repmgrd: add parameter "failover_delay"

This parameter is not documented and intended for use during testing.
It should not be used in production.

HISTORY

@@ -1,4 +1,4 @@
-5.1     2020-??-??
+5.1.0   2020-04-13
         repmgr: remove BDR 2.x support
         repmgr: don't query upstream's data directory (Ian)
         repmgr: rename --recovery-conf-only to --replication-conf-only (Ian)
@@ -14,6 +14,8 @@
         repmgr: improve "standby switchover" completion checks (Ian)
         repmgr: add replication configuration file ownership check to
           "standby switchover" (Ian)
+        repmgr: check the demotion candidate's registered repmgr.conf file can
+          be found (laixiong; GitHub 615)
         repmgr: consolidate replication connection code (Ian)
         repmgr: check permissions for "pg_promote()" and fall back to pg_ctl
           if necessary (Ian)

Makefile.in

@@ -11,6 +11,7 @@ EXTENSION = repmgr
 
 DATA = \
   repmgr--unpackaged--4.0.sql \
+  repmgr--unpackaged--5.1.sql \
   repmgr--4.0.sql \
   repmgr--4.0--4.1.sql \
   repmgr--4.1.sql \

configfile.h

@@ -29,7 +29,7 @@
 #define TARGET_TIMELINE_LATEST 0
 
 /*
- * This is defined src/include/utils.h, however it's not practical
+ * This is defined in src/include/utils.h, however it's not practical
  * to include that from a frontend application.
  */
 #define PG_AUTOCONF_FILENAME "postgresql.auto.conf"
@@ -189,6 +189,7 @@ typedef struct
 
 	/* undocumented test settings */
 	int			promote_delay;
+	int			failover_delay;
 } t_configuration_options;
 
 /*
@@ -243,8 +244,9 @@ typedef struct
 		/* barman settings */ \
 		"", "", "",	 \
 		/* rsync/ssh settings */ \
-		 "", "", \
+		"", "", \
 		/* undocumented test settings */ \
+		0, \
 		0 \
  }
 

configure

@@ -1,6 +1,6 @@
 #! /bin/sh
 # Guess values for system-dependent variables and create Makefiles.
-# Generated by GNU Autoconf 2.69 for repmgr 5.1.
+# Generated by GNU Autoconf 2.69 for repmgr 5.1.0.
 #
 # Report bugs to <repmgr@googlegroups.com>.
 #
@@ -582,8 +582,8 @@ MAKEFLAGS=
 # Identity of this package.
 PACKAGE_NAME='repmgr'
 PACKAGE_TARNAME='repmgr'
-PACKAGE_VERSION='5.1'
+PACKAGE_VERSION='5.1.0'
-PACKAGE_STRING='repmgr 5.1'
+PACKAGE_STRING='repmgr 5.1.0'
 PACKAGE_BUGREPORT='repmgr@googlegroups.com'
 PACKAGE_URL='https://repmgr.org/'
 
@@ -1181,7 +1181,7 @@ if test "$ac_init_help" = "long"; then
   # Omit some internal or obsolete options to make the list less imposing.
   # This message is too long to be a string in the A/UX 3.1 sh.
   cat <<_ACEOF
-\`configure' configures repmgr 5.1 to adapt to many kinds of systems.
+\`configure' configures repmgr 5.1.0 to adapt to many kinds of systems.
 
 Usage: $0 [OPTION]... [VAR=VALUE]...
 
@@ -1242,7 +1242,7 @@ fi
 
 if test -n "$ac_init_help"; then
   case $ac_init_help in
-     short | recursive ) echo "Configuration of repmgr 5.1:";;
+     short | recursive ) echo "Configuration of repmgr 5.1.0:";;
    esac
   cat <<\_ACEOF
 
@@ -1316,7 +1316,7 @@ fi
 test -n "$ac_init_help" && exit $ac_status
 if $ac_init_version; then
   cat <<\_ACEOF
-repmgr configure 5.1
+repmgr configure 5.1.0
 generated by GNU Autoconf 2.69
 
 Copyright (C) 2012 Free Software Foundation, Inc.
@@ -1335,7 +1335,7 @@ cat >config.log <<_ACEOF
 This file contains any messages produced by compilers while
 running configure, to aid debugging if configure makes a mistake.
 
-It was created by repmgr $as_me 5.1, which was
+It was created by repmgr $as_me 5.1.0, which was
 generated by GNU Autoconf 2.69.  Invocation command line was
 
   $ $0 $@
@@ -2487,7 +2487,7 @@ cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 # report actual input values of CONFIG_FILES etc. instead of their
 # values after options handling.
 ac_log="
-This file was extended by repmgr $as_me 5.1, which was
+This file was extended by repmgr $as_me 5.1.0, which was
 generated by GNU Autoconf 2.69.  Invocation command line was
 
   CONFIG_FILES    = $CONFIG_FILES
@@ -2550,7 +2550,7 @@ _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 ac_cs_config="`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`"
 ac_cs_version="\\
-repmgr config.status 5.1
+repmgr config.status 5.1.0
 configured by $0, generated by GNU Autoconf 2.69,
   with options \\"\$ac_cs_config\\"
 

configure.in

@@ -1,4 +1,4 @@
-AC_INIT([repmgr], [5.1], [repmgr@googlegroups.com], [repmgr], [https://repmgr.org/])
+AC_INIT([repmgr], [5.1.0], [repmgr@googlegroups.com], [repmgr], [https://repmgr.org/])
 
 AC_COPYRIGHT([Copyright (c) 2010-2020, 2ndQuadrant Ltd.])
 

dbutils.c

@@ -2951,7 +2951,7 @@ get_child_nodes(PGconn *conn, int node_id, NodeInfoList *node_list)
 					  "     WHERE n.upstream_node_id = %i ",
 					  node_id);
 
-		log_verbose(LOG_DEBUG, "get_active_sibling_node_records():\n%s", query.data);
+	log_verbose(LOG_DEBUG, "get_child_nodes():\n%s", query.data);
 
 	res = PQexec(conn, query.data);
 
@@ -5720,16 +5720,16 @@ get_node_replication_stats(PGconn *conn, t_node_info *node_info)
 
 
 NodeAttached
-is_downstream_node_attached(PGconn *conn, char *node_name)
+is_downstream_node_attached(PGconn *conn, char *node_name, char **node_state)
 {
 	PQExpBufferData query;
 	PGresult   *res = NULL;
-	int			c = 0;
+	const char *state = NULL;
 
 	initPQExpBuffer(&query);
 
 	appendPQExpBuffer(&query,
-					  " SELECT pg_catalog.count(*) "
+					  " SELECT pid, state "
 					  "   FROM pg_catalog.pg_stat_replication "
 					  "  WHERE application_name = '%s'",
 					  node_name);
@@ -5748,31 +5748,53 @@ is_downstream_node_attached(PGconn *conn, char *node_name)
 		return NODE_ATTACHED_UNKNOWN;
 	}
 
-	if (PQntuples(res) != 1)
+	termPQExpBuffer(&query);
-	{
+
-		log_verbose(LOG_WARNING, _("unexpected number of tuples (%i) returned"), PQntuples(res));
+	/*
+	 * If there's more than one entry in pg_stat_application, there's no
+	 * way we can reliably determine which one belongs to the node we're
+	 * checking, so there's nothing more we can do.
+	 */
+	if (PQntuples(res) > 1)
+	{
+		log_error(_("multiple entries with \"application_name\" set to  \"%s\" found in \"pg_stat_replication\""),
+				  node_name);
+		log_hint(_("verify that a unique node name is configured for each node"));
 
-		termPQExpBuffer(&query);
 		PQclear(res);
 
 		return NODE_ATTACHED_UNKNOWN;
 	}
 
-	c = atoi(PQgetvalue(res, 0, 0));
+	if (PQntuples(res) == 0)
-
-	termPQExpBuffer(&query);
-	PQclear(res);
-
-	if (c == 0)
 	{
-		log_verbose(LOG_WARNING, _("node \"%s\" not found in \"pg_stat_replication\""), node_name);
+		log_warning(_("node \"%s\" not found in \"pg_stat_replication\""), node_name);
+
+		PQclear(res);
 
 		return NODE_DETACHED;
 	}
 
-	if (c > 1)
+	state = PQgetvalue(res, 0, 1);
-		log_verbose(LOG_WARNING, _("multiple entries with \"application_name\" set to  \"%s\" found in \"pg_stat_replication\""),
+
-					node_name);
+	if (node_state != NULL)
+	{
+		*node_state = palloc0(strlen(state) + 1);
+		strncpy(*node_state, state, strlen(state));
+	}
+
+	if (strcmp(state, "streaming") != 0)
+	{
+		log_warning(_("node \"%s\" attached in state \"%s\""),
+					node_name,
+					state);
+
+		PQclear(res);
+
+		return NODE_NOT_ATTACHED;
+	}
+
+	PQclear(res);
 
 	return NODE_ATTACHED;
 }

dbutils.h

@@ -119,9 +119,14 @@ typedef enum
 
 typedef enum
 {
+	/* unable to query "pg_stat_replication" or other error */
 	NODE_ATTACHED_UNKNOWN = -1,
-	NODE_DETACHED,
+	/* node has record in "pg_stat_replication" and state is not "streaming" */
-	NODE_ATTACHED
+	NODE_ATTACHED,
+	/* node has record in "pg_stat_replication" but state is not "streaming" */
+	NODE_NOT_ATTACHED,
+	/* node has no record in "pg_stat_replication" */
+	NODE_DETACHED
 } NodeAttached;
 
 typedef enum
@@ -589,7 +594,7 @@ bool		get_replication_info(PGconn *conn, t_server_type node_type, ReplInfo *repl
 int			get_replication_lag_seconds(PGconn *conn);
 TimeLineID	get_node_timeline(PGconn *conn, char *timeline_id_str);
 void		get_node_replication_stats(PGconn *conn, t_node_info *node_info);
-NodeAttached is_downstream_node_attached(PGconn *conn, char *node_name);
+NodeAttached is_downstream_node_attached(PGconn *conn, char *node_name, char **node_state);
 void		set_upstream_last_seen(PGconn *conn, int upstream_node_id);
 int			get_upstream_last_seen(PGconn *conn, t_server_type node_type);
 

doc/Makefile

@@ -95,6 +95,7 @@ clean:
 	rm -f repmgr.html
 	rm -f repmgr-A4.pdf
 	rm -f repmgr-US.pdf
+	rm -f html/*
 
 maintainer-clean:
 	rm -rf html

doc/appendix-packages.xml

@@ -471,7 +471,7 @@ repmgr96-4.1.1-0.0git320.g5113ab0.1.el7.x86_64.rpm</programlisting>
       <title>Debian/Ubuntu</title>
       <para>
         An archive of old packages (<literal>3.3.2</literal> and later) for Debian/Ubuntu-based systems is available here:
-        <ulink url="http://atalia.postgresql.org/morgue/r/repmgr/">http://atalia.postgresql.org/morgue/r/repmgr/</ulink>
+        <ulink url="https://apt-archive.postgresql.org/">https://apt-archive.postgresql.org/</ulink>
       </para>
     </sect2>
 

doc/appendix-release-notes.xml

@@ -17,12 +17,12 @@
 
 <!-- remember to update the release date in ../repmgr_version.h.in -->
 
-  <sect1 id="release-5.1">
+  <sect1 id="release-5.1.0">
-    <title>Release 5.1</title>
+    <title id="release-current">Release 5.1.0</title>
-    <para><emphasis>?? ?? ??, 2020</emphasis></para>
+    <para><emphasis>Mon 13 April, 2020</emphasis></para>
 
     <para>
-      &repmgr; 5.1 is a major release.
+      &repmgr; 5.1.0 is a major release.
     </para>
     <para>
       For details on how to upgrade an existing &repmgr; installation, see
@@ -50,6 +50,23 @@
       <para>
         <itemizedlist>
 
+          <listitem>
+            <para>
+              The requirement that the &repmgr; user is a database superuser has been
+              removed as far as possible.
+            </para>
+            <para>
+              In theory, &repmgr; can be operated with a normal database user for managing
+              the &repmgr; database, and a separate replication user for managing replication
+              connections (and replication slots, if these are in use).
+            </para>
+            <para>
+              Some operations will still require superuser permissions, e.g. for issuing
+              a <command>CHECKPOINT</command> as par of a switchover operation; in this case
+              a valid superuser should be provided with the <option>-S</option>/<option>--superuser</option>
+              option.
+            </para>
+          </listitem>
 
           <listitem>
             <para>
@@ -73,6 +90,7 @@
               Improve logging and checking of potential failure situations.
             </para>
           </listitem>
+
           <listitem>
             <para>
               <link linkend="repmgr-standby-switchover"><command>repmgr standby switchover</command></link>:
@@ -82,12 +100,37 @@
               data directory.
             </para>
           </listitem>
+
+          <listitem>
+            <para>
+              <link linkend="repmgr-standby-switchover"><command>repmgr standby switchover</command></link>:
+              Provide additional information in <option>--dry-run mode</option> output.
+            </para>
+          </listitem>
+
+          <listitem>
+            <para>
+              <link linkend="repmgr-standby-switchover"><command>repmgr standby switchover</command></link>:
+              Checks that the demotion candidate's registered repmgr.conf file can be found, to
+              prevent confusing references to an incorrectly configured data directory. GitHub 615.
+            </para>
+          </listitem>
+
           <listitem>
             <para>
               <link linkend="repmgr-node-check"><command>repmgr node check</command></link>:
               accept option <option>-S</option>/<option>--superuser</option>. GitHub #621.
             </para>
           </listitem>
+
+          <listitem>
+            <para>
+              <link linkend="repmgr-node-check"><command>repmgr node check</command></link>:
+              add  <option>--upstream</option> option to check whether the node is attached
+              to the expected upstream node.
+            </para>
+          </listitem>
+
         </itemizedlist>
       </para>
     </sect2>
@@ -112,6 +155,15 @@
             </para>
           </listitem>
 
+
+          <listitem>
+            <para>
+              <link linkend="repmgr-standby-promote"><command>repmgr standby promote</command></link>:
+              in <option>--dry-run mode</option>, display promote command which will be executed.
+            </para>
+          </listitem>
+
+
           <listitem>
             <para>
               <link linkend="repmgr-standby-promote"><command>repmgr standby promote</command></link>
@@ -123,8 +175,17 @@
 
           <listitem>
             <para>
-              <link linkend="repmgr-standby-promote"><command>repmgr standby promote</command></link>:
+              <link linkend="repmgr-standby-switchover"><command>repmgr standby switchover</command></link>:
-              in PostgreSQL 12 and later, use <varname>service_promote_command</varname> if set.
+              check for demotion candidate reattachment as late as possible to avoid spurious failure
+              reports.
+            </para>
+          </listitem>
+
+
+          <listitem>
+            <para>
+              &repmgrd;: check for presence of <option>promote_command</option> and
+              <option>follow_command</option> on receipt of <literal>SIGHUP</literal>. GitHub 614.
             </para>
           </listitem>
 
@@ -150,7 +211,7 @@
   </sect1>
 
   <sect1 id="release-5.0">
-    <title id="release-current">Release 5.0</title>
+    <title>Release 5.0</title>
     <para><emphasis>Tue 15 October, 2019</emphasis></para>
 
     <para>

doc/configuration-file.xml

@@ -182,6 +182,14 @@ conninfo='host=node1 user=repmgr dbname=repmgr connect_timeout=2'</programlistin
    </itemizedlist>
   </para>
 
+  <para>
+    In examples provided in this documentation, it is assumed the configuration file is located
+    at <filename>/etc/repmgr.conf</filename>. If &repmgr; is installed from a package, the
+    configuration file will probably be located at another location specified by the packager;
+    see appendix <xref linkend="appendix-packages"/> for configuration file locations in
+    different packaging systems.
+  </para>
+
   <para>
    Note that if a file is explicitly specified with <literal>-f/--config-file</literal>,
    an error will be raised if it is not found or not readable, and no attempt will be made to
@@ -202,6 +210,61 @@ conninfo='host=node1 user=repmgr dbname=repmgr connect_timeout=2'</programlistin
       <filename>/path/to/repmgr.conf</filename>).
     </para>
    </note>
+  </sect2>
+
+  <sect2 id="configuration-file-postgresql-major-upgrades" xreflabel="configuration file and PostgreSQL major version upgrades">
+    <title>Configuration file and PostgreSQL major version upgrades</title>
+
+    <indexterm>
+      <primary>repmgr.conf</primary>
+      <secondary>ostgreSQL major version upgrades</secondary>
+    </indexterm>
+
+    <para>
+      When upgrading the PostgreSQL cluster to a new major version, <filename>repmgr.conf</filename>
+      will probably needed to be updated.
+    </para>
+    <para>
+      Usually <option>pg_bindir</option> and <option>data_directory</option> will need to be modified,
+      particularly if the default package locations are used, as these usually change.
+    </para>
+
+     <para>
+       It's also possible the location of <filename>repmgr.conf</filename> itself will change
+       (e.g. from <filename>/etc/repmgr/11/repmgr.conf</filename> to <filename>/etc/repmgr/12/repmgr.conf</filename>).
+       This is stored as part of the &repmgr; metadata and is used by &repmgr; to execute &repmgr; remotely
+       (e.g. during a <link linkend="performing-switchover">switchover operation</link>).
+     </para>
+     <para>
+       If the content and/or location of <filename>repmgr.conf</filename> has changed, the &repmgr; metadata
+       needs to be updated to reflect this. The &repmgr; metadata can be updated on each node with:
+       <itemizedlist spacing="compact" mark="bullet">
+         <listitem>
+           <simpara>
+             <link linkend="repmgr-primary-register">
+               <command>repmgr primary register --force -f /path/to/repmgr.conf</command>
+             </link>
+           </simpara>
+         </listitem>
+
+         <listitem>
+           <simpara>
+             <link linkend="repmgr-standby-register">
+               <command>repmgr standby register --force -f /path/to/repmgr.conf</command>
+             </link>
+           </simpara>
+         </listitem>
+
+
+         <listitem>
+           <simpara>
+             <link linkend="repmgr-witness-register">
+               <command>repmgr witness register --force -f /path/to/repmgr.conf -h primary_host</command>
+             </link>
+           </simpara>
+         </listitem>
+       </itemizedlist>
+     </para>
 
    </sect2>
  </sect1>

doc/configuration-password-management.xml

@@ -127,8 +127,31 @@ node2:5432:repmgr:repmgr:foo
 node2:5432:replication:repluser:foo
 node3:5432:repmgr:repmgr:foo
 node3:5432:replication:repluser:foo</programlisting>
-
+      If you are planning to use the <option>-S</option>/<option>--superuser</option> option,
+      there must also be an entry enabling the superuser to connect to the &repmgr; database.
+      Assuming the superuser is <literal>postgres</literal>, the file would look like this:
+        <programlisting>
+node1:5432:repmgr:repmgr:foo
+node1:5432:repmgr:postgres:foo
+node1:5432:replication:repluser:foo
+node2:5432:repmgr:repmgr:foo
+node2:5432:repmgr:postgres:foo
+node2:5432:replication:repluser:foo
+node3:5432:repmgr:repmgr:foo
+node3:5432:repmgr:postgres:foo
+node3:5432:replication:repluser:foo</programlisting>
     </para>
+
+    <para>
+      The <filename>~/.pgpass</filename> file can be simplified with the use of wildcards if
+      there is no requirement to restrict provision of passwords to particular hosts, ports
+      or databases. The preceding file could then be formatted like this:
+        <programlisting>
+*:*:*:repmgr:foo
+*:*:*:postgres:foo
+</programlisting>
+    </para>
+
     <note>
       <para>
         It's possible to specify an alternative location for the <filename>~/.pgpass</filename> file, either via
@@ -140,6 +163,11 @@ node3:5432:replication:repluser:foo</programlisting>
         location on all nodes, as when connecting to a remote node, the file referenced is the one on the
         local node.
       </para>
+      <para>
+        Additionally, you <emphasis>must</emphasis> specify the passfile location in <filename>repmgr.conf</filename>
+        with the <option>passfile</option> option so &repmgr; can write the correct path when creating the
+        <option>primary_conninfo</option> parameter for replication configuration on standbys.
+      </para>
     </note>
 
   </sect2>

doc/repmgr-node-rejoin.xml

@@ -43,7 +43,12 @@
       <programlisting>
       repmgr node rejoin -d '$conninfo'</programlisting>
 
-      where <literal>$conninfo</literal> is the conninfo string of any reachable node in the cluster.
+      where <literal>$conninfo</literal> is the PostgreSQL <literal>conninfo</literal> string of the
+      <emphasis>current</emphasis> primary node (or that of any reachable node in the cluster, but
+      <emphasis>not</emphasis> the local node). This is so that &repmgr; can fetch up-to-date information
+      about the current state of the cluster.
+    </para>
+    <para>
       <filename>repmgr.conf</filename> for the stopped node *must* be supplied explicitly if not
       otherwise available.
     </para>
@@ -283,7 +288,15 @@
       to execute <command>pg_rewind</command> to ensure the node can be rejoined successfully.
     </para>
 
-    <important>
+    <refsect2 id="repmgr-node-rejoin-pg-rewind-config-files" xreflabel="pg_rewind and configuration files">
+
+      <title><command>pg_rewind</command> and configuration file retention</title>
+
+      <indexterm>
+        <primary>pg_rewind</primary>
+        <secondary>configuration file retention</secondary>
+      </indexterm>
+
       <para>
         Be aware that if <command>pg_rewind</command> is executed and actually performs a
         rewind operation, any configuration files in the PostgreSQL data directory will be
@@ -291,17 +304,27 @@
       </para>
       <para>
         To prevent this happening, provide a comma-separated list of files to retain
-        using the <literal>--config-file</literal> command line option; the specified files
+        using the <option>--config-file</option> command line option; the specified files
         will be archived in a temporary directory (whose parent directory can be specified with
-        <literal>--config-archive-dir</literal>) and restored once the rewind operation is
+        <option>--config-archive-dir</option>, default: <filename>/tmp</filename>)
-        complete.
+        and restored once the rewind operation is complete.
       </para>
-    </important>
+    </refsect2>
 
-    <para>
+    <refsect2 id="repmgr-node-rejoin-pg-rewind-example" xreflabel="example using repmgr node rejoin and pg_rewind">
-      Example, first using <literal>--dry-run</literal>, then actually executing the
+
-      <literal>node rejoin command</literal>.
+      <title>Example using <command>repmgr node rejoin</command> and <command>pg_rewind</command></title>
-    <programlisting>
+
+      <indexterm>
+        <primary>pg_rewind</primary>
+        <secondary>configuration file retention</secondary>
+      </indexterm>
+
+
+      <para>
+        Example, first using <option>--dry-run</option>, then actually executing the
+        <literal>node rejoin command</literal>.
+        <programlisting>
     $ repmgr node rejoin -f /etc/repmgr.conf -d 'host=node3 dbname=repmgr user=repmgr' \
         --config-files=postgresql.local.conf,postgresql.conf --verbose --force-rewind --dry-run
     INFO: replication connection to the rejoin target node was successful
@@ -317,17 +340,17 @@
       pg_rewind -D '/var/lib/postgresql/data' --source-server='host=node3 dbname=repmgr user=repmgr'
     INFO: prerequisites for executing NODE REJOIN are met</programlisting>
 
-    <note>
+        <note>
-      <para>
+          <para>
-        If <option>--force-rewind</option> is used with the <option>--dry-run</option> option,
+            If <option>--force-rewind</option> is used with the <option>--dry-run</option> option,
-        this checks the prerequisites for using <application>pg_rewind</application>, but is
+            this checks the prerequisites for using <application>pg_rewind</application>, but is
-        not an absolute guarantee that actually executing <application>pg_rewind</application>
+            not an absolute guarantee that actually executing <application>pg_rewind</application>
-        will succeed. See also section <xref linkend="repmgr-node-rejoin-caveats"/> below.
+            will succeed. See also section <xref linkend="repmgr-node-rejoin-caveats"/> below.
-      </para>
+          </para>
 
-    </note>
+        </note>
 
-    <programlisting>
+        <programlisting>
     $ repmgr node rejoin -f /etc/repmgr.conf -d 'host=node3 dbname=repmgr user=repmgr' \
         --config-files=postgresql.local.conf,postgresql.conf --verbose --force-rewind
     NOTICE: pg_rewind execution required for this node to attach to rejoin target node 3
@@ -339,8 +362,8 @@
     NOTICE: starting server using "pg_ctl -l /var/log/postgres/startup.log -w -D '/var/lib/pgsql/data' start"
     NOTICE: NODE REJOIN successful
     DETAIL: node 2 is now attached to node 3</programlisting>
-    </para>
+      </para>
-
+    </refsect2>
   </refsect1>
 
   <refsect1 id="repmgr-node-rejoin-caveats" xreflabel="Caveats">
@@ -378,6 +401,51 @@
      is running in <option>--dry-run</option> mode.
    </para>
 
+   <warning>
+     <para>
+       In all current PostgreSQL versions (as of September 2020), <application>pg_rewind</application>
+       contains a corner-case bug which affects standbys in a very specific situation.
+     </para>
+     <para>
+       This situation occurs when a standby was shut down <emphasis>before</emphasis> its
+       primary node, and an attempt is made to attach this standby to another primary
+       in the same cluster (following a &quot;split brain&quot; situation where the standby
+       was connected to the wrong primary). In this case, &repmgr; will correctly determine
+       that <application>pg_rewind</application> should be executed, however
+       <application>pg_rewind</application> incorrectly decides that no action is necessary.
+     </para>
+     <para>
+       In this situation, &repmgr; will report something like:
+<programlisting>
+    NOTICE: pg_rewind execution required for this node to attach to rejoin target node 1
+    DETAIL: rejoin target server's timeline 3 forked off current database system timeline 2 before current recovery point 0/7019C10</programlisting>
+       but when executed, <application>pg_rewind</application> will report:
+<programlisting>
+    pg_rewind: servers diverged at WAL location 0/7015540 on timeline 2
+    pg_rewind: no rewind required</programlisting>
+       and if an attempt is made to attach the standby to the new primary, PostgreSQL logs on the standby
+       will contain errors like:
+<programlisting>
+    [2020-09-07 15:01:41 UTC]    LOG:  00000: replication terminated by primary server
+    [2020-09-07 15:01:41 UTC]    DETAIL:  End of WAL reached on timeline 2 at 0/7015540.
+    [2020-09-07 15:01:41 UTC]    LOG:  00000: new timeline 3 forked off current database system timeline 2 before current recovery point 0/7019C10</programlisting>
+     </para>
+     <para>
+       Currently it is not possible to resolve this situation using <application>pg_rewind</application>.
+       A <ulink url="https://www.postgresql.org/message-id/flat/CABvVfJU-LDWvoz4-Yow3Ay5LZYTuPD7eSjjE4kGyNZpXC6FrVQ@mail.gmail.com">patch</ulink>
+       has been submitted and will hopefully be included in a forthcoming PostgreSQL minor release.
+     </para>
+     <para>
+       As a workaround, start the primary server the standby was previously attached to,
+       and ensure the standby can be attached to it. If <application>pg_rewind</application> was actually executed,
+       it will have copied in the <filename>.history</filename> file from the target primary server; this must
+       be removed. <command>repmgr node rejoin</command> can then be used to attach the standby to the original
+       primary. Ensure any changes pending on the primary have propogated to the standby. Then shut down the primary
+       server <emphasis>first</emphasis>, before shutting down the standby. It should then be possible to
+       use <command>repmgr node rejoin</command> to attach the standby to the new primary.
+     </para>
+   </warning>
+
   </refsect1>
 
   <refsect1>

doc/repmgr-standby-promote.xml

@@ -95,7 +95,6 @@
       NOTICE: promoting standby to primary
       DETAIL: promoting server "node2" (ID: 2) using "pg_ctl -l /var/log/postgres/startup.log -w -D '/var/lib/postgres/data' promote"
       server promoting
-      DEBUG: setting node 2 as primary and marking existing primary as failed
       NOTICE: STANDBY PROMOTE successful
       DETAIL: server "node2" (ID: 2) was successfully promoted to primary</programlisting>
     </para>
@@ -170,6 +169,42 @@
         </listitem>
       </varlistentry>
 
+      <varlistentry>
+        <term><option>-F</option></term>
+        <term><option>--force</option></term>
+        <listitem>
+          <para>
+            Ignore warnings and continue anyway.
+          </para>
+          <para>
+            This option is relevant in the following situations if <option>--siblings-follow</option> was specified:
+            <itemizedlist spacing="compact" mark="bullet">
+              <listitem>
+                <simpara>
+                  If one or more sibling nodes was not reachable via SSH, the standby will be promoted anyway.
+                </simpara>
+              </listitem>
+              <listitem>
+                <simpara>
+                  If the promotion candidate has insufficient free walsenders to accomodate the standbys which will
+                  be attached to it, the standby will be promoted anyway.
+                </simpara>
+              </listitem>
+              <listitem>
+                <simpara>
+                  If replication slots are in use but the promotion candidate has insufficient free replication slots
+                  to accomodate the standbys which will be attached to it, the standby will be promoted anyway.
+                </simpara>
+              </listitem>
+            </itemizedlist>
+          </para>
+          <para>
+            Note that if the <option>-F</option>/<option>--force</option> option is used when any of the above
+            situations is encountered, the onus is on the user to manually resolve any resulting issues.
+          </para>
+        </listitem>
+      </varlistentry>
+
     </variablelist>
   </refsect1>
 

doc/repmgr-witness-register.xml

@@ -63,6 +63,34 @@
   </refsect1>
 
 
+
+  <refsect1>
+
+    <title>Options</title>
+   <variablelist>
+
+      <varlistentry>
+        <term><option>--dry-run</option></term>
+        <listitem>
+          <para>
+            Check prerequisites but don't actually register the witness
+          </para>
+        </listitem>
+      </varlistentry>
+
+
+      <varlistentry>
+       <term><option>-F</option>/<option>--force</option></term>
+        <listitem>
+          <para>
+            Overwrite an existing node record
+          </para>
+        </listitem>
+      </varlistentry>
+
+   </variablelist>
+  </refsect1>
+
   <refsect1 id="repmgr-witness-register-events">
     <title>Event notifications</title>
     <para>

doc/repmgrd-configuration.xml

@@ -15,9 +15,13 @@
   </para>
   <para>
     &repmgrd; can be configured to provide failover
-    capability in case the primary upstream node becomes unreachable, and/or
+    capability in case the primary or upstream node becomes unreachable, and/or
     provide monitoring data to the &repmgr; metadatabase.
   </para>
+  <para>
+    From &repmgr; 4.4, when running on the primary node, &repmgrd; can also monitor
+    standby disconnections/reconnections (see <xref linkend="repmgrd-primary-child-disconnection"/>).
+  </para>
 
   <sect1 id="repmgrd-basic-configuration">
     <title>repmgrd configuration</title>
@@ -583,7 +587,8 @@ repmgrd_service_stop_command='sudo systemctl repmgr12 stop'
         the option <option>monitor_interval_secs</option> (see above).
       </para>
       <para>
-        For more details on monitoring, see <xref linkend="repmgrd-monitoring"/>.
+        For more details on monitoring, see <xref linkend="repmgrd-monitoring"/>. For information on
+        monitoring standby disconnections, see <xref linkend="repmgrd-primary-child-disconnection"/>.
       </para>
     </sect2>
 

doc/upgrading-repmgr.xml

@@ -201,9 +201,13 @@ ALTER EXTENSION repmgr UPDATE</programlisting>
 	</para>
 	<tip>
 	  <para>
-		If the &repmgr; upgrade requires a PostgreSQL restart, combine the &repmgr; upgrade
+        If the &repmgr; upgrade requires a PostgreSQL restart, combine the &repmgr; upgrade
-		with a PostgreSQL minor version upgrade, which will require a restart in any case.
+        with a PostgreSQL minor version upgrade, which will require a restart in any case.
-		New PostgreSQL minor version are usually released every couple of months.
+      </para>
+      <para>
+		New PostgreSQL minor versions are usually released every couple of months;
+        see the <ulink url="https://www.postgresql.org/developer/roadmap/">Roadmap</ulink>
+        for the current schedule.
 	  </para>
 	</tip>
   </sect2>

repmgr--unpackaged--5.1.sql

@@ -0,0 +1,265 @@
+-- complain if script is sourced in psql, rather than via CREATE EXTENSION
+\echo Use "CREATE EXTENSION repmgr" to load this file. \quit
+
+-- extract the current schema name
+-- NOTE: this assumes there will be only one schema matching 'repmgr_%';
+-- user is responsible for ensuring this is the case
+
+CREATE TEMPORARY TABLE repmgr_old_schema (schema_name TEXT);
+INSERT INTO repmgr_old_schema (schema_name)
+SELECT nspname AS schema_name
+  FROM pg_catalog.pg_namespace
+ WHERE nspname LIKE 'repmgr_%'
+ LIMIT 1;
+
+-- move old objects into new schema
+DO $repmgr$
+DECLARE
+  old_schema TEXT;
+BEGIN
+  SELECT schema_name FROM repmgr_old_schema
+    INTO old_schema;
+  EXECUTE format('ALTER TABLE %I.repl_nodes SET SCHEMA repmgr', old_schema);
+  EXECUTE format('ALTER TABLE %I.repl_events SET SCHEMA repmgr', old_schema);
+  EXECUTE format('ALTER TABLE %I.repl_monitor SET SCHEMA repmgr', old_schema);
+  EXECUTE format('DROP VIEW IF EXISTS %I.repl_show_nodes', old_schema);
+  EXECUTE format('DROP VIEW IF EXISTS %I.repl_status', old_schema);
+END$repmgr$;
+
+-- convert "repmgr_$cluster.repl_nodes" to "repmgr.nodes"
+CREATE TABLE repmgr.nodes (
+  node_id          INTEGER     PRIMARY KEY,
+  upstream_node_id INTEGER     NULL REFERENCES repmgr.nodes (node_id) DEFERRABLE,
+  active           BOOLEAN     NOT NULL DEFAULT TRUE,
+  node_name        TEXT        NOT NULL,
+  type             TEXT        NOT NULL CHECK (type IN('primary','standby','witness','bdr')),
+  location         TEXT        NOT NULL DEFAULT 'default',
+  priority         INT         NOT NULL DEFAULT 100,
+  conninfo         TEXT        NOT NULL,
+  repluser         VARCHAR(63) NOT NULL,
+  slot_name        TEXT        NULL,
+  config_file      TEXT        NOT NULL
+);
+
+INSERT INTO repmgr.nodes
+  (node_id, upstream_node_id, active, node_name, type, location, priority, conninfo, repluser, slot_name, config_file)
+SELECT id, upstream_node_id, active, name,
+       CASE WHEN type = 'master' THEN 'primary' ELSE type END,
+       'default', priority, conninfo, 'unknown', slot_name, 'unknown'
+  FROM repmgr.repl_nodes
+ ORDER BY id;
+
+
+-- convert "repmgr_$cluster.repl_event" to "event"
+
+ALTER TABLE repmgr.repl_events RENAME TO events;
+
+-- create new table "repmgr.voting_term"
+CREATE TABLE repmgr.voting_term (
+  term INT NOT NULL
+);
+
+CREATE UNIQUE INDEX voting_term_restrict
+ON repmgr.voting_term ((TRUE));
+
+CREATE RULE voting_term_delete AS
+   ON DELETE TO repmgr.voting_term
+   DO INSTEAD NOTHING;
+
+INSERT INTO repmgr.voting_term (term) VALUES (1);
+
+
+-- convert "repmgr_$cluster.repl_monitor" to "monitoring_history"
+
+
+DO $repmgr$
+DECLARE
+  DECLARE server_version_num INT;
+BEGIN
+  SELECT setting
+    FROM pg_catalog.pg_settings
+   WHERE name = 'server_version_num'
+    INTO server_version_num;
+  IF server_version_num >= 90400 THEN
+    EXECUTE $repmgr_func$
+CREATE TABLE repmgr.monitoring_history (
+  primary_node_id                INTEGER NOT NULL,
+  standby_node_id                INTEGER NOT NULL,
+  last_monitor_time              TIMESTAMP WITH TIME ZONE NOT NULL,
+  last_apply_time                TIMESTAMP WITH TIME ZONE,
+  last_wal_primary_location      PG_LSN NOT NULL,
+  last_wal_standby_location      PG_LSN,
+  replication_lag                BIGINT NOT NULL,
+  apply_lag                      BIGINT NOT NULL
+)
+    $repmgr_func$;
+    INSERT INTO repmgr.monitoring_history
+      (primary_node_id, standby_node_id, last_monitor_time,  last_apply_time, last_wal_primary_location, last_wal_standby_location, replication_lag, apply_lag)
+    SELECT primary_node, standby_node, last_monitor_time,  last_apply_time, last_wal_primary_location::pg_lsn, last_wal_standby_location::pg_lsn, replication_lag, apply_lag
+      FROM repmgr.repl_monitor;
+  ELSE
+    EXECUTE $repmgr_func$
+CREATE TABLE repmgr.monitoring_history (
+  primary_node_id                INTEGER NOT NULL,
+  standby_node_id                INTEGER NOT NULL,
+  last_monitor_time              TIMESTAMP WITH TIME ZONE NOT NULL,
+  last_apply_time                TIMESTAMP WITH TIME ZONE,
+  last_wal_primary_location      TEXT NOT NULL,
+  last_wal_standby_location      TEXT,
+  replication_lag                BIGINT NOT NULL,
+  apply_lag                      BIGINT NOT NULL
+)
+    $repmgr_func$;
+    INSERT INTO repmgr.monitoring_history
+      (primary_node_id, standby_node_id, last_monitor_time,  last_apply_time, last_wal_primary_location, last_wal_standby_location, replication_lag, apply_lag)
+    SELECT primary_node, standby_node, last_monitor_time,  last_apply_time, last_wal_primary_location, last_wal_standby_location, replication_lag, apply_lag
+      FROM repmgr.repl_monitor;
+
+  END IF;
+END$repmgr$;
+
+CREATE INDEX idx_monitoring_history_time
+          ON repmgr.monitoring_history (last_monitor_time, standby_node_id);
+
+CREATE VIEW repmgr.show_nodes AS
+   SELECT n.node_id,
+          n.node_name,
+          n.active,
+          n.upstream_node_id,
+          un.node_name AS upstream_node_name,
+          n.type,
+          n.priority,
+          n.conninfo
+     FROM repmgr.nodes n
+LEFT JOIN repmgr.nodes un
+       ON un.node_id = n.upstream_node_id;
+
+
+/* ================= */
+/* repmgrd functions */
+/* ================= */
+
+/* monitoring functions */
+
+CREATE FUNCTION set_local_node_id(INT)
+  RETURNS VOID
+  AS 'MODULE_PATHNAME', 'set_local_node_id'
+  LANGUAGE C STRICT;
+
+CREATE FUNCTION get_local_node_id()
+  RETURNS INT
+  AS 'MODULE_PATHNAME', 'get_local_node_id'
+  LANGUAGE C STRICT;
+
+CREATE FUNCTION standby_set_last_updated()
+  RETURNS TIMESTAMP WITH TIME ZONE
+  AS 'MODULE_PATHNAME', 'standby_set_last_updated'
+  LANGUAGE C STRICT;
+
+CREATE FUNCTION standby_get_last_updated()
+  RETURNS TIMESTAMP WITH TIME ZONE
+  AS 'MODULE_PATHNAME', 'standby_get_last_updated'
+  LANGUAGE C STRICT;
+
+CREATE FUNCTION set_upstream_last_seen(INT)
+  RETURNS VOID
+  AS 'MODULE_PATHNAME', 'set_upstream_last_seen'
+  LANGUAGE C STRICT;
+
+CREATE FUNCTION get_upstream_last_seen()
+  RETURNS INT
+  AS 'MODULE_PATHNAME', 'get_upstream_last_seen'
+  LANGUAGE C STRICT;
+
+CREATE FUNCTION get_upstream_node_id()
+  RETURNS INT
+  AS 'MODULE_PATHNAME', 'get_upstream_node_id'
+  LANGUAGE C STRICT;
+
+CREATE FUNCTION set_upstream_node_id(INT)
+  RETURNS VOID
+  AS 'MODULE_PATHNAME', 'set_upstream_node_id'
+  LANGUAGE C STRICT;
+
+/* failover functions */
+
+CREATE FUNCTION notify_follow_primary(INT)
+  RETURNS VOID
+  AS 'MODULE_PATHNAME', 'notify_follow_primary'
+  LANGUAGE C STRICT;
+
+CREATE FUNCTION get_new_primary()
+  RETURNS INT
+  AS 'MODULE_PATHNAME', 'get_new_primary'
+  LANGUAGE C STRICT;
+
+CREATE FUNCTION reset_voting_status()
+  RETURNS VOID
+  AS 'MODULE_PATHNAME', 'reset_voting_status'
+  LANGUAGE C STRICT;
+
+CREATE FUNCTION get_repmgrd_pid()
+  RETURNS INT
+  AS 'MODULE_PATHNAME', 'get_repmgrd_pid'
+  LANGUAGE C STRICT;
+
+CREATE FUNCTION get_repmgrd_pidfile()
+  RETURNS TEXT
+  AS 'MODULE_PATHNAME', 'get_repmgrd_pidfile'
+  LANGUAGE C STRICT;
+
+CREATE FUNCTION set_repmgrd_pid(INT, TEXT)
+  RETURNS VOID
+  AS 'MODULE_PATHNAME', 'set_repmgrd_pid'
+  LANGUAGE C CALLED ON NULL INPUT;
+
+CREATE FUNCTION repmgrd_is_running()
+  RETURNS BOOL
+  AS 'MODULE_PATHNAME', 'repmgrd_is_running'
+  LANGUAGE C STRICT;
+
+CREATE FUNCTION repmgrd_pause(BOOL)
+  RETURNS VOID
+  AS 'MODULE_PATHNAME', 'repmgrd_pause'
+  LANGUAGE C STRICT;
+
+CREATE FUNCTION repmgrd_is_paused()
+  RETURNS BOOL
+  AS 'MODULE_PATHNAME', 'repmgrd_is_paused'
+  LANGUAGE C STRICT;
+
+CREATE FUNCTION get_wal_receiver_pid()
+  RETURNS INT
+  AS 'MODULE_PATHNAME', 'get_wal_receiver_pid'
+  LANGUAGE C STRICT;
+
+
+/* views */
+
+CREATE VIEW repmgr.replication_status AS
+  SELECT m.primary_node_id, m.standby_node_id, n.node_name AS standby_name,
+ 	     n.type AS node_type, n.active, last_monitor_time,
+         CASE WHEN n.type='standby' THEN m.last_wal_primary_location ELSE NULL END AS last_wal_primary_location,
+         m.last_wal_standby_location,
+         CASE WHEN n.type='standby' THEN pg_catalog.pg_size_pretty(m.replication_lag) ELSE NULL END AS replication_lag,
+         CASE WHEN n.type='standby' THEN
+           CASE WHEN replication_lag > 0 THEN age(now(), m.last_apply_time) ELSE '0'::INTERVAL END
+           ELSE NULL
+         END AS replication_time_lag,
+         CASE WHEN n.type='standby' THEN pg_catalog.pg_size_pretty(m.apply_lag) ELSE NULL END AS apply_lag,
+         AGE(NOW(), CASE WHEN pg_catalog.pg_is_in_recovery() THEN repmgr.standby_get_last_updated() ELSE m.last_monitor_time END) AS communication_time_lag
+    FROM repmgr.monitoring_history m
+    JOIN repmgr.nodes n ON m.standby_node_id = n.node_id
+   WHERE (m.standby_node_id, m.last_monitor_time) IN (
+	          SELECT m1.standby_node_id, MAX(m1.last_monitor_time)
+			    FROM repmgr.monitoring_history m1 GROUP BY 1
+         );
+
+
+
+/* drop old tables */
+DROP TABLE repmgr.repl_nodes;
+DROP TABLE repmgr.repl_monitor;
+
+-- remove temporary table
+DROP TABLE repmgr_old_schema;

repmgr-action-cluster.c

@@ -55,10 +55,8 @@ typedef enum
 struct ColHeader headers_show[SHOW_HEADER_COUNT];
 struct ColHeader headers_event[EVENT_HEADER_COUNT];
 
-
+static int	build_cluster_matrix(t_node_matrix_rec ***matrix_rec_dest, ItemList *warnings, int *error_code);
-
+static int	build_cluster_crosscheck(t_node_status_cube ***cube_dest, ItemList *warnings, int *error_code);
-static int	build_cluster_matrix(t_node_matrix_rec ***matrix_rec_dest, int *name_length, ItemList *warnings, int *error_code);
-static int	build_cluster_crosscheck(t_node_status_cube ***cube_dest, int *name_length, ItemList *warnings, int *error_code);
 static void cube_set_node_status(t_node_status_cube **cube, int n, int node_id, int matrix_node_id, int connection_node_id, int connection_status);
 
 /*
@@ -538,9 +536,6 @@ do_cluster_crosscheck(void)
 {
 	int			i = 0,
 				n = 0;
-	char		c;
-	const char *node_header = "Name";
-	int			name_length = strlen(node_header);
 
 	t_node_status_cube **cube;
 
@@ -548,7 +543,7 @@ do_cluster_crosscheck(void)
 	int			error_code = SUCCESS;
 	ItemList	warnings = {NULL, NULL};
 
-	n = build_cluster_crosscheck(&cube, &name_length, &warnings, &error_code);
+	n = build_cluster_crosscheck(&cube, &warnings, &error_code);
 
 	if (runtime_options.output_mode == OM_CSV)
 	{
@@ -582,24 +577,56 @@ do_cluster_crosscheck(void)
 	}
 	else
 	{
-		printf("%*s | Id ", name_length, node_header);
+		/* output header contains node name, node ID and one column for each node in the cluster */
-		for (i = 0; i < n; i++)
+		struct ColHeader *headers_crosscheck = NULL;
-			printf("| %2d ", cube[i]->node_id);
+
-		printf("\n");
+		int header_count = n + 2;
+		int header_id = 2;
+
+		headers_crosscheck = palloc0(sizeof(ColHeader) * header_count);
+
+		/* Initialize column headers  */
+		strncpy(headers_crosscheck[0].title, _("Name"), MAXLEN);
+		strncpy(headers_crosscheck[1].title, _("ID"), MAXLEN);
 
-		for (i = 0; i < name_length; i++)
-			printf("-");
-		printf("-+----");
 		for (i = 0; i < n; i++)
-			printf("+----");
+		{
-		printf("\n");
+			maxlen_snprintf(headers_crosscheck[header_id].title, "%i", cube[i]->node_id);
+			header_id++;
+		}
+
+		/* Initialize column max values */
+		for (i = 0; i < header_count; i++)
+		{
+			headers_crosscheck[i].display = true;
+			headers_crosscheck[i].max_length = strlen(headers_crosscheck[i].title);
+			headers_crosscheck[i].cur_length = headers_crosscheck[i].max_length;
+
+			/* We can derive the maximum node ID length for the ID column from
+			 * the generated matrix node ID headers
+			 */
+			if (i >= 2 && headers_crosscheck[i].max_length > headers_crosscheck[1].max_length)
+				headers_crosscheck[1].max_length = headers_crosscheck[i].max_length;
+		}
+
+		for (i = 0; i < n; i++)
+		{
+			if (strlen(cube[i]->node_name) > headers_crosscheck[0].max_length)
+			{
+				headers_crosscheck[0].max_length = strlen(cube[i]->node_name);
+			}
+		}
+
+		print_status_header(header_count, headers_crosscheck);
 
 		for (i = 0; i < n; i++)
 		{
 			int			column_node_ix;
 
-			printf("%*s | %2d ", name_length,
+			printf(" %-*s | %-*i ",
+				   headers_crosscheck[0].max_length,
 				   cube[i]->node_name,
+				   headers_crosscheck[1].max_length,
 				   cube[i]->node_id);
 
 			for (column_node_ix = 0; column_node_ix < n; column_node_ix++)
@@ -607,6 +634,8 @@ do_cluster_crosscheck(void)
 				int			max_node_status = -2;
 				int			node_ix = 0;
 
+				char		c;
+
 				/*
 				 * The value of entry (i,j) is equal to the maximum value of all
 				 * the (i,j,k). Indeed:
@@ -646,7 +675,7 @@ do_cluster_crosscheck(void)
 						exit(ERR_INTERNAL);
 				}
 
-				printf("|  %c ", c);
+				printf("| %-*c ", headers_crosscheck[column_node_ix + 2].max_length, c);
 			}
 
 			printf("\n");
@@ -708,16 +737,13 @@ do_cluster_matrix()
 				j = 0,
 				n = 0;
 
-	const char *node_header = "Name";
-	int			name_length = strlen(node_header);
-
 	t_node_matrix_rec **matrix_rec_list;
 
 	bool		connection_error_found = false;
 	int			error_code = SUCCESS;
 	ItemList	warnings = {NULL, NULL};
 
-	n = build_cluster_matrix(&matrix_rec_list, &name_length, &warnings, &error_code);
+	n = build_cluster_matrix(&matrix_rec_list, &warnings, &error_code);
 
 	if (runtime_options.output_mode == OM_CSV)
 	{
@@ -740,27 +766,60 @@ do_cluster_matrix()
 	}
 	else
 	{
-		char		c;
+		/* output header contains node name, node ID and one column for each node in the cluster */
+		struct ColHeader *headers_matrix = NULL;
 
-		printf("%*s | Id ", name_length, node_header);
+		int header_count = n + 2;
-		for (i = 0; i < n; i++)
+		int header_id = 2;
-			printf("| %2d ", matrix_rec_list[i]->node_id);
-		printf("\n");
 
-		for (i = 0; i < name_length; i++)
+		headers_matrix = palloc0(sizeof(ColHeader) * header_count);
-			printf("-");
+
-		printf("-+----");
+		/* Initialize column headers  */
-		for (i = 0; i < n; i++)
+		strncpy(headers_matrix[0].title, _("Name"), MAXLEN);
-			printf("+----");
+		strncpy(headers_matrix[1].title, _("ID"), MAXLEN);
-		printf("\n");
 
 		for (i = 0; i < n; i++)
 		{
-			printf("%*s | %2d ", name_length,
+			maxlen_snprintf(headers_matrix[header_id].title, "%i", matrix_rec_list[i]->node_id);
+			header_id++;
+		}
+
+		/* Initialize column max values */
+		for (i = 0; i < header_count; i++)
+		{
+			headers_matrix[i].display = true;
+			headers_matrix[i].max_length = strlen(headers_matrix[i].title);
+			headers_matrix[i].cur_length = headers_matrix[i].max_length;
+
+			/* We can derive the maximum node ID length for the ID column from
+			 * the generated matrix node ID headers
+			 */
+			if (i >= 2 && headers_matrix[i].max_length > headers_matrix[1].max_length)
+				headers_matrix[1].max_length = headers_matrix[i].max_length;
+		}
+
+
+		for (i = 0; i < n; i++)
+		{
+			if (strlen(matrix_rec_list[i]->node_name) > headers_matrix[0].max_length)
+			{
+				headers_matrix[0].max_length = strlen(matrix_rec_list[i]->node_name);
+			}
+		}
+
+		print_status_header(header_count, headers_matrix);
+
+		for (i = 0; i < n; i++)
+		{
+			printf(" %-*s | %-*i ",
+				   headers_matrix[0].max_length,
 				   matrix_rec_list[i]->node_name,
+				   headers_matrix[1].max_length,
 				   matrix_rec_list[i]->node_id);
 			for (j = 0; j < n; j++)
 			{
+				char		c;
+
 				switch (matrix_rec_list[i]->node_status_list[j]->node_status)
 				{
 					case -2:
@@ -778,7 +837,7 @@ do_cluster_matrix()
 						exit(ERR_INTERNAL);
 				}
 
-				printf("|  %c ", c);
+				printf("| %-*c ", headers_matrix[j + 2].max_length, c);
 			}
 			printf("\n");
 		}
@@ -838,7 +897,7 @@ matrix_set_node_status(t_node_matrix_rec **matrix_rec_list, int n, int node_id,
 
 
 static int
-build_cluster_matrix(t_node_matrix_rec ***matrix_rec_dest, int *name_length, ItemList *warnings, int *error_code)
+build_cluster_matrix(t_node_matrix_rec ***matrix_rec_dest, ItemList *warnings, int *error_code)
 {
 	PGconn	   *conn = NULL;
 	int			i = 0,
@@ -896,7 +955,6 @@ build_cluster_matrix(t_node_matrix_rec ***matrix_rec_dest, int *name_length, Ite
 	/* Initialise matrix structure for each node */
 	for (cell = nodes.head; cell; cell = cell->next)
 	{
-		int			name_length_cur;
 		NodeInfoListCell *cell_j;
 
 		matrix_rec_list[i] = (t_node_matrix_rec *) pg_malloc0(sizeof(t_node_matrix_rec));
@@ -906,13 +964,6 @@ build_cluster_matrix(t_node_matrix_rec ***matrix_rec_dest, int *name_length, Ite
 				cell->node_info->node_name,
 				sizeof(matrix_rec_list[i]->node_name));
 
-		/*
-		 * Find the maximum length of a node name
-		 */
-		name_length_cur = strlen(matrix_rec_list[i]->node_name);
-		if (name_length_cur > *name_length)
-			*name_length = name_length_cur;
-
 		matrix_rec_list[i]->node_status_list = (t_node_status_rec **) pg_malloc0(sizeof(t_node_status_rec) * nodes.node_count);
 
 		j = 0;
@@ -1077,7 +1128,7 @@ build_cluster_matrix(t_node_matrix_rec ***matrix_rec_dest, int *name_length, Ite
 
 
 static int
-build_cluster_crosscheck(t_node_status_cube ***dest_cube, int *name_length, ItemList *warnings, int *error_code)
+build_cluster_crosscheck(t_node_status_cube ***dest_cube, ItemList *warnings, int *error_code)
 {
 	PGconn	   *conn = NULL;
 	int			h,
@@ -1126,20 +1177,12 @@ build_cluster_crosscheck(t_node_status_cube ***dest_cube, int *name_length, Item
 
 	for (cell = nodes.head; cell; cell = cell->next)
 	{
-		int			name_length_cur = 0;
 		NodeInfoListCell *cell_i = NULL;
 
 		cube[h] = (t_node_status_cube *) pg_malloc(sizeof(t_node_status_cube));
 		cube[h]->node_id = cell->node_info->node_id;
 		strncpy(cube[h]->node_name, cell->node_info->node_name, sizeof(cube[h]->node_name));
 
-		/*
-		 * Find the maximum length of a node name
-		 */
-		name_length_cur = strlen(cube[h]->node_name);
-		if (name_length_cur > *name_length)
-			*name_length = name_length_cur;
-
 		cube[h]->matrix_list_rec = (t_node_matrix_rec **) pg_malloc(sizeof(t_node_matrix_rec) * nodes.node_count);
 
 		i = 0;

repmgr-action-node.c

@@ -43,7 +43,7 @@ static void _do_node_restore_config(void);
 
 static void do_node_check_replication_connection(void);
 static CheckStatus do_node_check_archive_ready(PGconn *conn, OutputMode mode, CheckStatusList *list_output);
-static CheckStatus do_node_check_downstream(PGconn *conn, OutputMode mode, CheckStatusList *list_output);
+static CheckStatus do_node_check_downstream(PGconn *conn, OutputMode mode, t_node_info *node_info, CheckStatusList *list_output);
 static CheckStatus do_node_check_upstream(PGconn *conn, OutputMode mode, t_node_info *node_info, CheckStatusList *list_output);
 static CheckStatus do_node_check_replication_lag(PGconn *conn, OutputMode mode, t_node_info *node_info, CheckStatusList *list_output);
 static CheckStatus do_node_check_role(PGconn *conn, OutputMode mode, t_node_info *node_info, CheckStatusList *list_output);
@@ -204,7 +204,16 @@ do_node_status(void)
 
 		if (enabled == false && recovery_type == RECTYPE_STANDBY)
 		{
-			appendPQExpBufferStr(&archiving_status, " (on standbys \"archive_mode\" must be set to \"always\" to be effective)");
+			if (PQserverVersion(conn) >= 90500)
+			{
+				appendPQExpBufferStr(&archiving_status,
+									 " (on standbys \"archive_mode\" must be set to \"always\" to be effective)");
+			}
+			else
+			{
+				appendPQExpBufferStr(&archiving_status,
+									 " (\"archive_mode\" has no effect on standbys)");
+			}
 		}
 
 		key_value_list_set(&node_status,
@@ -294,7 +303,7 @@ do_node_status(void)
 				continue;
 			}
 
-			if (is_downstream_node_attached(conn, node_cell->node_info->node_name) != NODE_ATTACHED)
+			if (is_downstream_node_attached(conn, node_cell->node_info->node_name, NULL) != NODE_ATTACHED)
 			{
 				missing_nodes_count++;
 				item_list_append_format(&missing_nodes,
@@ -797,6 +806,7 @@ do_node_check(void)
 	{
 		return_code = do_node_check_downstream(conn,
 											   runtime_options.output_mode,
+											   &node_info,
 											   NULL);
 		PQfinish(conn);
 		exit(return_code);
@@ -888,7 +898,7 @@ do_node_check(void)
 	if (do_node_check_upstream(conn, runtime_options.output_mode, &node_info, &status_list) != CHECK_STATUS_OK)
 		issue_detected = true;
 
-	if (do_node_check_downstream(conn, runtime_options.output_mode, &status_list) != CHECK_STATUS_OK)
+	if (do_node_check_downstream(conn, runtime_options.output_mode, &node_info, &status_list) != CHECK_STATUS_OK)
 		issue_detected = true;
 
 	if (do_node_check_slots(conn, runtime_options.output_mode, &node_info, &status_list) != CHECK_STATUS_OK)
@@ -1183,7 +1193,7 @@ do_node_check_archive_ready(PGconn *conn, OutputMode mode, CheckStatusList *list
 
 
 static CheckStatus
-do_node_check_downstream(PGconn *conn, OutputMode mode, CheckStatusList *list_output)
+do_node_check_downstream(PGconn *conn, OutputMode mode, t_node_info *node_info, CheckStatusList *list_output)
 {
 	NodeInfoList downstream_nodes = T_NODE_INFO_LIST_INITIALIZER;
 	NodeInfoListCell *cell = NULL;
@@ -1217,7 +1227,7 @@ do_node_check_downstream(PGconn *conn, OutputMode mode, CheckStatusList *list_ou
 			continue;
 		}
 
-		if (is_downstream_node_attached(conn, cell->node_info->node_name) != NODE_ATTACHED)
+		if (is_downstream_node_attached(conn, cell->node_info->node_name, NULL) != NODE_ATTACHED)
 		{
 			missing_nodes_count++;
 			item_list_append_format(&missing_nodes,
@@ -1234,7 +1244,13 @@ do_node_check_downstream(PGconn *conn, OutputMode mode, CheckStatusList *list_ou
 		}
 	}
 
-	if (missing_nodes_count == 0)
+	if (node_info->type == WITNESS)
+	{
+		/* witness is not connecting to any upstream */
+		appendPQExpBufferStr(&details,
+							 _("N/A - node is a witness"));
+	}
+	else if (missing_nodes_count == 0)
 	{
 		if (expected_nodes_count == 0)
 			appendPQExpBufferStr(&details,
@@ -1367,7 +1383,13 @@ do_node_check_upstream(PGconn *conn, OutputMode mode, t_node_info *node_info, Ch
 
 	initPQExpBuffer(&details);
 
-	if (get_node_record(conn, node_info->upstream_node_id, &upstream_node_info) != RECORD_FOUND)
+	if (node_info->type == WITNESS)
+	{
+		/* witness is not connecting to any upstream */
+		appendPQExpBufferStr(&details,
+							 _("N/A - node is a witness"));
+	}
+	else if (get_node_record(conn, node_info->upstream_node_id, &upstream_node_info) != RECORD_FOUND)
 	{
 		if (get_recovery_type(conn) == RECTYPE_STANDBY)
 		{
@@ -1388,7 +1410,7 @@ do_node_check_upstream(PGconn *conn, OutputMode mode, t_node_info *node_info, Ch
 		upstream_conn = establish_db_connection(upstream_node_info.conninfo, true);
 
 		/* check our node is connected */
-		if (is_downstream_node_attached(upstream_conn, config_file_options.node_name) != NODE_ATTACHED)
+		if (is_downstream_node_attached(upstream_conn, config_file_options.node_name, NULL) != NODE_ATTACHED)
 		{
 			appendPQExpBuffer(&details,
 							  _("node \"%s\" (ID: %i) is not attached to expected upstream node \"%s\" (ID: %i)"),
@@ -2458,7 +2480,7 @@ do_node_rejoin(void)
 		log_hint(_("check the local node is registered with the current primary \"%s\" (ID: %i)"),
 				 primary_node_record.node_name,
 				 primary_node_record.node_id);
-		PQfinish(upstream_conn);
+
 		PQfinish(primary_conn);
 		exit(ERR_BAD_CONFIG);
 	}
@@ -2777,7 +2799,7 @@ do_node_rejoin(void)
 						   config_file_options.node_rejoin_timeout);
 			}
 			else {
-				log_detail(_("no record for local node \"%s\" found in node \"%s\"'s \"pg_stat_replication\" table"),
+				log_detail(_("no active record for local node \"%s\" found in node \"%s\"'s \"pg_stat_replication\" table"),
 						   config_file_options.node_name,
 						   primary_node_record.node_name);
 			}
@@ -2789,7 +2811,7 @@ do_node_rejoin(void)
 	else
 	{
 		/* -W/--no-wait provided - check once */
-		NodeAttached node_attached = is_downstream_node_attached(primary_conn, config_file_options.node_name);
+		NodeAttached node_attached = is_downstream_node_attached(primary_conn, config_file_options.node_name, NULL);
 		if (node_attached == NODE_ATTACHED)
 			success = true;
 	}

repmgr-action-standby.c

@@ -986,14 +986,15 @@ check_barman_config(void)
 /*
  * _do_create_replication_conf()
  *
- * Create recovery.conf for a previously cloned instance.
+ * Create replication configuration for a previously cloned instance.
  *
  * Prerequisites:
  *
- * - data directory must be provided
+ * - data directory must be provided, either explicitly or via
+ *   repmgr.conf
  * - the instance should not be running
  * - an existing "recovery.conf" file can only be overwritten with
- *   -F/--force
+ *   -F/--force (Pg11 and earlier)
  * - connection parameters for an existing, running node must be provided
  * - --upstream-node-id, if provided, will be "primary_conninfo",
  *   otherwise primary node id; node must exist; unless -F/--force
@@ -1168,7 +1169,7 @@ _do_create_replication_conf(void)
 		}
 		else
 		{
-			log_hint(_("standby must be registered before a new recovery.conf file can be created"));
+			log_hint(_("standby must be registered before replication can be configured"));
 		}
 
 		exit(ERR_BAD_CONFIG);
@@ -1796,7 +1797,7 @@ do_standby_register(void)
 			else
 			{
 				/* check our standby is connected */
-				if (is_downstream_node_attached(upstream_conn, config_file_options.node_name) == NODE_ATTACHED)
+				if (is_downstream_node_attached(upstream_conn, config_file_options.node_name, NULL) == NODE_ATTACHED)
 				{
 					log_verbose(LOG_INFO, _("local node is attached to specified upstream node %i"), runtime_options.upstream_node_id);
 				}
@@ -1855,7 +1856,7 @@ do_standby_register(void)
 						primary_node_id);
 
 			/* check our standby is connected */
-			if (is_downstream_node_attached(primary_conn, config_file_options.node_name) == NODE_ATTACHED)
+			if (is_downstream_node_attached(primary_conn, config_file_options.node_name, NULL) == NODE_ATTACHED)
 			{
 				log_verbose(LOG_INFO, _("local node is attached to primary"));
 			}
@@ -2031,6 +2032,9 @@ do_standby_register(void)
 				if (node_record_on_standby.priority != node_record_on_primary.priority)
 					records_match = false;
 
+				if (strcmp(node_record_on_standby.location, node_record_on_primary.location) != 0)
+					records_match = false;
+
 				if (node_record_on_standby.active != node_record_on_primary.active)
 					records_match = false;
 
@@ -2388,7 +2392,7 @@ do_standby_promote(void)
 	 */
 	if (check_free_wal_senders(available_wal_senders, &sibling_nodes_stats, &dry_run_success) == false)
 	{
-		if (runtime_options.dry_run == false)
+		if (runtime_options.dry_run == false || runtime_options.force == false)
 		{
 			PQfinish(local_conn);
 			exit(ERR_BAD_CONFIG);
@@ -2402,7 +2406,7 @@ do_standby_promote(void)
 	 */
 	if (check_free_slots(&local_node_record, &sibling_nodes_stats, &dry_run_success) == false)
 	{
-		if (runtime_options.dry_run == false)
+		if (runtime_options.dry_run == false || runtime_options.force == false)
 		{
 			PQfinish(local_conn);
 			exit(ERR_BAD_CONFIG);
@@ -2503,7 +2507,7 @@ _do_standby_promote_internal(PGconn *conn)
 	/*
 	 * Promote standby to primary.
 	 *
-	 * `pg_ctl promote` returns immediately and (prior to 10.0) has no -w
+	 * "pg_ctl promote: returns immediately and (prior to 10.0) has no -w
 	 * option so we can't be sure when or if the promotion completes. For now
 	 * we'll poll the server until the default timeout (60 seconds)
 	 *
@@ -3069,7 +3073,9 @@ do_standby_follow(void)
 
 	for (timer = 0; timer < config_file_options.standby_follow_timeout; timer++)
 	{
-		NodeAttached node_attached = is_downstream_node_attached(follow_target_conn, config_file_options.node_name);
+		NodeAttached node_attached = is_downstream_node_attached(follow_target_conn,
+																 config_file_options.node_name,
+																 NULL);
 
 		if (node_attached == NODE_ATTACHED)
 		{
@@ -3563,7 +3569,8 @@ do_standby_switchover(void)
 	{
 		if (runtime_options.dry_run == true)
 		{
-			log_info(_("validating database connection for superuser \"%s\""), runtime_options.superuser);
+			log_info(_("validating connection to local database for superuser \"%s\""),
+					 runtime_options.superuser);
 		}
 
 		superuser_conn = establish_db_connection_with_replacement_param(
@@ -3573,23 +3580,27 @@ do_standby_switchover(void)
 
 		if (PQstatus(superuser_conn) != CONNECTION_OK)
 		{
-			log_error(_("unable to connect as provided superuser \"%s\""),
+			log_error(_("unable to connect to local database \"%s\" as provided superuser \"%s\""),
+					  PQdb(superuser_conn),
 					  runtime_options.superuser);
 			exit(ERR_BAD_CONFIG);
 		}
 
 		if (is_superuser_connection(superuser_conn, NULL) == false)
 		{
-			log_error(_("database connection established for provided superuser \"%s\" is not a superuser connection"),
+			log_error(_("connection established to local database \"%s\" for provided superuser \"%s\" is not a superuser connection"),
+					  PQdb(superuser_conn),
 					  runtime_options.superuser);
 			exit(ERR_BAD_CONFIG);
 		}
 
 		if (runtime_options.dry_run == true)
 		{
-			log_info(_("successfully established database connection established for provided superuser \"%s\""),
+			log_info(_("successfully established connection to local database \"%s\" for provided superuser \"%s\""),
+					 PQdb(superuser_conn),
 					 runtime_options.superuser);
 		}
+
 	}
 
 	/*
@@ -3702,7 +3713,7 @@ do_standby_switchover(void)
 		exit(ERR_BAD_CONFIG);
 	}
 
-	if (is_downstream_node_attached(remote_conn, local_node_record.node_name) != NODE_ATTACHED)
+	if (is_downstream_node_attached(remote_conn, local_node_record.node_name, NULL) != NODE_ATTACHED)
 	{
 		log_error(_("local node \"%s\" (ID: %i) is not attached to demotion candidate \"%s\" (ID: %i)"),
 				  local_node_record.node_name,
@@ -4052,11 +4063,13 @@ do_standby_switchover(void)
 
 	if (parse_data_directory_config(command_output.data) == false)
 	{
-		log_error(_("\"data_directory\" parameter in repmgr.conf on \"%s\" is incorrectly configured"),
+		log_error(_("\"data_directory\" parameter in \"repmgr.conf\" on \"%s\" (ID: %i) is incorrectly configured"),
-				  remote_node_record.node_name);
+				  remote_node_record.node_name,
+				  remote_node_record.node_id);
 
-		log_hint(_("execute \"repmgr node check --data-directory-config\" on \"%s\" to diagnose the issue"),
+		log_hint(_("execute \"repmgr node check --data-directory-config\" on \"%s\" (ID: %i) to diagnose the issue"),
-				 remote_node_record.node_name);
+				 remote_node_record.node_name,
+				 remote_node_record.node_id);
 
 		PQfinish(remote_conn);
 		PQfinish(local_conn);
@@ -5184,7 +5197,8 @@ do_standby_switchover(void)
 		 */
 
 		 node_attached = is_downstream_node_attached(local_conn,
-													 remote_node_record.node_name);
+													 remote_node_record.node_name,
+													 NULL);
 		if (node_attached == NODE_ATTACHED)
 		{
 			switchover_success = true;
@@ -5473,6 +5487,7 @@ check_source_server()
 					{
 						uint64		test_system_identifier = system_identifier(cell->node_info->conn);
 						PQfinish(cell->node_info->conn);
+						cell->node_info->conn = NULL;
 
 						if (test_system_identifier != UNKNOWN_SYSTEM_IDENTIFIER)
 						{
@@ -5496,6 +5511,7 @@ check_source_server()
 					else
 					{
 						PQfinish(cell->node_info->conn);
+						cell->node_info->conn = NULL;
 					}
 				}
 				clear_node_info_list(&all_nodes);
@@ -6691,6 +6707,11 @@ run_basebackup(t_node_info *node_record)
 }
 
 
+/*
+ * Perform a filesystem backup using rsync.
+ *
+ * From repmgr 4 this is only used for Barman backups.
+ */
 static int
 run_file_backup(t_node_info *local_node_record)
 {
@@ -6721,10 +6742,11 @@ run_file_backup(t_node_info *local_node_record)
 		/*
 		 * Read the list of backup files into a local file. In the process:
 		 *
-		 * - determine the backup ID; - check, and remove, the prefix; -
+		 * - determine the backup ID
-		 * detect tablespaces; - filter files in one list per tablespace;
+		 * - check, and remove, the prefix
+		 * - detect tablespaces
+		 * - filter files in one list per tablespace
 		 */
-
 		{
 			FILE	   *fi;		/* input stream */
 			FILE	   *fd;		/* output for data.txt */
@@ -7007,11 +7029,13 @@ run_file_backup(t_node_info *local_node_record)
 
 		if (mode == barman)
 		{
-			create_pg_dir(cell_t->location, false);
+			create_pg_dir(tblspc_dir_dest, false);
 
 			if (cell_t->f != NULL)	/* cell_t->f == NULL iff the tablespace is
 									 * empty */
 			{
+				fclose(cell_t->f);
+
 				maxlen_snprintf(command,
 								"rsync --progress -a --files-from=%s/%s.txt %s:%s/%s/%s %s",
 								local_repmgr_tmp_directory,
@@ -7024,7 +7048,6 @@ run_file_backup(t_node_info *local_node_record)
 				(void) local_command(
 									 command,
 									 NULL);
-				fclose(cell_t->f);
 				maxlen_snprintf(filename,
 								"%s/%s.txt",
 								local_repmgr_tmp_directory,
@@ -7141,7 +7164,10 @@ stop_backup:
 
 	if (mode == barman)
 	{
-		/* In Barman mode, remove local_repmgr_directory */
+		/*
+		 * In Barman mode, remove local_repmgr_tmp_directory,
+		 * which contains various temporary files containing Barman metadata.
+		 */
 		rmtree(local_repmgr_tmp_directory, true);
 	}
 
@@ -8579,6 +8605,10 @@ do_standby_help(void)
 	puts("");
 	printf(_("  \"standby promote\" promotes a standby node to primary.\n"));
 	puts("");
+	printf(_("  --dry-run                           perform checks etc. but don't actually promote the node\n"));
+	printf(_("  -F, --force                         ignore warnings and continue anyway\n"));
+	printf(_("  --siblings-follow                   have other standbys follow new primary\n"));
+	puts("");
 
 	printf(_("STANDBY FOLLOW\n"));
 	puts("");

repmgr-client.c

@@ -1196,7 +1196,7 @@ main(int argc, char **argv)
 
 	/*
 	 * If --dry-run specified, ensure log_level is at least LOG_INFO, regardless
-	 * of what's in the configuration file or -L/--log-level paremeter, otherwise
+	 * of what's in the configuration file or -L/--log-level parameter, otherwise
 	 * some or output might not be displayed.
 	 */
 	if (runtime_options.dry_run == true)
@@ -2351,6 +2351,7 @@ format_node_status(t_node_info *node_info, PQExpBufferData *node_status, PQExpBu
 			 * connected to the upstream
 			 */
 			NodeAttached attached_to_upstream = NODE_ATTACHED_UNKNOWN;
+			char *replication_state = NULL;
 			t_node_info upstream_node_rec = T_NODE_INFO_INITIALIZER;
 			RecordStatus upstream_node_rec_found = get_node_record(node_info->conn,
 																   node_info->upstream_node_id,
@@ -2378,7 +2379,7 @@ format_node_status(t_node_info *node_info, PQExpBufferData *node_status, PQExpBu
 				}
 				else
 				{
-					attached_to_upstream = is_downstream_node_attached(upstream_conn, node_info->node_name);
+					attached_to_upstream = is_downstream_node_attached(upstream_conn, node_info->node_name, &replication_state);
 				}
 
 				PQfinish(upstream_conn);
@@ -2394,6 +2395,18 @@ format_node_status(t_node_info *node_info, PQExpBufferData *node_status, PQExpBu
 										upstream_node_rec.node_name,
 										upstream_node_rec.node_id);
 			}
+			if (attached_to_upstream == NODE_NOT_ATTACHED)
+			{
+				appendPQExpBufferStr(upstream, "? ");
+				item_list_append_format(warnings,
+										"node \"%s\" (ID: %i) attached to its upstream node \"%s\" (ID: %i) in state \"%s\"",
+										node_info->node_name,
+										node_info->node_id,
+										upstream_node_rec.node_name,
+										upstream_node_rec.node_id,
+										replication_state);
+			}
+
 			else if (attached_to_upstream == NODE_DETACHED)
 			{
 				appendPQExpBufferStr(upstream, "! ");
@@ -3986,8 +3999,10 @@ check_standby_join(PGconn *upstream_conn, t_node_info *upstream_node_record, t_n
 
 	 for (; i < config_file_options.node_rejoin_timeout; i++)
 	 {
+		 char *node_state = NULL;
 		 NodeAttached node_attached = is_downstream_node_attached(upstream_conn,
-																  standby_node_record->node_name);
+																  standby_node_record->node_name,
+																  &node_state);
 		 if (node_attached == NODE_ATTACHED)
 		 {
 			 log_verbose(LOG_INFO, _("node \"%s\" (ID: %i) has attached to its upstream node"),
@@ -4004,9 +4019,19 @@ check_standby_join(PGconn *upstream_conn, t_node_info *upstream_node_record, t_n
 					  i + 1,
 					  config_file_options.node_rejoin_timeout);
 
-			 log_detail(_("checking for record in node \"%s\"'s \"pg_stat_replication\" table where \"application_name\" is \"%s\""),
+			 if (node_attached == NODE_NOT_ATTACHED)
-						upstream_node_record->node_name,
+			 {
-						standby_node_record->node_name);
+				 log_detail(_("node \"%s\" (ID: %i) is currrently attached to its upstream node in state \"%s\""),
+							upstream_node_record->node_name,
+							standby_node_record->node_id,
+							node_state);
+			 }
+			 else
+			 {
+				 log_detail(_("checking for record in node \"%s\"'s \"pg_stat_replication\" table where \"application_name\" is \"%s\""),
+							upstream_node_record->node_name,
+							standby_node_record->node_name);
+			 }
 		 }
 		 else
 		 {

repmgr.conf.sample

@@ -297,6 +297,7 @@ ssh_options='-q -o ConnectTimeout=10'	# Options to append to "ssh"
 #connection_check_type=ping		# How to check availability of the upstream node; valid options:
 					#  'ping': use PQping() to check if the node is accepting connections
 					#  'connection': execute a throwaway query on the current connection
+					#  'query': execute an SQL statement on the node via the existing connection
 #reconnect_attempts=6			# Number of attempts which will be made to reconnect to an unreachable
 					# primary (or other upstream node)
 #reconnect_interval=10			# Interval between attempts to reconnect to an unreachable

repmgr_version.h.in

@@ -1,5 +1,5 @@
 #define REPMGR_VERSION_DATE ""
-#define REPMGR_VERSION "5.1dev"
+#define REPMGR_VERSION "5.1.0"
 #define REPMGR_VERSION_NUM 50100
-#define REPMGR_RELEASE_DATE "2020-XX-XX"
+#define REPMGR_RELEASE_DATE "2020-04-13"
 #define PG_ACTUAL_VERSION_NUM 

repmgrd-physical.c

@@ -1328,6 +1328,7 @@ monitor_streaming_standby(void)
 	 */
 	if (PQstatus(upstream_conn) != CONNECTION_OK)
 	{
+		close_connection(&upstream_conn);
 		log_error(_("unable connect to upstream node (ID: %i), terminating"),
 				  local_node_info.upstream_node_id);
 		log_hint(_("upstream node must be running before repmgrd can start"));
@@ -1339,8 +1340,8 @@ monitor_streaming_standby(void)
 
 	if (upstream_node_info.node_id == local_node_info.node_id)
 	{
-		PQfinish(upstream_conn);
+		close_connection(&upstream_conn);
-		upstream_conn = NULL;
+
 		return;
 	}
 
@@ -1364,6 +1365,8 @@ monitor_streaming_standby(void)
 
 		if (PQstatus(primary_conn) != CONNECTION_OK)
 		{
+			close_connection(&primary_conn);
+
 			log_error(_("unable to connect to primary node"));
 			log_hint(_("ensure the primary node is reachable from this node"));
 
@@ -1440,6 +1443,7 @@ monitor_streaming_standby(void)
 	while (true)
 	{
 		log_verbose(LOG_DEBUG, "checking %s", upstream_node_info.conninfo);
+
 		if (check_upstream_connection(&upstream_conn, upstream_node_info.conninfo) == true)
 		{
 			set_upstream_last_seen(local_conn, upstream_node_info.node_id);
@@ -1557,8 +1561,9 @@ monitor_streaming_standby(void)
 
 							log_notice(_("current upstream node \"%s\" (ID: %i) is not primary, restarting monitoring"),
 									   upstream_node_info.node_name, upstream_node_info.node_id);
-							PQfinish(upstream_conn);
+
-							upstream_conn = NULL;
+							close_connection(&upstream_conn);
+
 							local_node_info.upstream_node_id = UNKNOWN_NODE_ID;
 
 							/* check local connection */
@@ -1568,7 +1573,7 @@ monitor_streaming_standby(void)
 							{
 								int i;
 
-								PQfinish(local_conn);
+								close_connection(&local_conn);
 
 								for (i = 0; i < config_file_options.repmgrd_standby_startup_timeout; i++)
 								{
@@ -1577,6 +1582,8 @@ monitor_streaming_standby(void)
 									if (PQstatus(local_conn) == CONNECTION_OK)
 										break;
 
+									close_connection(&local_conn);
+
 									log_debug("sleeping 1 second; %i of %i attempts to reconnect to local node",
 											  i + 1,
 											  config_file_options.repmgrd_standby_startup_timeout);
@@ -1706,7 +1713,12 @@ monitor_streaming_standby(void)
 					}
 					else
 					{
-						if (primary_conn == NULL || PQstatus(primary_conn) != CONNECTION_OK)
+						if (primary_conn != NULL && PQstatus(primary_conn) != CONNECTION_OK)
+						{
+							close_connection(&primary_conn);
+						}
+
+						if (primary_conn == NULL)
 						{
 							primary_conn = establish_primary_db_connection(upstream_conn, false);
 						}
@@ -1715,7 +1727,8 @@ monitor_streaming_standby(void)
 					initPQExpBuffer(&event_details);
 
 					appendPQExpBuffer(&event_details,
-									  _("reconnected to upstream node %i after %i seconds, resuming monitoring"),
+									  _("reconnected to upstream node \"%s\" (ID: %i) after %i seconds, resuming monitoring"),
+									  upstream_node_info.node_name,
 									  upstream_node_info.node_id,
 									  degraded_monitoring_elapsed);
 
@@ -1851,7 +1864,9 @@ monitor_streaming_standby(void)
 
 							if (PQstatus(cell->node_info->conn) != CONNECTION_OK)
 							{
+								close_connection(&cell->node_info->conn);
 								log_debug("unable to connect to %i ... ", cell->node_info->node_id);
+								close_connection(&cell->node_info->conn);
 								continue;
 							}
 
@@ -2036,7 +2051,8 @@ loop:
 
 				if (last_known_upstream_node_id != local_node_info.upstream_node_id)
 				{
-					log_notice(_("local node %i upstream appears to have changed, restarting monitoring"),
+					log_notice(_("upstream for local node \"%s\" (ID: %i) appears to have changed, restarting monitoring"),
+							   local_node_info.node_name,
 							   local_node_info.node_id);
 					log_detail(_("currently monitoring upstream %i; new upstream is %i"),
 							   last_known_upstream_node_id,
@@ -2050,7 +2066,8 @@ loop:
 				 */
 				if (local_node_info.type != STANDBY)
 				{
-					log_notice(_("local node %i is no longer a standby, restarting monitoring"),
+					log_notice(_("local node \"%s\" (ID: %i) is no longer a standby, restarting monitoring"),
+							   local_node_info.node_name,
 							   local_node_info.node_id);
 					close_connection(&upstream_conn);
 					return;
@@ -2075,8 +2092,8 @@ loop:
 				{
 					log_notice(_("current upstream node \"%s\" (ID: %i) is not primary, restarting monitoring"),
 							   upstream_node_info.node_name, upstream_node_info.node_id);
-					PQfinish(primary_conn);
+
-					primary_conn = NULL;
+					close_connection(&primary_conn);
 
 					local_node_info.upstream_node_id = UNKNOWN_NODE_ID;
 					return;
@@ -2144,7 +2161,8 @@ loop:
 
 			if (last_known_upstream_node_id != local_node_info.upstream_node_id)
 			{
-				log_notice(_("local node %i's upstream appears to have changed, restarting monitoring"),
+				log_notice(_("local node \"%s\" (ID: %i)'s upstream appears to have changed, restarting monitoring"),
+						   local_node_info.node_name,
 						   local_node_info.node_id);
 				log_detail(_("currently monitoring upstream %i; new upstream is %i"),
 						   last_known_upstream_node_id,
@@ -2335,8 +2353,9 @@ monitor_streaming_witness(void)
 					{
 						log_notice(_("current upstream node \"%s\" (ID: %i) is not primary, restarting monitoring"),
 								   upstream_node_info.node_name, upstream_node_info.node_id);
-						PQfinish(primary_conn);
+
-						primary_conn = NULL;
+						close_connection(&primary_conn);
+
 						termPQExpBuffer(&event_details);
 						return;
 					}
@@ -2397,7 +2416,8 @@ monitor_streaming_witness(void)
 					initPQExpBuffer(&event_details);
 
 					appendPQExpBuffer(&event_details,
-									  _("reconnected to upstream node %i after %i seconds, resuming monitoring"),
+									  _("reconnected to upstream node \"%s\" (ID: %i) after %i seconds, resuming monitoring"),
+									  upstream_node_info.node_name,
 									  upstream_node_info.node_id,
 									  degraded_monitoring_elapsed);
 
@@ -2407,9 +2427,11 @@ monitor_streaming_witness(void)
 					if (get_recovery_type(primary_conn) != RECTYPE_PRIMARY)
 					{
 						log_notice(_("current upstream node \"%s\" (ID: %i) is not primary, restarting monitoring"),
-								   upstream_node_info.node_name, upstream_node_info.node_id);
+								   upstream_node_info.node_name,
-						PQfinish(primary_conn);
+								   upstream_node_info.node_id);
-						primary_conn = NULL;
+
+						close_connection(&primary_conn);
+
 						termPQExpBuffer(&event_details);
 						return;
 					}
@@ -2463,7 +2485,9 @@ monitor_streaming_witness(void)
 
 						if (PQstatus(cell->node_info->conn) != CONNECTION_OK)
 						{
+							close_connection(&cell->node_info->conn);
 							log_debug("unable to connect to %i ... ", cell->node_info->node_id);
+							close_connection(&cell->node_info->conn);
 							continue;
 						}
 
@@ -2605,8 +2629,9 @@ loop:
 				{
 					log_notice(_("current upstream node \"%s\" (ID: %i) is not primary, restarting monitoring"),
 							   upstream_node_info.node_name, upstream_node_info.node_id);
-					PQfinish(primary_conn);
+
-					primary_conn = NULL;
+					close_connection(&primary_conn);
+
 					return;
 				}
 
@@ -2716,24 +2741,32 @@ do_primary_failover(void)
 			{
 				for (cell = check_sibling_nodes.head; cell; cell = cell->next)
 				{
-					pid_t sibling_wal_receiver_pid;
-
 					if (cell->node_info->conn == NULL)
 						cell->node_info->conn = establish_db_connection(cell->node_info->conninfo, false);
 
-					sibling_wal_receiver_pid = (pid_t)get_wal_receiver_pid(cell->node_info->conn);
+					if (PQstatus(cell->node_info->conn) != CONNECTION_OK)
-
-					if (sibling_wal_receiver_pid == UNKNOWN_PID)
 					{
-						log_warning(_("unable to query WAL receiver PID on node %i"),
+						log_warning(_("unable to query WAL receiver PID on node \"%s\" (ID: %i)"),
+									cell->node_info->node_name,
 									cell->node_info->node_id);
+						close_connection(&cell->node_info->conn);
 					}
-					else if (sibling_wal_receiver_pid > 0)
+					else
 					{
-						log_info(_("WAL receiver PID on node %i is %i"),
+						pid_t sibling_wal_receiver_pid = (pid_t)get_wal_receiver_pid(cell->node_info->conn);
-								 cell->node_info->node_id,
+
-								 sibling_wal_receiver_pid);
+						if (sibling_wal_receiver_pid == UNKNOWN_PID)
-						sibling_node_wal_receiver_connected = true;
+						{
+							log_warning(_("unable to query WAL receiver PID on node %i"),
+										cell->node_info->node_id);
+						}
+						else if (sibling_wal_receiver_pid > 0)
+						{
+							log_info(_("WAL receiver PID on node %i is %i"),
+									 cell->node_info->node_id,
+									 sibling_wal_receiver_pid);
+							sibling_node_wal_receiver_connected = true;
+						}
 					}
 				}
 
@@ -2895,7 +2928,8 @@ do_primary_failover(void)
 
 					initPQExpBuffer(&event_details);
 					appendPQExpBuffer(&event_details,
-									  _("node %i is in manual failover mode and is now disconnected from streaming replication"),
+									  _("node \"%s\" (ID: %i) is in manual failover mode and is now disconnected from streaming replication"),
+									  local_node_info.node_name,
 									  local_node_info.node_id);
 
 					new_primary_conn = establish_db_connection(new_primary.conninfo, false);
@@ -3055,8 +3089,6 @@ do_primary_failover(void)
 }
 
 
-
-
 static void
 update_monitoring_history(void)
 {
@@ -3291,6 +3323,8 @@ do_upstream_standby_failover(void)
 		if (PQstatus(local_conn) == CONNECTION_OK)
 			break;
 
+		close_connection(&local_conn);
+
 		log_debug("sleeping 1 second; %i of %i (\"repmgrd_standby_startup_timeout\") attempts to reconnect to local node",
 				  i + 1,
 				  config_file_options.repmgrd_standby_startup_timeout);
@@ -3299,7 +3333,8 @@ do_upstream_standby_failover(void)
 
 	if (PQstatus(local_conn) != CONNECTION_OK)
 	{
-		log_error(_("unable to reconnect to local node %i"),
+		log_error(_("unable to reconnect to local node \"%s\" (ID: %i)"),
+				  local_node_info.node_name,
 				  local_node_info.node_id);
 		return FAILOVER_STATE_FOLLOW_FAIL;
 	}
@@ -3334,7 +3369,8 @@ do_upstream_standby_failover(void)
 
 			initPQExpBuffer(&event_details);
 			appendPQExpBuffer(&event_details,
-							  _("unable to set node %i's new upstream ID to %i"),
+							  _("unable to set node \"%s\" (ID: %i)'s new upstream ID to %i"),
+							  local_node_info.node_name,
 							  local_node_info.node_id,
 							  primary_node_info.node_id);
 
@@ -3372,8 +3408,10 @@ do_upstream_standby_failover(void)
 		initPQExpBuffer(&event_details);
 
 		appendPQExpBuffer(&event_details,
-						  _("node %i is now following primary node %i"),
+						  _("node \"%s\" (ID: %i) is now following primary node \"%s\" (ID: %i)"),
+						  local_node_info.node_name,
 						  local_node_info.node_id,
+						  primary_node_info.node_name,
 						  primary_node_info.node_id);
 
 		log_notice("%s", event_details.data);
@@ -3438,12 +3476,16 @@ promote_self(void)
 
 	r = system(promote_command);
 
+	log_debug("result of promote_command: %i", WEXITSTATUS(r));
+
 	/* connection should stay up, but check just in case */
 	if (PQstatus(local_conn) != CONNECTION_OK)
 	{
 		log_warning(_("local database connection not available"));
 		log_detail("\n%s", PQerrorMessage(local_conn));
 
+		close_connection(&local_conn);
+
 		local_conn = establish_db_connection(local_node_info.conninfo, true);
 
 		/* assume node failed */
@@ -3451,24 +3493,37 @@ promote_self(void)
 		{
 			log_error(_("unable to reconnect to local node"));
 			log_detail("\n%s", PQerrorMessage(local_conn));
+
+			close_connection(&local_conn);
+
 			/* XXX handle this */
 			return FAILOVER_STATE_LOCAL_NODE_FAILURE;
 		}
 	}
 
-	if (r != 0)
+	if (WIFEXITED(r) && WEXITSTATUS(r))
 	{
-		int			primary_node_id;
+		int			primary_node_id = UNKNOWN_NODE_ID;
+
+		log_error(_("promote command failed"));
+		log_detail(_("promote command exited with error code %i"), WEXITSTATUS(r));
+
+		log_info(_("checking if original primary node has reappeared"));
 
 		upstream_conn = get_primary_connection(local_conn,
 											   &primary_node_id,
 											   NULL);
 
-		if (PQstatus(upstream_conn) == CONNECTION_OK && primary_node_id == failed_primary.node_id)
+		if (PQstatus(upstream_conn) != CONNECTION_OK)
+		{
+			close_connection(&upstream_conn);
+		}
+		else if (primary_node_id == failed_primary.node_id)
 		{
 			PQExpBufferData event_details;
 
-			log_notice(_("original primary (ID: %i) reappeared before this standby was promoted - no action taken"),
+			log_notice(_("original primary \"%s\" (ID: %i) reappeared before this standby was promoted - no action taken"),
+					   failed_primary.node_name,
 					   failed_primary.node_id);
 
 			initPQExpBuffer(&event_details);
@@ -3493,9 +3548,6 @@ promote_self(void)
 			return FAILOVER_STATE_PRIMARY_REAPPEARED;
 		}
 
-
-		log_error(_("promote command failed"));
-
 		create_event_notification(NULL,
 								  &config_file_options,
 								  local_node_info.node_id,
@@ -3506,6 +3558,11 @@ promote_self(void)
 		return FAILOVER_STATE_PROMOTION_FAILED;
 	}
 
+	/*
+	 * Promotion has succeeded - verify local connection is still available
+	 */
+	try_reconnect(&local_conn, &local_node_info);
+
 	/* bump the electoral term */
 	increment_current_term(local_conn);
 
@@ -3521,8 +3578,10 @@ promote_self(void)
 		initPQExpBuffer(&event_details);
 
 		appendPQExpBuffer(&event_details,
-						  _("node %i promoted to primary; old primary %i marked as failed"),
+						  _("node \"%s\" (ID: %i) promoted to primary; old primary \"%s\" (ID: %i) marked as failed"),
+						  local_node_info.node_name,
 						  local_node_info.node_id,
+						  failed_primary.node_name,
 						  failed_primary.node_id);
 
 		/* local_conn is now the primary connection */
@@ -3566,6 +3625,8 @@ notify_followers(NodeInfoList *standby_nodes, int follow_node_id)
 					 cell->node_info->node_name,
 					 cell->node_info->node_id);
 
+			close_connection(&cell->node_info->conn);
+
 			cell->node_info->conn = establish_db_connection(cell->node_info->conninfo, false);
 		}
 
@@ -3576,6 +3637,7 @@ notify_followers(NodeInfoList *standby_nodes, int follow_node_id)
 						cell->node_info->node_id);
 			log_detail("\n%s", PQerrorMessage(cell->node_info->conn));
 
+			close_connection(&cell->node_info->conn);
 			continue;
 		}
 
@@ -3800,15 +3862,18 @@ follow_new_primary(int new_primary_id)
 		if (PQstatus(local_conn) == CONNECTION_OK)
 			break;
 
+		close_connection(&local_conn);
+
 		log_debug("sleeping 1 second; %i of %i attempts to reconnect to local node",
 				  i + 1,
 				  config_file_options.repmgrd_standby_startup_timeout);
 		sleep(1);
 	}
 
-	if (PQstatus(local_conn) != CONNECTION_OK)
+	if (local_conn == NULL || PQstatus(local_conn) != CONNECTION_OK)
 	{
-		log_error(_("unable to reconnect to local node %i"),
+		log_error(_("unable to reconnect to local node \"%s\" (ID: %i)"),
+				  local_node_info.node_name,
 				  local_node_info.node_id);
 		return FAILOVER_STATE_FOLLOW_FAIL;
 	}
@@ -3822,8 +3887,10 @@ follow_new_primary(int new_primary_id)
 
 		initPQExpBuffer(&event_details);
 		appendPQExpBuffer(&event_details,
-						  _("node %i now following new upstream node %i"),
+						  _("node \"%s\" (ID: %i) now following new upstream node \"%s\" (ID: %i)"),
+						  local_node_info.node_name,
 						  local_node_info.node_id,
+						  upstream_node_info.node_name,
 						  upstream_node_info.node_id);
 
 		log_notice("%s", event_details.data);
@@ -3918,7 +3985,7 @@ witness_follow_new_primary(int new_primary_id)
 	record_status = get_node_record(upstream_conn, local_node_info.node_id, &local_node_info);
 	if (record_status != RECORD_FOUND)
 	{
-		log_error(_("unable to retrieve metadata record found for node %i"),
+		log_error(_("unable to retrieve metadata record for node %i"),
 				  local_node_info.node_id);
 		return FAILOVER_STATE_FOLLOW_FAIL;
 	}
@@ -3928,8 +3995,10 @@ witness_follow_new_primary(int new_primary_id)
 
 		initPQExpBuffer(&event_details);
 		appendPQExpBuffer(&event_details,
-						  _("witness node %i now following new primary node %i"),
+						  _("witness node \"%s\" (ID: %i) now following new primary node \"%s\" (ID: %i)"),
+						  local_node_info.node_name,
 						  local_node_info.node_id,
+						  upstream_node_info.node_name,
 						  upstream_node_info.node_id);
 
 		log_notice("%s", event_details.data);
@@ -4031,6 +4100,12 @@ do_election(NodeInfoList *sibling_nodes, int *new_primary_id)
 
 		return ELECTION_NOT_CANDIDATE;
 	}
+	if (config_file_options.failover_delay > 0)
+	{
+		log_debug("sleeping %i seconds (\"failover_delay\") before initiating failover",
+				  config_file_options.failover_delay);
+		sleep(config_file_options.failover_delay);
+	}
 
 	/* node priority is set to zero - don't become a candidate, and lose by default */
 	if (local_node_info.priority <= 0)
@@ -4165,6 +4240,8 @@ do_election(NodeInfoList *sibling_nodes, int *new_primary_id)
 
 		if (PQstatus(cell->node_info->conn) != CONNECTION_OK)
 		{
+			close_connection(&cell->node_info->conn);
+
 			continue;
 		}
 
@@ -4295,7 +4372,8 @@ do_election(NodeInfoList *sibling_nodes, int *new_primary_id)
 		{
 			if (sibling_replication_info.upstream_node_id != upstream_node_info.node_id)
 			{
-				log_warning(_("assumed sibling node %i monitoring different upstream node %i"),
+				log_warning(_("assumed sibling node \"%s\" (ID: %i) monitoring different upstream node %i"),
+							cell->node_info->node_name,
 							cell->node_info->node_id,
 							sibling_replication_info.upstream_node_id);
 
@@ -4335,7 +4413,8 @@ do_election(NodeInfoList *sibling_nodes, int *new_primary_id)
 		/* don't check 0-priority nodes */
 		if (cell->node_info->priority <= 0)
 		{
-			log_info(_("node %i has priority of %i, skipping"),
+			log_info(_("node \"%s\" (ID: %i) has priority of %i, skipping"),
+					   cell->node_info->node_name,
 					   cell->node_info->node_id,
 					   cell->node_info->priority);
 			continue;
@@ -4596,8 +4675,8 @@ check_connection(t_node_info *node_info, PGconn **conn)
 					node_info->node_name,
 					node_info->node_id);
 		log_detail("\n%s", PQerrorMessage(*conn));
-		PQfinish(*conn);
+
-		*conn = NULL;
+		close_connection(conn);
 	}
 
 	if (PQstatus(*conn) != CONNECTION_OK)
@@ -4606,13 +4685,14 @@ check_connection(t_node_info *node_info, PGconn **conn)
 				 node_info->node_name,
 				 node_info->node_id);
 
-		PQfinish(*conn);
+		close_connection(conn);
+
 		*conn = establish_db_connection(node_info->conninfo, false);
 
 		if (PQstatus(*conn) != CONNECTION_OK)
 		{
-			PQfinish(*conn);
+			close_connection(conn);
-			*conn = NULL;
+
 			log_warning(_("reconnection to node \"%s\" (ID: %i) failed"),
 						node_info->node_name,
 						node_info->node_id);
@@ -4686,7 +4766,8 @@ handle_sighup(PGconn **conn, t_server_type server_type)
 
 	if (reload_config(&config_file_options, server_type))
 	{
-		PQfinish(*conn);
+		close_connection(conn);
+
 		*conn = establish_db_connection(config_file_options.conninfo, true);
 	}
 
@@ -4847,6 +4928,8 @@ check_node_can_follow(PGconn *local_conn, XLogRecPtr local_xlogpos, PGconn *foll
 	if (PQstatus(follow_target_repl_conn) != CONNECTION_OK)
 	{
 		log_error(_("unable to establish a replication connection to the follow target node"));
+
+		PQfinish(follow_target_repl_conn);
 		return false;
 	}
 
@@ -4978,7 +5061,6 @@ check_node_can_follow(PGconn *local_conn, XLogRecPtr local_xlogpos, PGconn *foll
 	if (follow_target_history)
 		pfree(follow_target_history);
 
-
 	return can_follow;
 }
 
@@ -5152,11 +5234,8 @@ parse_child_nodes_disconnect_command(char *parsed_command, char *template, int r
 int
 try_primary_reconnect(PGconn **conn, PGconn *local_conn, t_node_info *node_info)
 {
-	PGconn	   *our_conn;
 	t_conninfo_param_list conninfo_params = T_CONNINFO_PARAM_LIST_INITIALIZER;
-
 	int			i;
-
 	int			max_attempts = config_file_options.reconnect_attempts;
 
 	initialize_conninfo_params(&conninfo_params, false);
@@ -5170,11 +5249,18 @@ try_primary_reconnect(PGconn **conn, PGconn *local_conn, t_node_info *node_info)
 
 	for (i = 0; i < max_attempts; i++)
 	{
-		log_info(_("checking state of node %i, %i of %i attempts"),
+		log_info(_("checking state of node \"%s\" (ID: %i), %i of %i attempts"),
-				 node_info->node_id, i + 1, max_attempts);
+				 node_info->node_name,
+				 node_info->node_id,
+				 i + 1, max_attempts);
+
 		if (is_server_available_params(&conninfo_params) == true)
 		{
-			log_notice(_("node %i has recovered, reconnecting"), node_info->node_id);
+			PGconn	   *our_conn;
+
+			log_notice(_("node \"%s\" (ID: %i) has recovered, reconnecting"),
+					   node_info->node_name,
+					   node_info->node_id);
 
 			/*
 			 * Note: we could also handle the case where node is pingable but
@@ -5187,7 +5273,9 @@ try_primary_reconnect(PGconn **conn, PGconn *local_conn, t_node_info *node_info)
 			{
 				free_conninfo_params(&conninfo_params);
 
-				log_info(_("connection to node %i succeeded"), node_info->node_id);
+				log_info(_("connection to node \"%s\" (ID: %i) succeeded"),
+						 node_info->node_name,
+						 node_info->node_id);
 
 				if (PQstatus(*conn) == CONNECTION_BAD)
 				{
@@ -5244,6 +5332,8 @@ try_primary_reconnect(PGconn **conn, PGconn *local_conn, t_node_info *node_info)
 					{
 						log_notice(_("received notification that new primary is node %i"), new_primary_node_id);
 					}
+
+					free_conninfo_params(&conninfo_params);
 					return new_primary_node_id;
 				}
 				sleep(1);
@@ -5251,7 +5341,8 @@ try_primary_reconnect(PGconn **conn, PGconn *local_conn, t_node_info *node_info)
 		}
 	}
 
-	log_warning(_("unable to reconnect to node %i after %i attempts"),
+	log_warning(_("unable to reconnect to node \"%s\" (ID: %i) after %i attempts"),
+				node_info->node_name,
 				node_info->node_id,
 				max_attempts);
 

repmgrd.c

@@ -817,32 +817,65 @@ check_upstream_connection(PGconn **conn, const char *conninfo)
 	/* Check the connection status twice in case it changes after reset */
 	bool		twice = false;
 
-	if (config_file_options.connection_check_type == CHECK_PING)
-		return is_server_available(conninfo);
 
-	if (config_file_options.connection_check_type == CHECK_CONNECTION)
+	log_debug("connection check type is \"%s\"",
+			  print_connection_check_type(config_file_options.connection_check_type));
+	/*
+	 * For the check types which do not involve using the existing database
+	 * connection, we'll perform the actual check, then as an additional
+	 * safeguard verify that the connection is still valid (as it might have
+	 * gone away during a brief outage between checks).
+	 */
+	if (config_file_options.connection_check_type != CHECK_QUERY)
 	{
 		bool success = true;
-		PGconn *test_conn = PQconnectdb(conninfo);
 
-		log_debug("check_upstream_connection(): attempting to connect to \"%s\"", conninfo);
+		if (config_file_options.connection_check_type == CHECK_PING)
-
-		if (PQstatus(test_conn) != CONNECTION_OK)
 		{
-			log_warning(_("unable to connect to \"%s\""), conninfo);
+			success = is_server_available(conninfo);
-			log_detail("\n%s", PQerrorMessage(test_conn));
-			success = false;
 		}
-		PQfinish(test_conn);
+		else if (config_file_options.connection_check_type == CHECK_CONNECTION)
+		{
+			PGconn *test_conn = PQconnectdb(conninfo);
 
-		return success;
+			log_debug("check_upstream_connection(): attempting to connect to \"%s\"", conninfo);
+
+			if (PQstatus(test_conn) != CONNECTION_OK)
+			{
+				log_warning(_("unable to connect to \"%s\""), conninfo);
+				log_detail("\n%s", PQerrorMessage(test_conn));
+				success = false;
+			}
+			PQfinish(test_conn);
+		}
+
+		if (success == false)
+			return false;
+
+		if (PQstatus(*conn) == CONNECTION_OK)
+			return true;
+
+		/*
+		 * Checks have succeeded, but the open connection to the primary has gone away,
+		 * possibly due to a brief outage between monitoring intervals - attempt to
+		 * reset it.
+		 */
+		log_notice(_("upstream is available but upstream connection has gone away, resetting"));
+
+		PQfinish(*conn);
+		*conn = PQconnectdb(conninfo);
+
+		if (PQstatus(*conn) == CONNECTION_OK)
+			return true;
+
+		return false;
 	}
 
 	for (;;)
 	{
 		if (PQstatus(*conn) != CONNECTION_OK)
 		{
-			log_debug("check_upstream_connection(): connection not OK");
+			log_debug("check_upstream_connection(): upstream connection has gone away, resetting");
 			if (twice)
 				return false;
 			/* reconnect */
@@ -877,6 +910,7 @@ check_upstream_connection(PGconn **conn, const char *conninfo)
 				return false;
 
 			/* reconnect */
+			log_debug("check_upstream_connection(): upstream connection not available, resetting");
 			PQfinish(*conn);
 			*conn = PQconnectdb(conninfo);
 			twice = true;