mirror of
https://github.com/EnterpriseDB/repmgr.git
synced 2026-03-22 22:56:29 +00:00
Add section on promoting standby
This commit is contained in:
Ian Barwick
@@ -308,7 +308,7 @@
|
||||
</para>
|
||||
<para>
|
||||
After starting the standby, the cluster will look like this, showing that <literal>node3</literal>
|
||||
is attached to <literal>node3</literal>, not the primary (<literal>node1</literal>).
|
||||
is attached to <literal>node2</literal>, not the primary (<literal>node1</literal>).
|
||||
<programlisting>
|
||||
$ repmgr -f /etc/repmgr.conf cluster show
|
||||
ID | Name | Role | Status | Upstream | Location | Connection string
|
||||
|
||||
@@ -148,6 +148,26 @@
|
||||
</sect2>
|
||||
</sect1>
|
||||
|
||||
<sect1 id="repmgr-standby-promote" xreflabel="repmgr standby promote">
|
||||
<indexterm>
|
||||
<primary>repmgr standby promote</primary>
|
||||
</indexterm>
|
||||
<title>repmgr standby promote</title>
|
||||
<para>
|
||||
Promotes a standby to a primary if the current primary has failed. This
|
||||
command requires a valid <filename>repmgr.conf</filename> file for the standby, either
|
||||
specified explicitly with <literal>-f/--config-file</literal> or located in a
|
||||
default location; no additional arguments are required.
|
||||
</para>
|
||||
<para>
|
||||
If the standby promotion succeeds, the server will not need to be
|
||||
restarted. However any other standbys will need to follow the new server,
|
||||
by using <xref linkend="repmgr-standby-follow">; if <command>repmgrd</command> is active, it will
|
||||
handle this automatically.
|
||||
</para>
|
||||
|
||||
</sect1>
|
||||
|
||||
<sect1 id="repmgr-standby-follow" xreflabel="repmgr standby follow">
|
||||
<indexterm>
|
||||
<primary>repmgr standby follow</primary>
|
||||
@@ -170,6 +190,7 @@
|
||||
</sect1>
|
||||
|
||||
|
||||
|
||||
<sect1 id="repmgr-node-rejoin" xreflabel="repmgr node rejoin">
|
||||
<indexterm>
|
||||
<primary>repmgr node rejoin</primary>
|
||||
@@ -179,8 +200,191 @@
|
||||
Enables a dormant (stopped) node to be rejoined to the replication cluster.
|
||||
</para>
|
||||
<para>
|
||||
This can optionally use `pg_rewind` to re-integrate a node which has diverged
|
||||
This can optionally use <command>pg_rewind</command> to re-integrate a node which has diverged
|
||||
from the rest of the cluster, typically a failed primary.
|
||||
</para>
|
||||
</sect1>
|
||||
|
||||
<sect1 id="repmgr-cluster-show" xreflabel="repmgr cluster show">
|
||||
<indexterm>
|
||||
<primary>repmgr cluster show</primary>
|
||||
</indexterm>
|
||||
<title>repmgr cluster show</title>
|
||||
<para>
|
||||
Displays information about each active node in the replication cluster. This
|
||||
command polls each registered server and shows its role (<literal>primary</literal> /
|
||||
<literal>standby</literal> / <literal>bdr</literal>) and status. It polls each server
|
||||
directly and can be run on any node in the cluster; this is also useful when analyzing
|
||||
connectivity from a particular node.
|
||||
</para>
|
||||
<para>
|
||||
This command requires either a valid <filename>repmgr.conf</filename> file or a database
|
||||
connection string to one of the registered nodes; no additional arguments are needed.
|
||||
</para>
|
||||
|
||||
<para>
|
||||
Example:
|
||||
<programlisting>
|
||||
$ repmgr -f /etc/repmgr.conf cluster show
|
||||
|
||||
ID | Name | Role | Status | Upstream | Location | Connection string
|
||||
----+-------+---------+-----------+----------+----------+-----------------------------------------
|
||||
1 | node1 | primary | * running | | default | host=db_node1 dbname=repmgr user=repmgr
|
||||
2 | node2 | standby | running | node1 | default | host=db_node2 dbname=repmgr user=repmgr
|
||||
3 | node3 | standby | running | node1 | default | host=db_node3 dbname=repmgr user=repmgr</programlisting>
|
||||
</para>
|
||||
|
||||
<para>
|
||||
To show database connection errors when polling nodes, run the command in
|
||||
<literal>--verbose</literal> mode.
|
||||
</para>
|
||||
<para>
|
||||
The `cluster show` command accepts an optional parameter <literal>--csv</literal>, which
|
||||
outputs the replication cluster's status in a simple CSV format, suitable for
|
||||
parsing by scripts:
|
||||
<programlisting>
|
||||
$ repmgr -f /etc/repmgr.conf cluster show --csv
|
||||
1,-1,-1
|
||||
2,0,0
|
||||
3,0,1</programlisting>
|
||||
</para>
|
||||
<para>
|
||||
The columns have following meanings:
|
||||
<itemizedlist spacing="compact" mark="bullet">
|
||||
<listitem>
|
||||
<simpara>
|
||||
node ID
|
||||
</simpara>
|
||||
<simpara>
|
||||
availability (0 = available, -1 = unavailable)
|
||||
</simpara>
|
||||
<simpara>
|
||||
recovery state (0 = not in recovery, 1 = in recovery, -1 = unknown)
|
||||
</simpara>
|
||||
</listitem>
|
||||
</itemizedlist>
|
||||
</para>
|
||||
|
||||
<para>
|
||||
Note that the availability is tested by connecting from the node where
|
||||
<command>repmgr cluster show</command> is executed, and does not necessarily imply the node
|
||||
is down. See <xref linkend="repmgr-cluster-matrix"> and <xref linkend="repmgr-cluster-crosscheck"> to get
|
||||
a better overviews of connections between nodes.
|
||||
</para>
|
||||
</sect1>
|
||||
|
||||
<sect1 id="repmgr-cluster-matrix" xreflabel="repmgr cluster matrix">
|
||||
<indexterm>
|
||||
<primary>repmgr cluster matrix</primary>
|
||||
</indexterm>
|
||||
<title>repmgr cluster matric</title>
|
||||
<para>
|
||||
<command>repmgr cluster matrix</command> runs <command>repmgr cluster show</command> on each
|
||||
node and arranges the results in a matrix, recording success or failure.
|
||||
</para>
|
||||
<para>
|
||||
<command>repmgr cluster matrix</command> requires a valid <filename>repmgr.conf</filename>
|
||||
file on each node. Additionally passwordless `ssh` connections are required between
|
||||
all nodes.
|
||||
</para>
|
||||
<para>
|
||||
Example 1 (all nodes up):
|
||||
<programlisting>
|
||||
$ repmgr -f /etc/repmgr.conf cluster matrix
|
||||
|
||||
Name | Id | 1 | 2 | 3
|
||||
-------+----+----+----+----
|
||||
node1 | 1 | * | * | *
|
||||
node2 | 2 | * | * | *
|
||||
node3 | 3 | * | * | *</programlisting>
|
||||
</para>
|
||||
<para>
|
||||
Example 2 (<literal>node1</literal> and <literal>node2</literal> up, <literal>node3</literal> down):
|
||||
<programlisting>
|
||||
$ repmgr -f /etc/repmgr.conf cluster matrix
|
||||
|
||||
Name | Id | 1 | 2 | 3
|
||||
-------+----+----+----+----
|
||||
node1 | 1 | * | * | x
|
||||
node2 | 2 | * | * | x
|
||||
node3 | 3 | ? | ? | ?
|
||||
</programlisting>
|
||||
</para>
|
||||
<para>
|
||||
Each row corresponds to one server, and indicates the result of
|
||||
testing an outbound connection from that server.
|
||||
</para>
|
||||
<para>
|
||||
Since <literal>node3</literal> is down, all the entries in its row are filled with
|
||||
<literal>?</literal>, meaning that there we cannot test outbound connections.
|
||||
</para>
|
||||
<para>
|
||||
The other two nodes are up; the corresponding rows have <literal>x</literal> in the
|
||||
column corresponding to <literal>node3</literal>, meaning that inbound connections to
|
||||
that node have failed, and `*` in the columns corresponding to
|
||||
<literal>node1</literal> and <literal>node2</literal>, meaning that inbound connections
|
||||
to these nodes have succeeded.
|
||||
</para>
|
||||
<para>
|
||||
Example 3 (all nodes up, firewall dropping packets originating
|
||||
from <literal>node1</literal> and directed to port 5432 on <literal>node3</literal>) -
|
||||
running <command>repmgr cluster matrix</command> from <literal>node1</literal> gives the following output:
|
||||
<programlisting>
|
||||
$ repmgr -f /etc/repmgr.conf cluster matrix
|
||||
|
||||
Name | Id | 1 | 2 | 3
|
||||
-------+----+----+----+----
|
||||
node1 | 1 | * | * | x
|
||||
node2 | 2 | * | * | *
|
||||
node3 | 3 | ? | ? | ?</programlisting>
|
||||
</para>
|
||||
<para>
|
||||
Note this may take some time depending on the <varname>connect_timeout</varname>
|
||||
setting in the node <varname>conninfo</varname> strings; default is
|
||||
<literal>1 minute</literal> which means without modification the above
|
||||
command would take around 2 minutes to run; see comment elsewhere about setting
|
||||
<varname>connect_timeout</varname>)
|
||||
</para>
|
||||
<para>
|
||||
The matrix tells us that we cannot connect from <literal>node1</literal> to <literal>node3</literal>,
|
||||
and that (therefore) we don't know the state of any outbound
|
||||
connection from <literal>node3</literal>.
|
||||
</para>
|
||||
<para>
|
||||
In this case, the <xref linkend="repmgr-cluster-crosscheck"> command will produce a more
|
||||
useful result.
|
||||
</para>
|
||||
</sect1>
|
||||
|
||||
|
||||
<sect1 id="repmgr-cluster-crosscheck" xreflabel="repmgr cluster crosscheck">
|
||||
<indexterm>
|
||||
<primary>repmgr cluster crosscheck</primary>
|
||||
</indexterm>
|
||||
<title>repmgr cluster crosscheck</title>
|
||||
<para>
|
||||
<command>repmgr cluster crosscheck</command> is similar to <xref linkend="repmgr-cluster-matrix">,
|
||||
but cross-checks connections between each combination of nodes. In "Example 3" in
|
||||
<xref linkend="repmgr-cluster-matrix"> we have no information about the state of <literal>node3</literal>.
|
||||
However by running <command>repmgr cluster crosscheck</command> it's possible to get a better
|
||||
overview of the cluster situation:
|
||||
<programlisting>
|
||||
$ repmgr -f /etc/repmgr.conf cluster crosscheck
|
||||
|
||||
Name | Id | 1 | 2 | 3
|
||||
-------+----+----+----+----
|
||||
node1 | 1 | * | * | x
|
||||
node2 | 2 | * | * | *
|
||||
node3 | 3 | * | * | *</programlisting>
|
||||
</para>
|
||||
<para>
|
||||
What happened is that <command>repmgr cluster crosscheck</command> merged its own
|
||||
<command>repmgr cluster matrix</command> with the <command>repmgr cluster matrix</command>
|
||||
output from <literal>node2</literal>; the latter is able to connect to <literal>node3</literal>
|
||||
and therefore determine the state ofx outbound connections from that node.
|
||||
</para>
|
||||
|
||||
</sect1>
|
||||
|
||||
|
||||
</chapter>
|
||||
|
||||
@@ -40,6 +40,7 @@
|
||||
<!ENTITY configuration-file SYSTEM "configuration-file.sgml">
|
||||
<!ENTITY configuration-file-settings SYSTEM "configuration-file-settings.sgml">
|
||||
<!ENTITY cloning-standbys SYSTEM "cloning-standbys.sgml">
|
||||
<!ENTITY promoting-standby SYSTEM "promoting-standby.sgml">
|
||||
<!ENTITY command-reference SYSTEM "command-reference.sgml">
|
||||
<!ENTITY appendix-signatures SYSTEM "appendix-signatures.sgml">
|
||||
|
||||
|
||||
74
doc/promoting-standby.sgml
Normal file
74
doc/promoting-standby.sgml
Normal file
@@ -0,0 +1,74 @@
|
||||
<chapter id="promoting-standby" xreflabel="promoting a standby">
|
||||
<title>Promoting a standby server with repmgr</title>
|
||||
<para>
|
||||
If a primary server fails or needs to be removed from the replication cluster,
|
||||
a new primary server must be designated, to ensure the cluster continues
|
||||
to function correctly. This can be done with <xref linkend="repmgr-standby-promote">,
|
||||
which promotes the standby on the current server to primary.
|
||||
</para>
|
||||
|
||||
<para>
|
||||
To demonstrate this, set up a replication cluster with a primary and two attached
|
||||
standby servers so that the cluster looks like this:
|
||||
<programlisting>
|
||||
$ repmgr -f /etc/repmgr.conf cluster show
|
||||
ID | Name | Role | Status | Upstream | Location | Connection string
|
||||
----+-------+---------+-----------+----------+----------+--------------------------------------
|
||||
1 | node1 | primary | * running | | default | host=node1 dbname=repmgr user=repmgr
|
||||
2 | node2 | standby | running | node1 | default | host=node2 dbname=repmgr user=repmgr
|
||||
3 | node3 | standby | running | node1 | default | host=node3 dbname=repmgr user=repmgr</programlisting>
|
||||
</para>
|
||||
<para>
|
||||
Stop the current primary with e.g.:
|
||||
<programlisting>
|
||||
$ pg_ctl -D /var/lib/postgresql/data -m fast stop</programlisting>
|
||||
</para>
|
||||
<para>
|
||||
At this point the replication cluster will be in a partially disabled state, with
|
||||
both standbys accepting read-only connections while attempting to connect to the
|
||||
stopped primary. Note that the &repmgr; metadata table will not yet have been updated;
|
||||
executing <xref linkend="repmgr-cluster-show"> will note the discrepancy:
|
||||
<programlisting>
|
||||
$ repmgr -f /etc/repmgr.conf cluster show
|
||||
ID | Name | Role | Status | Upstream | Location | Connection string
|
||||
----+-------+---------+---------------+----------+----------+--------------------------------------
|
||||
1 | node1 | primary | ? unreachable | | default | host=node1 dbname=repmgr user=repmgr
|
||||
2 | node2 | standby | running | node1 | default | host=node2 dbname=repmgr user=repmgr
|
||||
3 | node3 | standby | running | node1 | default | host=node3 dbname=repmgr user=repmgr
|
||||
|
||||
WARNING: following issues were detected
|
||||
node "node1" (ID: 1) is registered as an active primary but is unreachable</programlisting>
|
||||
</para>
|
||||
<para>
|
||||
Now promote the first standby with:
|
||||
<programlisting>
|
||||
$ repmgr -f /etc/repmgr.conf standby promote</programlisting>
|
||||
</para>
|
||||
<para>
|
||||
This will produce output similar to the following:
|
||||
<programlisting>
|
||||
INFO: connecting to standby database
|
||||
NOTICE: promoting standby
|
||||
DETAIL: promoting server using "pg_ctl -l /var/log/postgresql/startup.log -w -D '/var/lib/postgresql/data' promote"
|
||||
server promoting
|
||||
INFO: reconnecting to promoted server
|
||||
NOTICE: STANDBY PROMOTE successful
|
||||
DETAIL: node 2 was successfully promoted to primary</programlisting>
|
||||
</para>
|
||||
<para>
|
||||
Executing <xref linkend="repmgr-cluster-show"> will show the current state; as there is now an
|
||||
active primary, the previous warning will not be displayed:
|
||||
<programlisting>
|
||||
$ repmgr -f /etc/repmgr.conf cluster show
|
||||
ID | Name | Role | Status | Upstream | Location | Connection string
|
||||
----+-------+---------+-----------+----------+----------+--------------------------------------
|
||||
1 | node1 | primary | - failed | | default | host=node1 dbname=repmgr user=repmgr
|
||||
2 | node2 | primary | * running | | default | host=node2 dbname=repmgr user=repmgr
|
||||
3 | node3 | standby | running | node1 | default | host=node3 dbname=repmgr user=repmgr</programlisting>
|
||||
</para>
|
||||
<para>
|
||||
However the sole remaining standby (<literal>node3</literal>) is still trying to replicate from the failed
|
||||
primary; <xref linkend="repmgr-standby-follow"> must now be executed to rectify this situation.
|
||||
</para>
|
||||
</chapter>
|
||||
|
||||
@@ -69,6 +69,7 @@
|
||||
|
||||
&configuration;
|
||||
&cloning-standbys;
|
||||
&promoting-standby;
|
||||
&command-reference;
|
||||
</part>
|
||||
|
||||
|
||||
Reference in New Issue
Block a user