting>
+ You could add following to /etc/sysctl.conf instead.
+ <programlisting>
+ net.core.somaxconn = 256
+ </programlisting>
+ </para>
+ <para>
+ Number of connections to
+ each <productname>PostgreSQL</productname> is roughly
+ max_pool*num_init_children.
+ </para>
+
+ <para>
+ However, canceling a query creates another
+ connection to the backend; thus, a query cannot be canceled if
+ all the connections are in use. If you want to ensure that queries can
+ be canceled, set this value to twice the expected connections.
+ </para>
+ <para>
+ In addition, <productname>PostgreSQL</productname> allows concurrent
+ connections for non superusers up to max_connections -
+ superuser_reserved_connections.
+ </para>
+ <para>
+ In summary, max_pool, num_init_children, max_connections,
+ superuser_reserved_connections must satisfy the following formula:
+ <programlisting>
+ max_pool*num_init_children <= (max_connections - superuser_reserved_connections) (no query canceling needed)
+ max_pool*num_init_children*2 <= (max_connections - superuser_reserved_connections) (query canceling needed)
+ </programlisting>
+ </para>
+ <para>
+ This parameter can only be set at server start.
+ </para>
</listitem>
- </varlistentry>
- </variablelist>
- </sect2>
-
- <sect2 id="runtime-config-authentication-settings">
- <title>Authentication Settings</title>
- <variablelist>
-
- <varlistentry id="guc-enable-pool-hba" xreflabel="enable_pool_hba">
- <term><varname>enable_pool_hba</varname> (<type>boolean</type>)
- <indexterm>
- <primary><varname>enable_pool_hba</varname> configuration parameter</primary>
- </indexterm>
- </term>
- <listitem>
- <para>
- If <literal>true</literal>, <productname>Pgpool-II</productname> will use the
- <filename>pool_hba.conf</filename> for the client authentication.
- See <xref linkend="auth-pool-hba-conf"> for details on how to configure
- <filename>pool_hba.conf</filename> for client authentication.
- Default is <literal>false</literal>
- </para>
- <para>
- This parameter can be changed by reloading the <productname>Pgpool-II</> configurations.
- </para>
- </listitem>
- </varlistentry>
-
- <varlistentry id="guc-pool-passwd" xreflabel="pool_passwd">
- <term><varname>pool_passwd</varname> (<type>string</type>)
- <indexterm>
- <primary><varname>pool_passwd</varname> configuration parameter</primary>
- </indexterm>
- </term>
- <listitem>
- <para>
- Specify the password file name for md5 authentication.
- Default value is <literal>"pool_passwd"</literal>.
- Specifying <literal>''</literal> (empty) disables the use of password file.
- See <xref linkend="auth-md5"> for more details.
- </para>
- <para>
- This parameter can only be set at server start.
- </para>
- </listitem>
- </varlistentry>
-
- <varlistentry id="guc-authentication-timeout" xreflabel="authentication_timeout">
- <term><varname>authentication_timeout</varname> (<type>integer</type>)
- <indexterm>
- <primary><varname>authentication_timeout</varname> configuration parameter</primary>
- </indexterm>
- </term>
- <listitem>
- <para>
- Specify the timeout in seconds for <productname>Pgpool-II</productname>
- authentication. Specifying 0 disables the time out.
- Default value is 60
- </para>
- <para>
- This parameter can be changed by reloading the <productname>Pgpool-II</> configurations.
- </para>
- </listitem>
- </varlistentry>
-
- </variablelist>
-
- </sect2>
+ </varlistentry>
+ </variablelist>
+ </sect2>
+
+ <sect2 id="runtime-config-authentication-settings">
+ <title>Authentication Settings</title>
+ <variablelist>
+
+ <varlistentry id="guc-enable-pool-hba" xreflabel="enable_pool_hba">
+ <term><varname>enable_pool_hba</varname> (<type>boolean</type>)
+ <indexterm>
+ <primary><varname>enable_pool_hba</varname> configuration parameter</primary>
+ </indexterm>
+ </term>
+ <listitem>
+ <para>
+ If <literal>true</literal>, <productname>Pgpool-II</productname> will use the
+ <filename>pool_hba.conf</filename> for the client
+ authentication. See <xref linkend="auth-pool-hba-conf">
+ for details on how to configure
+ <filename>pool_hba.conf</filename> for client authentication.
+ Default is <literal>false</literal>
+ </para>
+ <para>
+ This parameter can be changed by reloading
+ the <productname>Pgpool-II</productname> configurations.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry id="guc-pool-passwd" xreflabel="pool_passwd">
+ <term><varname>pool_passwd</varname> (<type>string</type>)
+ <indexterm>
+ <primary><varname>pool_passwd</varname> configuration parameter</primary>
+ </indexterm>
+ </term>
+ <listitem>
+ <para>
+ Specify the password file name for md5 authentication.
+ Default value is <literal>"pool_passwd"</literal>.
+ Specifying <literal>''</literal> (empty) disables the use
+ of password file. See <xref linkend="auth-md5"> for more
+ details.
+ </para>
+ <para>
+ This parameter can only be set at server start.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry id="guc-authentication-timeout" xreflabel="authentication_timeout">
+ <term><varname>authentication_timeout</varname> (<type>integer</type>)
+ <indexterm>
+ <primary><varname>authentication_timeout</varname> configuration parameter</primary>
+ </indexterm>
+ </term>
+ <listitem>
+ <para>
+ Specify the timeout in seconds
+ for <productname>Pgpool-II</productname>
+ authentication. Specifying 0 disables the time out.
+ Default value is 60
+ </para>
+ <para>
+ This parameter can be changed by reloading
+ the <productname>Pgpool-II</productname> configurations.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ </variablelist>
+
+ </sect2>
</sect1>
<sect1 id="runtime-config-runnung-mode">
- <title>Running mode</title>
-
- <sect2 id="runtime-config-master-slave-mode">
- <title>Master slave mode</title>
-
- <para>
- This mode is used to couple <productname>Pgpool-II</productname>
- with another master/slave replication software (like <acronym>Slony-I</acronym>
- and Streaming replication), that is responsible for doing the actual data replication.
- </para>
-
- <note>
- <para>
- The number of slave nodes are not limited to 1 and
- <productname>Pgpool-II</productname> can have up to 127 slave nodes.
- master/slave mode can also work just master node without any slave nodes.
- </para>
- </note>
-
- <para>
- Load balancing (see <xref linkend="runtime-config-load-balancing"> ) can
- also be used with master/slave mode to distribute the read load on the
- standby backend nodes.
- </para>
- <para>
- Following options are required to be specified for master/slave mode.
- </para>
-
- <variablelist>
-
- <varlistentry id="guc-master-slave-mode" xreflabel="master_slave_mode">
- <term><varname>master_slave_mode</varname> (<type>boolean</type>)
- <indexterm>
- <primary><varname>master_slave_mode</varname> configuration parameter</primary>
- </indexterm>
- </term>
- <listitem>
- <para>
- Setting to on enables the master/slave mode.
- Default is off.
- </para>
- <note>
- <para>
- <xref linkend="guc-master-slave-mode"> and <xref linkend="guc-replication-mode">
- are mutually exclusive and only one can be enabled at a time.
- </para>
- </note>
- <para>
- This parameter can only be set at server start.
- </para>
- </listitem>
- </varlistentry>
-
- <varlistentry id="guc-master-slave-sub-mode" xreflabel="master_slave_sub_mode">
- <term><varname>master_slave_sub_mode</varname> (<type>enum</type>)
- <indexterm>
- <primary><varname>master_slave_sub_mode</varname> configuration parameter</primary>
- </indexterm>
- </term>
- <listitem>
- <para>
- Specifies the external replication system used for data replication between
- <productname>PostgreSQL</> nodes.
- Below table contains the list of valid values for the parameter.
- </para>
-
- <table id="master-slave-sub-mode-table">
- <title>master slave sub mode options</title>
- <tgroup cols="2">
- <thead>
- <row>
- <entry>Value</entry>
- <entry>Description</entry>
- </row>
- </thead>
-
- <tbody>
- <row>
- <entry><literal>'slony'</literal></entry>
- <entry>Suitable for <acronym>Slony-I</acronym></entry>
- </row>
-
- <row>
- <entry><literal>'stream'</literal></entry>
- <entry>Suitable for <productname>PostgreSQL</>'s built-in replication system (Streaming Replication)</entry>
- </row>
-
- </tbody>
- </tgroup>
- </table>
-
- <para>
- Default is <literal>'slony'</literal>.
- </para>
- <para>
- This parameter can only be set at server start.
- </para>
- </listitem>
- </varlistentry>
-
- </variablelist>
- </sect2>
-
- <sect2 id="runtime-config-replication-mode">
- <title>Replication mode</title>
-
- <para>
- This mode makes the <productname>Pgpool-II</productname> to replicate data
- between <productname>PostgreSQL</> backends.
- </para>
-
- <para>
- Load balancing (see <xref linkend="runtime-config-load-balancing"> ) can
- also be used with replication mode to distribute the load to the
- attached backend nodes.
- </para>
- <para>
- Following options affect the behavior of
- <productname>Pgpool-II</productname> in the replication mode.
- </para>
-
- <variablelist>
-
- <varlistentry id="guc-replication-mode" xreflabel="replication_mode">
- <term><varname>replication_mode</varname> (<type>boolean</type>)
- <indexterm>
- <primary><varname>replication_mode</varname> configuration parameter</primary>
- </indexterm>
- </term>
- <listitem>
- <para>
- Setting to on enables the replication mode.
- Default is off.
- </para>
- <note>
- <para>
- <xref linkend="guc-replication-mode"> and <xref linkend="guc-master-slave-mode">
- are mutually exclusive and only one can be enabled at a time.
- </para>
- </note>
- <para>
- This parameter can only be set at server start.
- </para>
- </listitem>
- </varlistentry>
-
- <varlistentry id="guc-replication-stop-on-mismatch" xreflabel="replication_stop_on_mismatch">
- <term><varname>replication_stop_on_mismatch</varname> (<type>boolean</type>)
- <indexterm>
- <primary><varname>replication_stop_on_mismatch</varname> configuration parameter</primary>
- </indexterm>
- </term>
- <listitem>
- <para>
- When set to on, and all nodes do not reply with the same packet kind
- to the query that was sent to all <productname>PostgreSQL</> backend nodes, then the
- backend node whose reply differs from the majority is degenerated
- by the <productname>Pgpool-II</productname>.
- If <varname>replication_stop_on_mismatch</varname> is set to off and
- a similar situation happens then the <productname>Pgpool-II</productname>
- only terminates the current user session but does not degenerate a backend node.
- </para>
-
- <note>
- <para>
- <productname>Pgpool-II</productname> does not examine the data returned
- by the backends and takes the decision only by comparing the result packet
- types.
- </para>
- </note>
-
- <para>
- A typical use case of enabling the <varname>replication_stop_on_mismatch</varname>
- is to guard against the data inconsistency among the backend nodes.
- For example, you may want to degenerate a backend node if an UPDATE statement
- fails on one backend node while passes on others.
- </para>
- <para>
- Default is off.
- </para>
- <para>
- This parameter can be changed by reloading the <productname>Pgpool-II</> configurations.
- </para>
- </listitem>
- </varlistentry>
-
- <varlistentry id="guc-failover-if-affected-tuples-mismatch" xreflabel="failover_if_affected_tuples_mismatch">
- <term><varname>failover_if_affected_tuples_mismatch</varname> (<type>boolean</type>)
- <indexterm>
- <primary><varname>failover_if_affected_tuples_mismatch</varname> configuration parameter</primary>
- </indexterm>
- </term>
- <listitem>
- <para>
- When set to on, and all nodes do not reply with the same number of
- affected tuples to the INSERT/UPDATE/DELETE query, then the
- backend node whose reply differs from the majority is degenerated
- by the <productname>Pgpool-II</productname>.
- If <varname>failover_if_affected_tuples_mismatch</varname> is set to off and
- a similar situation happens then the <productname>Pgpool-II</productname>
- only terminates the current user session but does not degenerate a backend node.
- </para>
-
- <note>
- <para>
- In case of a tie, when two or more groups have the same number of nodes,
- then the group containing the master node (backend node having
- the youngest node id) gets the precedence.
- </para>
- </note>
-
- <para>
- Default is off.
- </para>
- <para>
- This parameter can be changed by reloading the <productname>Pgpool-II</> configurations.
- </para>
- </listitem>
- </varlistentry>
-
- <varlistentry id="guc-replicate-select" xreflabel="replicate_select">
- <term><varname>replicate_select</varname> (<type>boolean</type>)
- <indexterm>
- <primary><varname>replicate_select</varname> configuration parameter</primary>
- </indexterm>
- </term>
- <listitem>
- <para>
- When set to on, <productname>Pgpool-II</productname> enables the
- SELECT query replication mode. i.e. The SELECT queries are sent
- to all backend nodes.
- </para>
-
- <table id="replicate-select-affect-table">
- <title>replicate_select with <xref linkend="guc-load-balance-mode"> affects on SELECT routing</title>
- <tgroup cols="6" align="center">
- <colspec colname="_1" colwidth="1*">
- <colspec colname="_2" colwidth="1*">
- <colspec colname="_3" colwidth="1*">
- <colspec colname="_4" colwidth="1*">
- <colspec colname="_5" colwidth="1*">
- <colspec colname="_6" colwidth="1*">
-
- <tbody>
- <row>
- <entry>replicate_select is true</entry>
- <entry align="center">Y</entry>
- <entry align="center" nameend="_6" namest="_3">N</entry>
- </row>
-
- <row>
- <entry>load_balance_mode is true</entry>
- <entry>ANY</entry>
- <entry align="center">Y</entry>
- <entry nameend="_6" namest="_4" align="center">N</entry>
- </row>
-
- <row>
- <entry>SELECT is inside a transaction block</entry>
- <entry align="center">ANY</entry>
- <entry nameend="_4" namest="_3" align="center"> Y </entry>
- <entry align="center">N</entry>
- <entry align="center">ANY</entry>
- </row>
-
- <row>
- <entry>
- Transaction isolation level is SERIALIZABLE and
- the transaction has issued a write query
- </entry>
- <entry align="center">Y</entry>
- <entry align="center">N</entry>
- <entry align="center">ANY</entry>
- <entry align="center">ANY</entry>
- <entry align="center">ANY</entry>
- </row>
-
- <row>
- <entry>
- results(R:replication, M: send only to master, L: load balance)
- </entry>
- <entry align="center">R</entry>
- <entry align="center">M</entry>
- <entry align="center">L</entry>
- <entry align="center">L</entry>
- <entry align="center">M</entry>
- </row>
-
- </tbody>
- </tgroup>
- </table>
-
- <para>
- Default is off.
- </para>
- <para>
- This parameter can be changed by reloading the <productname>Pgpool-II</> configurations.
- </para>
- </listitem>
- </varlistentry>
-
- <varlistentry id="guc-insert-lock" xreflabel="insert_lock">
- <term><varname>insert_lock</varname> (<type>boolean</type>)
- <indexterm>
- <primary><varname>insert_lock</varname> configuration parameter</primary>
- </indexterm>
- </term>
- <listitem>
- <para>
- When set to on, <productname>Pgpool-II</productname> will automatically lock
- the table on <productname>PostgreSQL</> before an INSERT statement is issued for that.
- </para>
-
- <para>
- When replicating a table with SERIAL data type,
- the SERIAL column value may get different values
- on the different backends.
- The workaround to this problem is to explicitly lock the table
- before issuing the INSERT.
- </para>
- <para>
- So for automatically locking the table <productname>Pgpool-II</productname> do the following transformation:
- <programlisting>
-INSERT INTO ...
- </programlisting>
- to
- <programlisting>
-BEGIN;
-LOCK TABLE ...
-INSERT INTO ...
-COMMIT;
- </programlisting>
- </para>
- <caution>
- <para>
- This approach severely degrades the transactions' parallelism
- </para>
- </caution>
-
- <para>
- <productname>Pgpool-II</productname> <emphasis>V2.2</emphasis> or later,
- automatically detects whether the table has a SERIAL columns or not,
- so it never locks the table if it desn't have the SERIAL columns.
- </para>
-
- <para>
- <productname>Pgpool-II</productname> <emphasis>V3.0</emphasis> until
- <productname>Pgpool-II</productname> <emphasis>V3.0.4</emphasis> uses a row lock
- against the sequence relation, rather than table lock.
- This is intended to minimize lock conflict with <acronym>VACUUM</acronym>
- (including autovacuum).
- However this can lead to another problem. After transaction wraparound happens,
- row locking against the sequence relation causes PostgreSQL internal error
- (more precisely, access error on pg_clog, which keeps transaction status).
- To prevent this, <productname>PostgreSQL</> core developers decided to disallow row locking
- against sequences and this broke the <productname>Pgpool-II</productname>,
- of course (the "fixed" version of PostgreSQL was released as
- 9.0.5, 8.4.9, 8.3.16 and 8.2.22).
- </para>
-
- <para>
- <productname>Pgpool-II</productname> <emphasis>V3.0.5</emphasis> or later
- uses a row lock against <literal>pgpool_catalog.insert_lock</literal> table
- because new PostgreSQL disallows a row lock against the sequence relation.
- So creating insert_lock table in all databases which are accessed via
- <productname>Pgpool-II</productname> beforehand is required.
- See <xref linkend="create-installlock-table"> for more details.
- If does not exist insert_lock table, <productname>Pgpool-II</productname>
- locks the insert target table.
- This behavior is same as <productname>Pgpool-II</productname> <emphasis>V2.2</emphasis>
- and <emphasis>V2.3</emphasis> series.
- </para>
- <para>
- If you want to use <varname>insert_lock</varname> which is compatible
- with older releases, you can specify lock method by configure script.
- See <xref linkend="install-pgpool"> for more details.
- </para>
-
- <para>
- For fine (per statement) control:
- </para>
-
- <itemizedlist>
- <listitem>
- <para>
- set insert_lock to true, and add /*NO INSERT LOCK*/ at the beginning of an
- INSERT statement for which you do not want to acquire the table lock.
- </para>
- </listitem>
- <listitem>
- <para>
- set insert_lock to false, and add /*INSERT LOCK*/ at the beginning of an
- INSERT statement for which you want to acquire the table lock.
- </para>
- </listitem>
-
- </itemizedlist>
- <note>
- <para>
- If insert_lock is enabled, the regression tests for PostgreSQL 8.0 gets fail
- in transactions, privileges, rules, and alter_table.
- </para>
- <para>
- The reason for this is that <productname>Pgpool-II</productname>
- tries to LOCK the VIEW for the rule test, and it produces the below error message:
- <programlisting>
-! ERROR: current transaction is aborted, commands ignored until
-end of transaction block
- </programlisting>
- For example, the transactions test tries an INSERT into a table which does not exist,
- and <productname>Pgpool-II</productname> causes <productname>PostgreSQL</productname> to
- acquire the lock for the table. Of cause this results in an error.
- The transaction will be aborted, and the following INSERT statement produces the above error message.
- </para>
- </note>
- <para>
- Default is off.
- </para>
- <para>
- This parameter can be changed by reloading the <productname>Pgpool-II</> configurations.
- </para>
- </listitem>
- </varlistentry>
-
- <varlistentry id="guc-lobj-lock-table" xreflabel="lobj_lock_table">
- <term><varname>lobj_lock_table</varname> (<type>string</type>)
- <indexterm>
- <primary><varname>lobj_lock_table</varname> configuration parameter</primary>
- </indexterm>
- </term>
- <listitem>
- <para>
- Specifies a table name used for large object replication control.
- If it is specified, <productname>Pgpool-II</productname> will lock
- the table specified by <varname>lobj_lock_table</varname> and generate
- a large object id by looking into <literal>pg_largeobject</literal>
- system catalog and then call <literal>lo_create</literal> to create
- the large object.
- This procedure guarantees that <productname>Pgpool-II</productname>
- will get the same large object id in all DB nodes in replication mode.
- </para>
- <note>
- <para>
- <productname>PostgreSQL</> 8.0 and older does not have <literal>lo_create</literal>,
- so this feature does not work with PostgreSQL 8.0 and older versions.
- </para>
- </note>
- <para>
- A call to the <literal>libpq</literal> function <literal>lo_creat()</literal>
- triggers this feature. Also large object creation through <acronym>Java</acronym>
- API (<acronym>JDBC</acronym> driver), <acronym>PHP</acronym> API
- (<literal>pg_lo_create</literal>, or similar API in PHP library such as PDO),
- and this same API in various programming languages are known to use a
- similar protocol, and thus should work.
- </para>
-
- <para>
- This feature does not works with following operations on large objects.
- </para>
- <itemizedlist>
- <listitem>
- <para>
- All APIs using <literal>lo_create</literal>, <literal>lo_import_with_oid</literal>.
- </para>
- </listitem>
- <listitem>
- <para>
- <literal>lo_import</literal> function in backend called in SELECT.
- </para>
- </listitem>
- <listitem>
- <para>
- <literal>lo_create</literal> function in backend called in SELECT.
- </para>
- </listitem>
- </itemizedlist>
-
- <note>
- <para>
- All <productname>PostgreSQL</> users must have a write access on <varname>lobj_lock_table</varname>
- and it can be created in any schema.
- </para>
- </note>
-
- <para>
- Example to create a large object lock table:
- <programlisting>
-CREATE TABLE public.my_lock_table ();
-GRANT ALL ON public.my_lock_table TO PUBLIC;
- </programlisting>
- </para>
- <para>
- Default is <literal>''</literal>(empty), which disables the feature.
- </para>
- </listitem>
- </varlistentry>
-
- </variablelist>
- </sect2>
+ <title>Running mode</title>
+
+ <sect2 id="runtime-config-master-slave-mode">
+ <title>Master slave mode</title>
+
+ <para>
+ This mode is used to couple <productname>Pgpool-II</productname>
+ with another master/slave replication software
+ (like <acronym>Slony-I</acronym> and Streaming replication),
+ that is responsible for doing the actual data replication.
+ </para>
+
+ <note>
+ <para>
+ The number of slave nodes are not limited to 1 and
+ <productname>Pgpool-II</productname> can have up to 127 slave nodes.
+ master/slave mode can also work just master node without any slave nodes.
+ </para>
+ </note>
+
+ <para>
+ Load balancing (see <xref linkend="runtime-config-load-balancing"> ) can
+ also be used with master/slave mode to distribute the read load on the
+ standby backend nodes.
+ </para>
+ <para>
+ Following options are required to be specified for master/slave mode.
+ </para>
+
+ <variablelist>
+
+ <varlistentry id="guc-master-slave-mode" xreflabel="master_slave_mode">
+ <term><varname>master_slave_mode</varname> (<type>boolean</type>)
+ <indexterm>
+ <primary><varname>master_slave_mode</varname> configuration parameter</primary>
+ </indexterm>
+ </term>
+ <listitem>
+ <para>
+ Setting to on enables the master/slave mode.
+ Default is off.
+ </para>
+ <note>
+ <para>
+ <xref linkend="guc-master-slave-mode"> and <xref linkend="guc-replication-mode">
+ are mutually exclusive and only one can be enabled at a time.
+ </para>
+ </note>
+ <para>
+ This parameter can only be set at server start.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry id="guc-master-slave-sub-mode" xreflabel="master_slave_sub_mode">
+ <term><varname>master_slave_sub_mode</varname> (<type>enum</type>)
+ <indexterm>
+ <primary><varname>master_slave_sub_mode</varname> configuration parameter</primary>
+ </indexterm>
+ </term>
+ <listitem>
+ <para>
+ Specifies the external replication system used for data
+ replication between
+ <productname>PostgreSQL</productname> nodes.
+ Below table contains the list of valid values for the parameter.
+ </para>
+
+ <table id="master-slave-sub-mode-table">
+ <title>master slave sub mode options</title>
+ <tgroup cols="2">
+ <thead>
+ <row>
+ <entry>Value</entry>
+ <entry>Description</entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry><literal>'slony'</literal></entry>
+ <entry>Suitable for <acronym>Slony-I</acronym></entry>
+ </row>
+
+ <row>
+ <entry><literal>'stream'</literal></entry>
+ <entry>Suitable
+ for <productname>PostgreSQL</productname>'s built-in
+ replication system (Streaming Replication)</entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
+ <para>
+ Default is <literal>'slony'</literal>.
+ </para>
+ <para>
+ This parameter can only be set at server start.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ </variablelist>
+ </sect2>
+
+ <sect2 id="runtime-config-replication-mode">
+ <title>Replication mode</title>
+
+ <para>
+ This mode makes the <productname>Pgpool-II</productname> to
+ replicate data between <productname>PostgreSQL</productname>
+ backends.
+ </para>
+
+ <para>
+ Load balancing
+ (see <xref linkend="runtime-config-load-balancing"> ) can also
+ be used with replication mode to distribute the load to the
+ attached backend nodes.
+ </para>
+ <para>
+ Following options affect the behavior of
+ <productname>Pgpool-II</productname> in the replication mode.
+ </para>
+
+ <variablelist>
+
+ <varlistentry id="guc-replication-mode" xreflabel="replication_mode">
+ <term><varname>replication_mode</varname> (<type>boolean</type>)
+ <indexterm>
+ <primary><varname>replication_mode</varname> configuration parameter</primary>
+ </indexterm>
+ </term>
+ <listitem>
+ <para>
+ Setting to on enables the replication mode.
+ Default is off.
+ </para>
+ <note>
+ <para>
+ <xref linkend="guc-replication-mode">
+ and <xref linkend="guc-master-slave-mode"> are
+ mutually exclusive and only one can be enabled at a
+ time.
+ </para>
+ </note>
+ <para>
+ This parameter can only be set at server start.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry id="guc-replication-stop-on-mismatch" xreflabel="replication_stop_on_mismatch">
+ <term><varname>replication_stop_on_mismatch</varname> (<type>boolean</type>)
+ <indexterm>
+ <primary><varname>replication_stop_on_mismatch</varname> configuration parameter</primary>
+ </indexterm>
+ </term>
+ <listitem>
+ <para>
+ When set to on, and all nodes do not reply with the same
+ packet kind to the query that was sent to
+ all <productname>PostgreSQL</productname> backend nodes,
+ then the backend node whose reply differs from the
+ majority is degenerated by
+ the <productname>Pgpool-II</productname>.
+ If <varname>replication_stop_on_mismatch</varname> is set
+ to off and a similar situation happens then
+ the <productname>Pgpool-II</productname> only terminates
+ the current user session but does not degenerate a backend
+ node.
+ </para>
+
+ <note>
+ <para>
+ <productname>Pgpool-II</productname> does not examine
+ the data returned by the backends and takes the decision
+ only by comparing the result packet types.
+ </para>
+ </note>
+
+ <para>
+ A typical use case of enabling
+ the <varname>replication_stop_on_mismatch</varname> is to
+ guard against the data inconsistency among the backend
+ nodes. For example, you may want to degenerate a backend
+ node if an UPDATE statement fails on one backend node
+ while passes on others.
+ </para>
+ <para>
+ Default is off.
+ </para>
+ <para>
+ This parameter can be changed by reloading
+ the <productname>Pgpool-II</productname> configurations.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry id="guc-failover-if-affected-tuples-mismatch" xreflabel="failover_if_affected_tuples_mismatch">
+ <term><varname>failover_if_affected_tuples_mismatch</varname> (<type>boolean</type>)
+ <indexterm>
+ <primary><varname>failover_if_affected_tuples_mismatch</varname> configuration parameter</primary>
+ </indexterm>
+ </term>
+ <listitem>
+ <para>
+ When set to on, and all nodes do not reply with the same
+ number of affected tuples to the INSERT/UPDATE/DELETE
+ query, then the backend node whose reply differs from the
+ majority is degenerated by
+ the <productname>Pgpool-II</productname>.
+ If <varname>failover_if_affected_tuples_mismatch</varname>
+ is set to off and a similar situation happens then
+ the <productname>Pgpool-II</productname> only terminates
+ the current user session but does not degenerate a backend
+ node.
+ </para>
+
+ <note>
+ <para>
+ In case of a tie, when two or more groups have the same
+ number of nodes, then the group containing the master
+ node (backend node having the youngest node id) gets the
+ precedence.
+ </para>
+ </note>
+
+ <para>
+ Default is off.
+ </para>
+ <para>
+ This parameter can be changed by reloading the <productname>Pgpool-II</productname> configurations.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry id="guc-replicate-select" xreflabel="replicate_select">
+ <term><varname>replicate_select</varname> (<type>boolean</type>)
+ <indexterm>
+ <primary><varname>replicate_select</varname> configuration parameter</primary>
+ </indexterm>
+ </term>
+ <listitem>
+ <para>
+ When set to on, <productname>Pgpool-II</productname> enables the
+ SELECT query replication mode. i.e. The SELECT queries are sent
+ to all backend nodes.
+ </para>
+
+ <table id="replicate-select-affect-table">
+ <title>replicate_select with <xref linkend="guc-load-balance-mode"> affects on SELECT routing</title>
+ <tgroup cols="6" align="center">
+ <colspec colname="_1" colwidth="1*">
+ <colspec colname="_2" colwidth="1*">
+ <colspec colname="_3" colwidth="1*">
+ <colspec colname="_4" colwidth="1*">
+ <colspec colname="_5" colwidth="1*">
+ <colspec colname="_6" colwidth="1*">
+
+ <tbody>
+ <row>
+ <entry>replicate_select is true</entry>
+ <entry align="center">Y</entry>
+ <entry align="center" nameend="_6" namest="_3">N</entry>
+ </row>
+
+ <row>
+ <entry>load_balance_mode is true</entry>
+ <entry>ANY</entry>
+ <entry align="center">Y</entry>
+ <entry nameend="_6" namest="_4" align="center">N</entry>
+ </row>
+
+ <row>
+ <entry>SELECT is inside a transaction block</entry>
+ <entry align="center">ANY</entry>
+ <entry nameend="_4" namest="_3" align="center"> Y </entry>
+ <entry align="center">N</entry>
+ <entry align="center">ANY</entry>
+ </row>
+
+ <row>
+ <entry>
+ Transaction isolation level is SERIALIZABLE and
+ the transaction has issued a write query
+ </entry>
+ <entry align="center">Y</entry>
+ <entry align="center">N</entry>
+ <entry align="center">ANY</entry>
+ <entry align="center">ANY</entry>
+ <entry align="center">ANY</entry>
+ </row>
+
+ <row>
+ <entry>
+ results(R:replication, M: send only to master, L: load balance)
+ </entry>
+ <entry align="center">R</entry>
+ <entry align="center">M</entry>
+ <entry align="center">L</entry>
+ <entry align="center">L</entry>
+ <entry align="center">M</entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
+ <para>
+ Default is off.
+ </para>
+ <para>
+ This parameter can be changed by reloading the <productname>Pgpool-II</productname> configurations.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry id="guc-insert-lock" xreflabel="insert_lock">
+ <term><varname>insert_lock</varname> (<type>boolean</type>)
+ <indexterm>
+ <primary><varname>insert_lock</varname> configuration parameter</primary>
+ </indexterm>
+ </term>
+ <listitem>
+ <para>
+ When set to on, <productname>Pgpool-II</productname> will
+ automatically lock the table
+ on <productname>PostgreSQL</productname> before an INSERT
+ statement is issued for that.
+ </para>
+
+ <para>
+ When replicating a table with SERIAL data type, the SERIAL
+ column value may get different values on the different
+ backends. The workaround to this problem is to explicitly
+ lock the table before issuing the INSERT.
+ </para>
+ <para>
+ So for automatically locking the
+ table <productname>Pgpool-II</productname> do the
+ following transformation:
+ <programlisting>
+ INSERT INTO ...
+ </programlisting>
+ to
+ <programlisting>
+ BEGIN;
+ LOCK TABLE ...
+ INSERT INTO ...
+ COMMIT;
+ </programlisting>
+ </para>
+ <caution>
+ <para>
+ This approach severely degrades the transactions'
+ parallelism
+ </para>
+ </caution>
+
+ <para>
+ <productname>Pgpool-II</productname> <emphasis>V2.2</emphasis>
+ or later, automatically detects whether the table has a
+ SERIAL columns or not, so it never locks the table if it
+ desn't have the SERIAL columns.
+ </para>
+
+ <para>
+ <productname>Pgpool-II</productname> <emphasis>V3.0</emphasis> until
+ <productname>Pgpool-II</productname> <emphasis>V3.0.4</emphasis>
+ uses a row lock against the sequence relation, rather than
+ table lock. This is intended to minimize lock conflict
+ with <acronym>VACUUM</acronym> (including autovacuum).
+ However this can lead to another problem. After
+ transaction wraparound happens, row locking against the
+ sequence relation causes PostgreSQL internal error (more
+ precisely, access error on pg_clog, which keeps
+ transaction status). To prevent
+ this, <productname>PostgreSQL</productname> core
+ developers decided to disallow row locking against
+ sequences and this broke
+ the <productname>Pgpool-II</productname>, of course (the
+ "fixed" version of PostgreSQL was released as 9.0.5,
+ 8.4.9, 8.3.16 and 8.2.22).
+ </para>
+
+ <para>
+ <productname>Pgpool-II</productname> <emphasis>V3.0.5</emphasis>
+ or later uses a row lock
+ against <literal>pgpool_catalog.insert_lock</literal>
+ table because new PostgreSQL disallows a row lock against
+ the sequence relation. So creating insert_lock table in
+ all databases which are accessed via
+ <productname>Pgpool-II</productname> beforehand is
+ required. See <xref linkend="create-installlock-table">
+ for more details. If does not exist insert_lock
+ table, <productname>Pgpool-II</productname> locks the
+ insert target table. This behavior is same
+ as <productname>Pgpool-II</productname> <emphasis>V2.2</emphasis>
+ and <emphasis>V2.3</emphasis> series.
+ </para>
+ <para>
+ If you want to use <varname>insert_lock</varname> which is
+ compatible with older releases, you can specify lock
+ method by configure script.
+ See <xref linkend="install-pgpool"> for more details.
+ </para>
+
+ <para>
+ For fine (per statement) control:
+ </para>
+
+ <itemizedlist>
+ <listitem>
+ <para>
+ set insert_lock to true, and add /*NO INSERT LOCK*/ at
+ the beginning of an INSERT statement for which you do
+ not want to acquire the table lock.
+ </para>
+ </listitem>
+ <listitem>
+ <para>
+ set insert_lock to false, and add /*INSERT LOCK*/ at
+ the beginning of an INSERT statement for which you
+ want to acquire the table lock.
+ </para>
+ </listitem>
+
+ </itemizedlist>
+ <note>
+ <para>
+ If insert_lock is enabled, the regression tests for
+ PostgreSQL 8.0 gets fail in transactions, privileges,
+ rules, and alter_table.
+ </para>
+ <para>
+ The reason for this is
+ that <productname>Pgpool-II</productname> tries to LOCK
+ the VIEW for the rule test, and it produces the below
+ error message:
+ <programlisting>
+ ! ERROR: current transaction is aborted, commands ignored until
+ end of transaction block
+ </programlisting>
+ For example, the transactions test tries an INSERT into
+ a table which does not exist,
+ and <productname>Pgpool-II</productname>
+ causes <productname>PostgreSQL</productname> to acquire
+ the lock for the table. Of cause this results in an
+ error. The transaction will be aborted, and the
+ following INSERT statement produces the above error
+ message.
+ </para>
+ </note>
+ <para>
+ Default is off.
+ </para>
+ <para>
+ This parameter can be changed by reloading
+ the <productname>Pgpool-II</productname> configurations.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry id="guc-lobj-lock-table" xreflabel="lobj_lock_table">
+ <term><varname>lobj_lock_table</varname> (<type>string</type>)
+ <indexterm>
+ <primary><varname>lobj_lock_table</varname> configuration parameter</primary>
+ </indexterm>
+ </term>
+ <listitem>
+ <para>
+ Specifies a table name used for large object replication control.
+ If it is specified, <productname>Pgpool-II</productname> will lock
+ the table specified by <varname>lobj_lock_table</varname> and generate
+ a large object id by looking into <literal>pg_largeobject</literal>
+ system catalog and then call <literal>lo_create</literal> to create
+ the large object.
+ This procedure guarantees that <productname>Pgpool-II</productname>
+ will get the same large object id in all DB nodes in replication mode.
+ </para>
+ <note>
+ <para>
+ <productname>PostgreSQL</productname> 8.0 and older does
+ not have <literal>lo_create</literal>, so this feature
+ does not work with PostgreSQL 8.0 and older versions.
+ </para>
+ </note>
+ <para>
+ A call to the <literal>libpq</literal>
+ function <literal>lo_creat()</literal> triggers this
+ feature. Also large object creation
+ through <acronym>Java</acronym> API
+ (<acronym>JDBC</acronym> driver), <acronym>PHP</acronym>
+ API (<literal>pg_lo_create</literal>, or similar API in
+ PHP library such as PDO), and this same API in various
+ programming languages are known to use a similar protocol,
+ and thus should work.
+ </para>
+
+ <para>
+ This feature does not works with following operations on large objects.
+ </para>
+ <itemizedlist>
+ <listitem>
+ <para>
+ All APIs using <literal>lo_create</literal>, <literal>lo_import_with_oid</literal>.
+ </para>
+ </listitem>
+ <listitem>
+ <para>
+ <literal>lo_import</literal> function in backend called in SELECT.
+ </para>
+ </listitem>
+ <listitem>
+ <para>
+ <literal>lo_create</literal> function in backend called in SELECT.
+ </para>
+ </listitem>
+ </itemizedlist>
+
+ <note>
+ <para>
+ All <productname>PostgreSQL</productname> users must
+ have a write access
+ on <varname>lobj_lock_table</varname> and it can be
+ created in any schema.
+ </para>
+ </note>
+
+ <para>
+ Example to create a large object lock table:
+ <programlisting>
+ CREATE TABLE public.my_lock_table ();
+ GRANT ALL ON public.my_lock_table TO PUBLIC;
+ </programlisting>
+ </para>
+ <para>
+ Default is <literal>''</literal>(empty), which disables the feature.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ </variablelist>
+ </sect2>
</sect1>
<sect1 id="runtime-config-backend-settings">
- <title>Backend Settings</title>
+ <title>Backend Settings</title>
- <sect2 id="runtime-config-backend-connection-settings">
- <title>Backend Connection Settings</title>
+ <sect2 id="runtime-config-backend-connection-settings">
+ <title>Backend Connection Settings</title>
- <variablelist>
+ <variablelist>
- <varlistentry id="guc-backend-hostname" xreflabel="backend_hostname">
+ <varlistentry id="guc-backend-hostname" xreflabel="backend_hostname">
<term><varname>backend_hostname</varname> (<type>string</type>)
- <indexterm>
-
<primary><varname>backend_hostname</varname> configuration parameter</primary>
- </indexterm>
+ <indexterm>
+ <primary><varname>backend_hostname</varname> configuration parameter</primary>
+ </indexterm>
</term>
<listitem>
- <para>
- <varname>backend_hostname</varname> specifies the
- <productname>PostgreSQL</productname> backend to be connected to.
- It is used by <productname>Pgpool-II</productname> to communicate
- with the server.
- </para>
-
- <para>
- For TCP/IP communication, this parameter can take a hostname
- or an IP address. If this begins with a slash(<literal>/</literal>), it specifies
- Unix-domain communication rather than TCP/IP; the value is
- the name of the directory in which the socket file is
- stored. The default behavior when backend_hostname is empty
- (<literal>''</literal>) is to connect to a Unix-domain socket in <filename>/tmp</>.
- </para>
-
- <para>
- Multiple backends can be specified by adding a number at the
- end of the parameter name (e.g.backend_hostname0). This
- number is referred to as "DB node ID", and it starts from
- 0. The backend which was given the DB node ID of 0 will be
- called "master node". When multiple backends are defined, the
- service can be continued even if the master node is down (not
- true in some modes). In this case, the youngest DB node ID
- alive will be the new master node.
- </para>
-
- <para>
- Please note that the DB node which has id 0 has no special
- meaning if operated in streaming replication mode. Rather,
- you should care about if the DB node is the "primary node" or
- not. See <xref linkend="runtime-config-load-balancing">,
- <xref linkend="runtime-config-failover">,
- <xref linkend="runtime-streaming-replication-check">
- for more details.
- </para>
-
- <para>
- If you plan to use only one <productname>PostgreSQL</> server, specify it by
- backend_hostname0.
- </para>
-
- <para>
- New nodes can be added by adding parameter rows and reloading a
- configuration file. However, the existing values cannot be updated, so
- you must restart <productname>Pgpool-II</productname> in
- that case.
- </para>
+ <para>
+ <varname>backend_hostname</varname> specifies the
+ <productname>PostgreSQL</productname> backend to be
+ connected to. It is used
+ by <productname>Pgpool-II</productname> to communicate
+ with the server.
+ </para>
+
+ <para>
+ For TCP/IP communication, this parameter can take a
+ hostname or an IP address. If this begins with a
+ slash(<literal>/</literal>), it specifies Unix-domain
+ communication rather than TCP/IP; the value is the name of
+ the directory in which the socket file is stored. The
+ default behavior when backend_hostname is empty
+ (<literal>''</literal>) is to connect to a Unix-domain
+ socket in <filename>/tmp</filename>.
+ </para>
+
+ <para>
+ Multiple backends can be specified by adding a number at the
+ end of the parameter name (e.g.backend_hostname0). This
+ number is referred to as "DB node ID", and it starts from
+ 0. The backend which was given the DB node ID of 0 will be
+ called "master node". When multiple backends are defined, the
+ service can be continued even if the master node is down (not
+ true in some modes). In this case, the youngest DB node ID
+ alive will be the new master node.
+ </para>
+
+ <para>
+ Please note that the DB node which has id 0 has no special
+ meaning if operated in streaming replication mode. Rather,
+ you should care about if the DB node is the "primary node" or
+ not. See <xref linkend="runtime-config-load-balancing">,
+ <xref linkend="runtime-config-failover">,
+ <xref linkend="runtime-streaming-replication-check">
+ for more details.
+ </para>
+
+ <para>
+ If you plan to use only
+ one <productname>PostgreSQL</productname> server, specify
+ it by backend_hostname0.
+ </para>
+
+ <para>
+ New nodes can be added by adding parameter rows and
+ reloading a configuration file. However, the existing
+ values cannot be updated, so you must
+ restart <productname>Pgpool-II</productname> in that case.
+ </para>
</listitem>
- </varlistentry>
+ </varlistentry>
- <varlistentry id="guc-backend-port" xreflabel="backend_port">
+ <varlistentry id="guc-backend-port" xreflabel="backend_port">
<term><varname>backend_port</varname> (<type>integer</type>)
- <indexterm>
-
<primary><varname>backend_port</varname> configuration parameter</primary>
- </indexterm>
+ <indexterm>
+ <primary><varname>backend_port</varname> configuration parameter</primary>
+ </indexterm>
</term>
<listitem>
- <varname>backend_port</varname> specifies the port number
- of the backends. Multiple backends can be specified by
- adding a number at the end of the parameter name
- (e.g. backend_port0). If you plan to use only one
- <productname>PostgreSQL</> server, specify it by backend_port0.
- </para>
- New backend ports can be added by adding parameter rows and reloading a
- configuration file. However, the existing values cannot be updated, so
- you must restart <productname>Pgpool-II</productname> in
- that case.
- </para>
+ <para>
+ <varname>backend_port</varname> specifies the port number
+ of the backends. Multiple backends can be specified by
+ adding a number at the end of the parameter name
+ (e.g. backend_port0). If you plan to use only one
+ <productname>PostgreSQL</productname> server, specify it by backend_port0.
+ </para>
+ <para>
+ New backend ports can be added by adding parameter rows
+ and reloading a configuration file. However, the existing
+ values cannot be updated, so you must
+ restart <productname>Pgpool-II</productname> in that case.
+ </para>
</listitem>
- </varlistentry>
+ </varlistentry>
- <varlistentry id="guc-backend-weight" xreflabel="backend_weight">
+ <varlistentry id="guc-backend-weight" xreflabel="backend_weight">
<term><varname>backend_weight</varname> (<type>floating point</type>)
- <indexterm>
-
<primary><varname>backend_weight</varname> configuration parameter</primary>
- </indexterm>
+ <indexterm>
+ <primary><varname>backend_weight</varname> configuration parameter</primary>
+ </indexterm>
</term>
<listitem>
- <para>
- <varname>backend_weight</varname> specifies the load balance
- ratio of the backends. It may be set to any interger or
- floating point value greater than or equeal zero.
- Multiple backends can be specified by
- adding a number at the end of the parameter name
- (e.g. backend_weight0). If you plan to use only one
- PostgreSQL server, specify it by backend_weight0.
- </para>
- <para>
- New <varname>backend_weight</varname> can be added in this parameter by
- reloading a configuration file. However, this will take
- effect only for new established client sessions.
- <productname>Pgpool-II</> <emphasis>V2.2.6</>, <emphasis>V2.3</> or later
- allows alllow updating the values by reloading a configuration file.
- This is useful if you want to prevent any query sent to
- slaves to perform some administrative work in master/slave mode.
- </para>
+ <para>
+ <varname>backend_weight</varname> specifies the load balance
+ ratio of the backends. It may be set to any interger or
+ floating point value greater than or equeal zero.
+ Multiple backends can be specified by
+ adding a number at the end of the parameter name
+ (e.g. backend_weight0). If you plan to use only one
+ PostgreSQL server, specify it by backend_weight0.
+ </para>
+ <para>
+ New <varname>backend_weight</varname> can be added in this parameter by
+ reloading a configuration file. However, this will take
+ effect only for new established client sessions.
+ <productname>Pgpool-II</productname> <emphasis>V2.2.6</emphasis>, <emphasis>V2.3</emphasis>
+ or later allows alllow updating the values by reloading a
+ configuration file. This is useful if you want to prevent
+ any query sent to slaves to perform some administrative
+ work in master/slave mode.
+ </para>
</listitem>
- </varlistentry>
+ </varlistentry>
- </variablelist>
- </sect2>
+ </variablelist>
+ </sect2>
- <sect2 id="runtime-config-backend-data">
- <title>Backend Data Settings</title>
+ <sect2 id="runtime-config-backend-data">
+ <title>Backend Data Settings</title>
- <variablelist>
+ <variablelist>
- <varlistentry id="guc-backend-data-directory" xreflabel="backend_data_directory">
+ <varlistentry id="guc-backend-data-directory" xreflabel="backend_data_directory">
<term><varname>backend_data_directory</varname> (<type>string</type>)
- <indexterm>
-
<primary><varname>backend_data_directory</varname> configuration parameter</primary>
- </indexterm>
+ <indexterm>
+ <primary><varname>backend_data_directory</varname> configuration parameter</primary>
+ </indexterm>
</term>
<listitem>
- <varname>backend_data_directory</varname> specifies the
- database cluster directory of the backend. Multiple backends can be
- specified by adding a number at the end of the parameter
- name (e.g. backend_data_directory0). If you plan to use
- only one PostgreSQL server, specify it by
- backend_data_directory0.
- </para>
- New <varname>backend data_directory</> can be added by adding parameter rows and reloading a
- configuration file. However, the existing values cannot be updated, so
-
you must restart <productname>Pgpool-II</productname> in
- that case.
- </para>
+ <para>
+ <varname>backend_data_directory</varname> specifies the
+ database cluster directory of the backend. Multiple backends can be
+ specified by adding a number at the end of the parameter
+ name (e.g. backend_data_directory0). If you plan to use
+ only one PostgreSQL server, specify it by
+ backend_data_directory0.
+ </para>
+ <para>
+ New <varname>backend data_directory</varname> can be added by adding parameter rows and reloading a
+ configuration file. However, the existing values cannot be updated, so
+ you must restart <productname>Pgpool-II</productname> in
+ that case.
+ </para>
</listitem>
- </varlistentry>
+ </varlistentry>
- <varlistentry id="guc-backend-flag" xreflabel="backend_flag">
+ <varlistentry id="guc-backend-flag" xreflabel="backend_flag">
<term><varname>backend_flag</varname> (<type>string</type>)
- <indexterm>
-
<primary><varname>backend_flag</varname> configuration parameter</primary>
- </indexterm>
+ <indexterm>
+ <primary><varname>backend_flag</varname> configuration parameter</primary>
+ </indexterm>
</term>
<listitem>
- <varname>backend_flag</varname> controls various backend
- behavior. Multiple backends can be specified by adding a
- number at the end of the parameter name
- (e.g. backend_flag0). If you plan to use only one
- PostgreSQL server, specify it by backend_flag0.
- </para>
- New backend flags can be added by adding parameter rows and reloading a
- configuration file. Currently followings are allowed. Multiple flags can
- be specified by using "|".
- </para>
-
-
<table id="backend-flag-table">
- <title>Backend flags</title>
- <tgroup cols="2">
- <thead>
+ <para>
+ <varname>backend_flag</varname> controls various backend
+ behavior. Multiple backends can be specified by adding a
+ number at the end of the parameter name
+ (e.g. backend_flag0). If you plan to use only one
+ PostgreSQL server, specify it by backend_flag0.
+ </para>
+ <para>
+ New backend flags can be added by adding parameter rows and reloading a
+ configuration file. Currently followings are allowed. Multiple flags can
+ be specified by using "|".
+ </para>
+
+ <table id="backend-flag-table">
+ <title>Backend flags</title>
+ <tgroup cols="2">
+ <thead>
<row>
- <entry>Flag</entry>
- <entry>Description</entry>
+ <entry>Flag</entry>
+ <entry>Description</entry>
</row>
- </thead>
+ </thead>
- <tbody>
+ <tbody>
<row>
- <entry><literal>ALLOW_TO_FAILOVER</literal></entry>
- <entry>Allow to failover or detaching backend. This
- is the default. You cannot specify with
- DISALLOW_TO_FAILOVER at a same time.</entry>
+ <entry><literal>ALLOW_TO_FAILOVER</literal></entry>
+ <entry>Allow to failover or detaching backend. This
+ is the default. You cannot specify with
+ DISALLOW_TO_FAILOVER at a same time.</entry>
</row>
<row>
- <entry><literal>DISALLOW_TO_FAILOVER</literal></entry>
- <entry>Disllow to failover or detaching backend
- This is useful when you protect backend by
- using HA (High Availability) softwares such as
- <productname>Heartbeat</> or <productname>Pacemaker</>. You cannot specify with
- ALLOW_TO_FAILOVER at a same time.
- </entry>
+ <entry><literal>DISALLOW_TO_FAILOVER</literal></entry>
+ <entry>Disllow to failover or detaching backend
+ This is useful when you protect backend by
+ using HA (High Availability) softwares such as
+ <productname>Heartbeat</productname>
+ or <productname>Pacemaker</productname>. You cannot specify with
+ ALLOW_TO_FAILOVER at a same time.
+ </entry>
</row>
- </tbody>
- </tgroup>
- </table>
- <para>
- This parameter can be changed by reloading the <productname>Pgpool-II</> configurations.
- </para>
+ </tbody>
+ </tgroup>
+ </table>
+ <para>
+ This parameter can be changed by reloading
+ the <productname>Pgpool-II</productname> configurations.
+ </para>
</listitem>
- </varlistentry>
+ </varlistentry>
- </variablelist>
+ </variablelist>
- </sect2>
+ </sect2>
</sect1>
projects / pgpool2.git / commitdiff