repmgr/doc/appendix-faq.sgml

<appendix id="appendix-faq" xreflabel="FAQ">

  <title>FAQ (Frequently Asked Questions)</title>

  <indexterm>
   <primary>FAQ (Frequently Asked Questions)</primary>
  </indexterm>

 <sect1 id="faq-general" xreflabel="General">
  <title>General</title>

  <sect2 id="faq-xrepmgr-version-diff" xreflabel="Version differences">
    <title>What's the difference between the repmgr versions?</title>
    <para>
      &repmgr; 4 is a complete rewrite of the existing &repmgr; code base
      and implements &repmgr; as a PostgreSQL extension. It
      supports all PostgreSQL versions from 9.3 (although some &repmgr;
      features are not available for PostgreSQL 9.3 and 9.4).
     </para>
     <para>
      &repmgr; 3.x builds on the improved replication facilities added
      in PostgreSQL 9.3, as well as improved automated failover support
      via &repmgrd;, and is not compatible with PostgreSQL 9.2
      and earlier. We recommend upgrading to &repmgr; 4, as the &repmgr; 3.x
      series is no longer maintained.
     </para>
     <para>
      &repmgr; 2.x supports PostgreSQL 9.0 ~ 9.3. While it is compatible
      with PostgreSQL 9.3, we recommend using repmgr 4.x. &repmgr; 2.x is
      no longer maintained.
     </para>
     <para>
       See also <link linkend="install-compatibility-matrix">&repmgr; compatibility matrix</link>
       and <link linkend="faq-upgrade-repmgr">Should I upgrade &repmgr;?</link>.
     </para>
  </sect2>

  <sect2 id="faq-replication-slots-advantage" xreflabel="Advantages of replication slots">
   <title>What's the advantage of using replication slots?</title>
   <para>
    Replication slots, introduced in PostgreSQL 9.4, ensure that the
    primary server will retain WAL files until they have been consumed
    by all standby servers. This means standby servers should never
    fail due to not being able to retrieve required WAL files from the
    primary.
   </para>
   <para>
    However this does mean that if a standby is no longer connected to the
    primary, the presence of the replication slot will cause WAL files
    to be retained indefinitely, and eventually lead to disk space
    exhaustion.
   </para>

   <tip>
     <para>
       2ndQuadrant's recommended configuration is to configure
       <ulink url="https://www.pgbarman.org/">Barman</ulink> as a fallback
       source of WAL files, rather than maintain replication slots for
       each standby. See also: <link linkend="cloning-from-barman-restore-command">Using Barman as a WAL file source</link>.
     </para>
   </tip>
  </sect2>

  <sect2 id="faq-replication-slots-number" xreflabel="Number of replication slots">
   <title>How many replication slots should I define in <varname>max_replication_slots</varname>?</title>
   <para>
    Normally at least same number as the number of standbys which will connect
    to the node. Note that changes to <varname>max_replication_slots</varname> require a server
    restart to take effect, and as there is no particular penalty for unused
    replication slots, setting a higher figure will make adding new nodes
    easier.
   </para>
  </sect2>

  <sect2 id="faq-hash-index" xreflabel="Hash indexes">
   <title>Does &repmgr; support hash indexes?</title>
   <para>
    Before PostgreSQL 10, hash indexes were not WAL logged and are therefore not suitable
    for use in streaming replication in PostgreSQL 9.6 and earlier. See the
    <ulink url="https://www.postgresql.org/docs/9.6/sql-createindex.html#AEN80279">PostgreSQL documentation</ulink>
    for details.
   </para>
   <para>
    From PostgreSQL 10, this restriction has been lifted and hash indexes can be used
    in a streaming replication cluster.
   </para>
  </sect2>

  <sect2 id="faq-upgrades" xreflabel="Upgrading PostgreSQL with repmgr">
   <title>Can &repmgr; assist with upgrading a PostgreSQL cluster?</title>
   <para>
     For <emphasis>minor</emphasis> version upgrades, e.g. from 9.6.7 to 9.6.8, a common
     approach is to upgrade a standby to the latest version, perform a
     <link linkend="performing-switchover">switchover</link> promoting it to a primary,
     then upgrade the former primary.
   </para>
   <para>
     For <emphasis>major</emphasis> version upgrades (e.g. from PostgreSQL 9.6 to PostgreSQL 10),
     the traditional approach is to "reseed" a cluster by upgrading a single
     node with <ulink url="https://www.postgresql.org/docs/current/pgupgrade.html">pg_upgrade</ulink>
     and recloning standbys from this.
   </para>
   <para>
     To minimize downtime during major upgrades from PostgreSQL 9.4 and later,
     <ulink url="https://www.2ndquadrant.com/en/resources/pglogical/">pglogical</ulink>
     can be used to set up a parallel cluster using the newer PostgreSQL version,
     which can be kept in sync with the existing production cluster until the
     new cluster is ready to be put into production.
   </para>
  </sect2>

  <sect2 id="faq-libdir-repmgr-error">
   <title>What does this error mean: <literal>ERROR: could not access file "$libdir/repmgr"</literal>?</title>
   <para>
     It means the &repmgr; extension code is not installed in the
     PostgreSQL application directory. This typically happens when using PostgreSQL
     packages provided by a third-party vendor, which often have different
     filesystem layouts.
   </para>
   <para>
     Either use PostgreSQL packages provided by the community or 2ndQuadrant; if this
     is not possible, contact your vendor for assistance.
   </para>
  </sect2>

  <sect2 id="faq-old-packages">
   <title>How can I obtain old versions of &repmgr; packages?</title>
   <para>
     See appendix <xref linkend="packages-old-versions"/> for details.
   </para>
  </sect2>

  <sect2 id="faq-repmgr-required-for-replication">
    <title>Is &repmgr; required for streaming replication?</title>
    <para>
      No.
    </para>
    <para>
     &repmgr; (together with &repmgrd;) assists with
     <emphasis>managing</emphasis> replication. It does not actually perform replication, which
     is part of the core PostgreSQL functionality.
    </para>
  </sect2>

  <sect2 id="faq-what-if-repmgr-uninstalled">
   <title>Will replication stop working if &repmgr; is uninstalled?</title>
   <para>
     No. See preceding question.
   </para>
  </sect2>

  <sect2 id="faq-version-mix">
   <title>Does it matter if different &repmgr; versions are present in the replication cluster?</title>
   <para>
     Yes. If different &quot;major&quot; &repmgr; versions (e.g. 3.3.x and 4.1.x) are present,
     &repmgr; (in particular &repmgrd;)
     may not run, or run properly, or in the worst case (if different &repmgrd;
     versions are running and there are differences in the failover implementation) break
     your replication cluster.
   </para>
   <para>
     If different &quot;minor&quot; &repmgr; versions (e.g. 4.1.1 and 4.1.6) are installed,
     &repmgr; will function, but we strongly recommend always running the same version
     to ensure there are no unexpected suprises, e.g. a newer version behaving slightly
     differently to the older version.
   </para>
   <para>
     See also <link linkend="faq-upgrade-repmgr">Should I upgrade &repmgr;?</link>.
   </para>
  </sect2>

  <sect2 id="faq-upgrade-repmgr">
    <title>Should I upgrade &repmgr;?</title>
    <para>
      Yes.
    </para>
    <para>
      We don't release new versions for fun, you know. Upgrading may require a little effort,
      but running an older &repmgr; version with bugs which have since been fixed may end up
      costing you more effort. The same applies to PostgreSQL itself.
    </para>

  </sect2>

  <sect2 id="faq-repmgr-conf-data-directory">
    <title>Why do I need to specify the data directory location in repmgr.conf?</title>
    <para>
      In some circumstances &repmgr; may need to access a PostgreSQL data
      directory while the PostgreSQL server is not running, e.g. to confirm
      it shut down cleanly during a <link linkend="performing-switchover">switchover</link>.
    </para>
    <para>
      Additionally, this provides support when using &repmgr; on PostgreSQL 9.6 and
      earlier, where the <literal>repmgr</literal> user is not a superuser; in that
      case the <literal>repmgr</literal> user will not be able to access the
      <literal>data_directory</literal> configuration setting, access to which is restricted
      to superusers. (In PostgreSQL 10 and later, non-superusers can be added to the
      group <option>pg_read_all_settings</option> which will enable them to read this setting).
    </para>
  </sect2>
 </sect1>

 <sect1 id="faq-repmgr" xreflabel="repmgr">
  <title><command>repmgr</command></title>

  <sect2 id="faq-register-existing-node" xreflabel="registering an existing node">
   <title>Can I register an existing PostgreSQL server with repmgr?</title>
   <para>
    Yes, any existing PostgreSQL server which is part of the same replication
    cluster can be registered with &repmgr;. There's no requirement for a
    standby to have been cloned using &repmgr;.
   </para>
  </sect2>

  <sect2 id="faq-repmgr-clone-other-source" >
   <title>Can I use a standby not cloned by &repmgr; as a &repmgr; node?</title>

   <para>
     For a standby which has been manually cloned or recovered from an external
     backup manager such as Barman, the command
     <command><link linkend="repmgr-standby-clone">repmgr standby clone --recovery-conf-only</link></command>
     can be used to create the correct <filename>recovery.conf</filename> file for
     use with &repmgr; (and will create a replication slot if required). Once this has been done,
     <link linkend="repmgr-standby-register">register the node</link> as usual.
   </para>
  </sect2>

  <sect2 id="faq-repmgr-recovery-conf" >
    <title>What does &repmgr; write in <filename>recovery.conf</filename>, and what options can be set there?</title>
   <para>
     See section <link linkend="repmgr-standby-clone-recovery-conf">Customising recovery.conf</link>.
   </para>
  </sect2>

  <sect2 id="faq-repmgr-failed-primary-standby" xreflabel="Reintegrate a failed primary as a standby">
   <title>How can a failed primary be re-added as a standby?</title>
   <para>
    This is a two-stage process. First, the failed primary's data directory
    must be re-synced with the current primary; secondly the failed primary
    needs to be re-registered as a standby.
   </para>
   <para>
    It's possible to use <command>pg_rewind</command> to re-synchronise the existing data
    directory, which will usually be much
    faster than re-cloning the server. However <command>pg_rewind</command> can only
    be used if PostgreSQL either has <varname>wal_log_hints</varname> enabled, or
    data checksums were enabled when the cluster was initialized.
   </para>
   <para>
     Note that <command>pg_rewind</command> is available as part of the core PostgreSQL
     distribution from PostgreSQL 9.5, and as a third-party utility for PostgreSQL 9.3 and 9.4.
   </para>
   <para>
    &repmgr; provides the command <command>repmgr node rejoin</command> which can
    optionally execute <command>pg_rewind</command>; see the <xref linkend="repmgr-node-rejoin"/>
    documentation for details, in particular the section <xref linkend="repmgr-node-rejoin-pg-rewind"/>.
   </para>
   <para>
    If <command>pg_rewind</command> cannot be used, then the data directory will need
    to be re-cloned from scratch.
   </para>

  </sect2>

  <sect2 id="faq-repmgr-check-configuration" xreflabel="Check PostgreSQL configuration">
   <title>Is there an easy way to check my primary server is correctly configured for use with &repmgr;?</title>
   <para>
    Execute <command><link linkend="repmgr-standby-clone">repmgr standby clone</link></command>
    with the <literal>--dry-run</literal> option; this will report any configuration problems
    which need to be rectified.
   </para>
  </sect2>

  <sect2 id="faq-repmgr-clone-skip-config-files" xreflabel="">
   <title>When cloning a standby, how can I get &repmgr; to copy
     <filename>postgresql.conf</filename> and <filename>pg_hba.conf</filename> from the PostgreSQL configuration
     directory in <filename>/etc</filename>?</title>
   <para>
    Use the command line option <literal>--copy-external-config-files</literal>. For more details
    see <xref linkend="repmgr-standby-clone-config-file-copying"/>.
   </para>
  </sect2>

  <sect2 id="faq-repmgr-shared-preload-libaries-no-repmgrd" xreflabel="shared_preload_libraries without repmgrd">
    <title>Do I need to include <literal>shared_preload_libraries = 'repmgr'</literal>
      in <filename>postgresql.conf</filename> if I'm not using &repmgrd;?</title>
   <para>
    No, the <literal>repmgr</literal> shared library is only needed when running &repmgrd;.
    If you later decide to run &repmgrd;, you just need to add
    <literal>shared_preload_libraries = 'repmgr'</literal> and restart PostgreSQL.
   </para>
  </sect2>

  <sect2 id="faq-repmgr-permissions" xreflabel="Replication permission problems">
   <title>I've provided replication permission for the <literal>repmgr</literal> user in <filename>pg_hba.conf</filename>
     but <command>repmgr</command>/&repmgrd; complains it can't connect to the server... Why?</title>
   <para>
    <command>repmgr</command> and &repmgrd; need to be able to connect to the repmgr database
    with a normal connection to query metadata. The <literal>replication</literal> connection
    permission is for PostgreSQL's streaming replication (and doesn't  necessarily need to be the <literal>repmgr</literal> user).
   </para>
  </sect2>

  <sect2 id="faq-repmgr-clone-provide-primary-conninfo" xreflabel="Providing primary connection parameters">
   <title>When cloning a standby, why do I need to provide the connection parameters
     for the primary server on the command line, not in the configuration file?</title>
   <para>
    Cloning a standby is a one-time action; the role of the server being cloned
    from could change, so fixing it in the configuration file would create
    confusion. If &repmgr; needs to establish a connection to the primary
    server, it can retrieve this from the <literal>repmgr.nodes</literal> table on the local
    node, and if necessary scan the replication cluster until it locates the active primary.
   </para>
  </sect2>

  <sect2 id="faq-repmgr-clone-waldir-xlogdir" xreflabel="Providing a custom WAL directory">
   <title>When cloning a standby, how do I ensure the WAL files are placed in a custom directory?</title>
   <para>
     Provide the option <literal>--waldir</literal>  (<literal>--xlogdir</literal> in PostgreSQL 9.6
     and earlier) with the absolute path to the WAL directory in <varname>pg_basebackup_options</varname>.
     For more details see <xref linkend="cloning-advanced-pg-basebackup-options"/>.
   </para>
  </sect2>

  <sect2 id="faq-repmgr-events-no-fkey" xreflabel="No foreign key on node_id in repmgr.events">
   <title>Why is there no foreign key on the <literal>node_id</literal> column in the <literal>repmgr.events</literal>
     table?</title>
   <para>
     Under some circumstances event notifications can be generated for servers
     which have not yet been registered; it's also useful to retain a record
     of events which includes servers removed from the replication cluster
     which no longer have an entry in the <literal>repmgr.nodes</literal> table.
   </para>
  </sect2>

  <sect2 id="faq-repmgr-recovery-conf-quoted-values" xreflabel="Quoted values in recovery.conf">
    <title>Why are some values in <filename>recovery.conf</filename> surrounded by pairs of single quotes?</title>
    <para>
      This is to ensure that user-supplied values which are written as parameter values in <filename>recovery.conf</filename>
      are escaped correctly and do not cause errors when <filename>recovery.conf</filename> is parsed.
    </para>
    <para>
      The escaping is performed by an internal PostgreSQL routine, which leaves strings consisting
      of digits and alphabetical characters only as-is, but wraps everything else in pairs of single quotes,
      even if the string does not contain any characters which need escaping.
    </para>
  </sect2>


 </sect1>

 <sect1 id="faq-repmgrd" xreflabel="repmgrd">
  <title>&repmgrd;</title>


  <sect2 id="faq-repmgrd-prevent-promotion" xreflabel="Prevent standby from being promoted to primary">
   <title>How can I prevent a node from ever being promoted to primary?</title>
   <para>
     In <filename>repmgr.conf</filename>, set its priority to a value of <literal>0</literal>; apply the changed setting with
    <command><link linkend="repmgr-standby-register">repmgr standby register --force</link></command>.
   </para>
   <para>
    Additionally, if <varname>failover</varname> is set to <literal>manual</literal>, the node will never
    be considered as a promotion candidate.
   </para>
  </sect2>

  <sect2 id="faq-repmgrd-delayed-standby" xreflabel="Delayed standby support">
   <title>Does &repmgrd; support delayed standbys?</title>
   <para>
    &repmgrd; can monitor delayed standbys - those set up with
    <varname>recovery_min_apply_delay</varname> set to a non-zero value
    in <filename>recovery.conf</filename> - but as it's not currently possible
    to directly examine the value applied to the standby, &repmgrd;
    may not be able to properly evaluate the node as a promotion candidate.
   </para>
   <para>
    We recommend that delayed standbys are explicitly excluded from promotion
    by setting <varname>priority</varname> to <literal>0</literal> in
    <filename>repmgr.conf</filename>.
   </para>
   <para>
    Note that after registering a delayed standby, &repmgrd; will only start
    once the metadata added in the primary node has been replicated.
   </para>
  </sect2>

  <sect2 id="faq-repmgrd-logfile-rotate" xreflabel="repmgrd logfile rotation">
   <title>How can I get &repmgrd; to rotate its logfile?</title>
   <para>
     Configure your system's <literal>logrotate</literal> service to do this; see <xref linkend="repmgrd-log-rotation"/>.
   </para>

  </sect2>

  <sect2 id="faq-repmgrd-recloned-no-start" xreflabel="repmgrd not restarting after node cloned">
   <title>I've recloned a failed primary as a standby, but &repmgrd; refuses to start?</title>
   <para>
    Check you registered the standby after recloning. If unregistered, the standby
    cannot be considered as a promotion candidate even if <varname>failover</varname> is set to
    <literal>automatic</literal>, which is probably not what you want. &repmgrd; will start if
    <varname>failover</varname> is set to <literal>manual</literal> so the node's replication status can still
  be monitored, if desired.
   </para>
  </sect2>

  <sect2 id="faq-repmgrd-pg-bindir" xreflabel="repmgrd does not apply pg_bindir to promote_command or follow_command">
    <title>
      &repmgrd; ignores pg_bindir when executing <varname>promote_command</varname> or <varname>follow_command</varname>
    </title>
    <para>
      <varname>promote_command</varname> or <varname>follow_command</varname> can be user-defined scripts,
      so &repmgr; will not apply <option>pg_bindir</option> even if excuting &repmgr;. Always provide the full
      path; see <xref linkend="repmgrd-automatic-failover-configuration"/> for more details.
    </para>
  </sect2>

  <sect2 id="faq-repmgrd-startup-no-upstream" xreflabel="repmgrd does not start if upstream node is not running">
    <title>
      &repmgrd; aborts startup with the error "<literal>upstream node must be running before repmgrd can start</literal>"
    </title>
    <para>
      &repmgrd; does this to avoid starting up on a replication cluster
      which is not in a healthy state. If the upstream is unavailable, &repmgrd;
      may initiate a failover immediately after starting up, which could have unintended side-effects,
      particularly if &repmgrd; is not running on other nodes.
    </para>
    <para>
      In particular, it's possible that the node's local copy of the <literal>repmgr.nodes</literal> copy
      is out-of-date, which may lead to incorrect failover behaviour.
    </para>
    <para>
      The onus is therefore on the adminstrator to manually set the cluster to a stable, healthy state before
      starting &repmgrd;.
    </para>
  </sect2>

 </sect1>
</appendix>