Fix alignment and syntax

This commit introduces three new options:
  - start_command
  - stop_command
  - restart_command

If these are set, repmgr will issue the specified command instead
of the default pg_ctl commands

FAQ.md

@@ -38,7 +38,7 @@ General
 
   No. Hash indexes and replication do not mix well and their use is
   explicitly discouraged; see:
-    http://www.postgresql.org/docs/current/interactive/sql-createindex.html#AEN74175
+    https://www.postgresql.org/docs/current/interactive/sql-createindex.html#AEN74175
 
 `repmgr`
 --------

HISTORY

@@ -1,3 +1,27 @@
+3.1.5   2016-08-15
+        repmgrd: in a failover situation, prevent endless looping when
+          attempting to establish the status of a node with
+          `failover=manual` (Ian)
+        repmgrd: improve handling of failover events on standbys with
+          `failover=manual`, and create a new event notification
+          for this, `standby_disconnect_manual` (Ian)
+        repmgr: add further event notifications (Gianni)
+        repmgr: when executing `standby switchover`, don't collect remote
+          command output unless required (Gianni, Ian)
+        repmgrd: improve standby monitoring query (Ian, based on suggestion
+          from  Álvaro)
+        repmgr: various command line handling improvements (Ian)
+
+3.1.4   2016-07-12
+        repmgr: new configuration option for setting "restore_command"
+          in the recovery.conf file generated by repmgr (Martín)
+        repmgr: add --csv option to "repmgr cluster show" (Gianni)
+        repmgr: enable provision of a conninfo string as the -d/--dbname
+          parameter, similar to other PostgreSQL utilities (Ian)
+        repmgr: during switchover operations improve detection of
+          demotion candidate shutdown (Ian)
+        various bugfixes and documentation updates (Ian, Martín)
+
 3.1.3   2016-05-17
         repmgrd: enable monitoring when a standby is catching up by
           replaying archived WAL (Ian)

Makefile

@@ -5,7 +5,7 @@
 HEADERS = $(wildcard *.h)
 
 repmgrd_OBJS = dbutils.o config.o repmgrd.o log.o strutil.o
-repmgr_OBJS = dbutils.o check_dir.o config.o repmgr.o log.o strutil.o
+repmgr_OBJS = dbutils.o check_dir.o config.o repmgr.o log.o strutil.o dirmod.o
 
 DATA = repmgr.sql uninstall_repmgr.sql
 

README.md

@@ -48,7 +48,7 @@ This guide assumes that you are familiar with PostgreSQL administration and
 streaming replication concepts. For further details on streaming
 replication, see this link:
 
-  http://www.postgresql.org/docs/current/interactive/warm-standby.html#STREAMING-REPLICATION
+  https://www.postgresql.org/docs/current/interactive/warm-standby.html#STREAMING-REPLICATION
 
 The following terms are used throughout the `repmgr` documentation.
 
@@ -155,9 +155,15 @@ system.
 
 - RedHat/CentOS: RPM packages for `repmgr` are available via Yum through
   the PostgreSQL Global Development Group RPM repository ( http://yum.postgresql.org/ ).
-  You need to follow the instructions for your distribution (RedHat, CentOS,
+  Follow the instructions for your distribution (RedHat, CentOS,
   Fedora, etc.) and architecture as detailed at yum.postgresql.org.
 
+  2ndQuadrant also provides its own RPM packages which are made available
+  at the same time as each `repmgr` release, as it can take some days for
+  them to become available via the main PGDG repository. See here for details:
+
+     http://repmgr.org/yum-repository.html
+
 - Debian/Ubuntu: the most recent `repmgr` packages are available from the
   PostgreSQL Community APT repository ( http://apt.postgresql.org/ ).
   Instructions can be found in the APT section of the PostgreSQL Wiki
@@ -215,6 +221,34 @@ command line options:
 - `-b/--pg_bindir`
 
 
+### Command line options and environment variables
+
+For some commands, e.g. `repmgr standby clone`, database connection parameters
+need to be provided. Like other PostgreSQL utilities, following standard
+parameters can be used:
+
+- `-d/--dbname=DBNAME`
+- `-h/--host=HOSTNAME`
+- `-p/--port=PORT`
+- `-U/--username=USERNAME`
+
+If `-d/--dbname` contains an `=` sign or starts with a valid URI prefix (`postgresql://`
+or `postgres://`), it is treated as a conninfo string. See the PostgreSQL
+documentation for further details:
+
+  https://www.postgresql.org/docs/current/static/libpq-connect.html#LIBPQ-CONNSTRING
+
+Note that if a `conninfo` string is provided, values set in this will override any
+provided as individual parameters. For example, with `-d 'host=foo' --host bar`, `foo`
+will be chosen over `bar`.
+
+Like other PostgreSQL utilities, `repmgr` will default to any values set in environment
+variables if explicit command line parameters are not provided. See the PostgreSQL
+documentation for further details:
+
+  https://www.postgresql.org/docs/current/static/libpq-envars.html
+
+
 Setting up a simple replication cluster with repmgr
 ---------------------------------------------------
 
@@ -237,15 +271,19 @@ both servers.
 On the master server, a PostgreSQL instance must be initialised and running.
 The following replication settings must be included in `postgresql.conf`:
 
+
+    # Enable replication connections; set this figure to at least one more
+    # than the number of standbys which will connect to this server
+    # (note that repmgr will execute `pg_basebackup` in WAL streaming mode,
+    # which requires two free WAL senders)
+
+    max_wal_senders = 10
+
     # Ensure WAL files contain enough information to enable read-only queries
     # on the standby
 
     wal_level = 'hot_standby'
 
-    # Enable up to 10 replication connections
-
-    max_wal_senders = 10
-
     # How much WAL to retain on the master to allow a temporarily
     # disconnected standby to catch up again. The larger this is, the
     # longer the standby can be disconnected. This is needed only in
@@ -259,6 +297,14 @@ The following replication settings must be included in `postgresql.conf`:
 
     hot_standby = on
 
+    # Enable WAL file archiving
+    archive_mode = on
+
+    # Set archive command to a script or application that will safely store
+    # you WALs in a secure place. /bin/true is an example of a command that
+    # ignores archiving. Use something more sensible.
+    archive_command = '/bin/true'
+
 
 * * *
 
@@ -371,14 +417,32 @@ Clone the standby with:
     [2016-01-07 17:21:28] [NOTICE] you can now start your PostgreSQL server
     [2016-01-07 17:21:28] [HINT] for example : pg_ctl -D /path/to/node2/data/ start
 
-This will clone the PostgreSQL data directory files from the master at repmgr_node1
-using PostgreSQL's pg_basebackup utility. A `recovery.conf` file containing the
+This will clone the PostgreSQL data directory files from the master at `repmgr_node1`
+using PostgreSQL's `pg_basebackup` utility. A `recovery.conf` file containing the
 correct parameters to start streaming from this master server will be created
-automatically, and unless otherwise the `postgresql.conf` and `pg_hba.conf`
+automatically, and unless otherwise specified, the `postgresql.conf` and `pg_hba.conf`
 files will be copied from the master.
 
-Make any adjustments to the PostgreSQL configuration files now, then start the
-standby server.
+Be aware that when initially cloning a standby, you will need to ensure
+that all required WAL files remain available while the cloning is taking
+place. To ensure this happens when using the default `pg_basebackup` method,
+`repmgr` will set `pg_basebackup`'s `--xlog-method` parameter to `stream`,
+which will ensure all WAL files generated during the cloning process are
+streamed in parallel with the main backup. Note that this requires two
+replication connections to be available.
+
+To override this behaviour, in `repmgr.conf` set `pg_basebackup`'s
+`--xlog-method` parameter to `fetch`:
+
+    pg_basebackup_options='--xlog-method=fetch'
+
+and ensure that `wal_keep_segments` is set to an appropriately high value.
+See the `pg_basebackup` documentation for details:
+
+    https://www.postgresql.org/docs/current/static/app-pgbasebackup.html
+
+Make any adjustments to the standby's PostgreSQL configuration files now,
+then start the server.
 
 * * *
 
@@ -458,7 +522,11 @@ so should be used with care.
 Further options can be passed to the `pg_basebackup` utility via
 the setting `pg_basebackup_options` in `repmgr.conf`. See the PostgreSQL
 documentation for more details of available options:
+<<<<<<< HEAD
   http://www.postgresql.org/docs/current/static/app-pgbasebackup.html
+=======
+  https://www.postgresql.org/docs/current/static/app-pgbasebackup.html
+>>>>>>> 72f9b0145afab1060dd1202c8f8937653c8b2e39
 
 ### Using rsync to clone a standby
 
@@ -476,7 +544,6 @@ and destination server as the contents of files existing on both servers need
 to be compared, meaning this method is not necessarily faster than making a
 fresh clone with `pg_basebackup`.
 
-
 ### Dealing with PostgreSQL configuration files
 
 By default, `repmgr` will attempt to copy the standard configuration files
@@ -491,6 +558,21 @@ which enables any valid `rsync` options to be passed to that command, e.g.:
 
     rsync_options='--exclude=postgresql.local.conf'
 
+### Controlling `primary_conninfo` in `recovery.conf`
+
+`repmgr` will create the `primary_conninfo` setting in `recovery.conf` based
+on the connection parameters provided to `repmgr standby clone` and PostgreSQL's
+standard connection defaults, including any environment variables set on the
+local node.
+
+To include specific connection parameters other than the standard host, port,
+username and database values (e.g. `sslmode`), include these in a `conninfo`-style
+tring passed to `repmgr` with `-d/--dbname` (see above for details), and/or set
+appropriate environment variables.
+
+Note that PostgreSQL will always set explicit defaults for `sslmode` and
+`sslcompression`.
+
 
 Setting up cascading replication with repmgr
 --------------------------------------------
@@ -564,6 +646,10 @@ To enable `repmgr` to use replication slots, set the boolean parameter
 Note that `repmgr` will fail with an error if this option is specified when
 working with PostgreSQL 9.3.
 
+Replication slots must be enabled in `postgresql.conf` by setting the parameter
+`max_replication_slots` to at least the number of expected standbys (changes
+to this parameter require a server restart).
+
 When cloning a standby, `repmgr` will automatically generate an appropriate
 slot name, which is stored in the `repl_nodes` table, and create the slot
 on the master:
@@ -586,21 +672,9 @@ Note that a slot name will be created by default for the master but not
 actually used unless the master is converted to a standby using e.g.
 `repmgr standby switchover`.
 
-Be aware that when initially cloning a standby, you will need to ensure
-that all required WAL files remain available while the cloning is taking
-place. If using the default `pg_basebackup` method, we recommend setting
-`pg_basebackup`'s `--xlog-method` parameter to `stream` like this:
-
-    pg_basebackup_options='--xlog-method=stream'
-
-See the `pg_basebackup` documentation for details:
-    http://www.postgresql.org/docs/current/static/app-pgbasebackup.html
-
-Otherwise it's necessary to set `wal_keep_segments` to an appropriately high
-value.
 
 Further information on replication slots in the PostgreSQL documentation:
-    http://www.postgresql.org/docs/current/interactive/warm-standby.html#STREAMING-REPLICATION-SLOTS
+    https://www.postgresql.org/docs/current/interactive/warm-standby.html#STREAMING-REPLICATION-SLOTS
 
 
 Promoting a standby server with repmgr
@@ -699,8 +773,9 @@ updated to reflect this:
 
 
 Note that with cascading replication, `repmgr standby follow` can also be
-used to detach a standby from its current upstream server and follow another
-upstream server, including the master.
+used to detach a standby from its current upstream server and follow the
+master. However it's currently not possible to have it follow another standby;
+we hope to improve this in a future release.
 
 
 Performing a switchover with repmgr
@@ -727,7 +802,7 @@ both passwordless SSH access and the path of `repmgr.conf` on that server.
 > careful preparation and with adequate attention. In particular you should
 > be confident that your network environment is stable and reliable.
 >
-> We recommend running `repmgr standby switchover`  at the most verbose
+> We recommend running `repmgr standby switchover` at the most verbose
 > logging level (`--log-level DEBUG --verbose`) and capturing all output
 > to assist troubleshooting any problems.
 >
@@ -793,7 +868,7 @@ should have been updated to reflect this:
 
 ### Caveats
 
-- the functionality provided `repmgr standby switchover` is primarily aimed
+- The functionality provided `repmgr standby switchover` is primarily aimed
   at a two-server master/standby replication cluster and currently does
   not support additional standbys.
 - `repmgr standby switchover` is designed to use the `pg_rewind` utility,
@@ -802,11 +877,16 @@ should have been updated to reflect this:
 - `pg_rewind` *requires* that either `wal_log_hints` is enabled, or that
    data checksums were enabled when the cluster was initialized. See the
   `pg_rewind` documentation for details:
-     http://www.postgresql.org/docs/current/static/app-pgrewind.html
+     https://www.postgresql.org/docs/current/static/app-pgrewind.html
 - `repmgrd` should not be running when a switchover is carried out, otherwise
   the `repmgrd` may try and promote a standby by itself.
 - Any other standbys attached to the old master will need to be manually
   instructed to point to the new master (e.g. with `repmgr standby follow`).
+- You must ensure that following a server start using `pg_ctl`, log output
+  is not send to STDERR (the default behaviour). If logging is not configured,
+  We recommend setting `logging_collector=on` in `postgresql.conf` and
+  providing an explicit `-l/--log` setting in `repmgr.conf`'s `pg_ctl_options`
+  parameter.
 
 We hope to remove some of these restrictions in future versions of `repmgr`.
 
@@ -860,8 +940,8 @@ Adjust schema and node ID accordingly. A future `repmgr` release
 will make it possible to unregister failed standbys.
 
 
-Automatic failover with repmgrd
--------------------------------
+Automatic failover with `repmgrd`
+---------------------------------
 
 `repmgrd` is a management and monitoring daemon which runs on standby nodes
 and which can automate actions such as failover and updating standbys to
@@ -889,7 +969,7 @@ actions happening, but we strongly recommend executing `repmgr` directly.
 
 `repmgrd` can be started simply with e.g.:
 
-    repmgrd -f /etc/repmgr.conf --verbose > $HOME/repmgr/repmgr.log 2>&1
+    repmgrd -f /etc/repmgr.conf --verbose >> $HOME/repmgr/repmgr.log 2>&1
 
 For permanent operation, we recommend using the options `-d/--daemonize` to
 detach the `repmgrd` process, and `-p/--pid-file` to write the process PID
@@ -911,7 +991,7 @@ table looks like this:
 
 
 Start `repmgrd` on each standby and verify that it's running by examining
-the log output, which at default log level will look like this:
+the log output, which at log level INFO will look like this:
 
     [2016-01-05 13:15:40] [INFO] checking cluster configuration with schema 'repmgr_test'
     [2016-01-05 13:15:40] [INFO] checking node 2 in cluster 'test'
@@ -981,8 +1061,8 @@ during the failover:
     (3 rows)
 
 
-repmgrd log rotation
---------------------
+`repmgrd` log rotation
+----------------------
 
 Note that currently `repmgrd` does not provide logfile rotation. To ensure
 the current logfile does not grow indefinitely, configure your system's `logrotate`
@@ -998,8 +1078,29 @@ for up to 52 weeks and rotation forced if a file grows beyond 100Mb:
         create 0600 postgres postgres
     }
 
-Monitoring
-----------
+
+`repmgrd` and PostgreSQL connection settings
+--------------------------------------------
+
+In addition to the `repmgr` configuration settings, parameters in the
+`conninfo` string influence how `repmgr` makes a network connection to
+PostgreSQL. In particular, if another server in the replication cluster
+is unreachable at network level, system network settings will influence
+the length of time it takes to determine that the connection is not possible.
+
+In particular explicitly setting a parameter for `connect_timeout` should
+be considered; the effective minimum value of `2` (seconds) will ensure
+that a connection failure at network level is reported as soon as possible,
+otherwise depending on the system settings (e.g. `tcp_syn_retries` in Linux)
+a delay of a minute or more is possible.
+
+For further details on `conninfo` network connection parameters, see:
+
+  https://www.postgresql.org/docs/current/static/libpq-connect.html#LIBPQ-PARAMKEYWORDS
+
+
+Monitoring with `repmgrd`
+-------------------------
 
 When `repmgrd` is running with the option `-m/--monitoring-history`, it will
 constantly write standby node status information to the `repl_monitor` table,
@@ -1031,9 +1132,16 @@ table , it's advisable to regularly purge historical data with
 `repmgr cluster cleanup`; use the `-k/--keep-history` to specify how
 many day's worth of data should be retained.
 
+It's possible to use `repmgrd` to provide monitoring only for some or all
+nodes by setting `failover = manual` in the node's `repmgr.conf`. In the
+event of the node's upstream failing, no failover action will be taken
+and the node will require manual intervention to be reattached to replication.
+If this occurs, event notification `standby_disconnect_manual` will be
+created.
+
 Note that when a standby node is not streaming directly from its upstream
-node, i.e. recovering WAL from an archive, `apply_lag` will always
-appear as `0 bytes`.
+node, e.g. recovering WAL from an archive, `apply_lag` will always appear as
+`0 bytes`.
 
 
 Using a witness server with repmgrd
@@ -1130,6 +1238,7 @@ The following event types are available:
   * `standby_promote`
   * `standby_follow`
   * `standby_switchover`
+  * `standby_disconnect_manual`
   * `witness_create`
   * `witness_create`
   * `repmgrd_start`
@@ -1291,17 +1400,32 @@ which contains connection details for the local database.
     when analyzing connectivity from a particular node.
 
     This command requires a valid `repmgr.conf` file to be provided; no
-    additional arguments are required.
+    additional arguments are needed.
 
     Example:
 
         $ repmgr -f /etc/repmgr.conf cluster show
 
         Role      | Name  | Upstream | Connection String
-        ----------+-------|----------|--------------------------------------------
-        * master  | node1 |          | host=repmgr_node1 dbname=repmgr user=repmgr
-          standby | node2 | node1    | host=repmgr_node1 dbname=repmgr user=repmgr
-          standby | node3 | node2    | host=repmgr_node1 dbname=repmgr user=repmgr
+        ----------+-------|----------|----------------------------------------
+        * master  | node1 |          | host=db_node1 dbname=repmgr user=repmgr
+          standby | node2 | node1    | host=db_node2 dbname=repmgr user=repmgr
+          standby | node3 | node2    | host=db_node3 dbname=repmgr user=repmgr
+
+    To show database connection errors when polling nodes, run the command in
+    `--verbose` mode.
+
+    The `cluster show` command now accepts the optional parameter `--csv`, which
+    outputs the replication cluster's status in a simple CSV format, suitable for
+    parsing by scripts:
+
+        $ repmgr -f /etc/repmgr.conf cluster show --csv
+        1,-1
+        2,0
+        3,1
+
+    The first column is the node's ID, and the second column represents the
+    node's status (0 = master, 1 = standby, -1 = failed).
 
 * `cluster cleanup`
 
@@ -1320,20 +1444,22 @@ which contains connection details for the local database.
 `repmgr` or `repmgrd` will return one of the following error codes on program
 exit:
 
-* SUCCESS (0)              Program ran successfully.
-* ERR_BAD_CONFIG (1)       Configuration file could not be parsed or was invalid
-* ERR_BAD_RSYNC (2)        An rsync call made by the program returned an error
-* ERR_NO_RESTART (4)       An attempt to restart a PostgreSQL instance failed
-* ERR_DB_CON (6)           Error when trying to connect to a database
-* ERR_DB_QUERY (7)         Error while executing a database query
-* ERR_PROMOTED (8)         Exiting program because the node has been promoted to master
-* ERR_BAD_PASSWORD (9)     Password used to connect to a database was rejected
-* ERR_STR_OVERFLOW (10)    String overflow error
-* ERR_FAILOVER_FAIL (11)   Error encountered during failover (repmgrd only)
-* ERR_BAD_SSH (12)         Error when connecting to remote host via SSH
-* ERR_SYS_FAILURE (13)     Error when forking (repmgrd only)
-* ERR_BAD_BASEBACKUP (14)  Error when executing pg_basebackup
-* ERR_MONITORING_FAIL (16) Unrecoverable error encountered during monitoring (repmgrd only)
+* SUCCESS (0)               Program ran successfully.
+* ERR_BAD_CONFIG (1)        Configuration file could not be parsed or was invalid
+* ERR_BAD_RSYNC (2)         An rsync call made by the program returned an error (repmgr only)
+* ERR_NO_RESTART (4)        An attempt to restart a PostgreSQL instance failed
+* ERR_DB_CON (6)            Error when trying to connect to a database
+* ERR_DB_QUERY (7)          Error while executing a database query
+* ERR_PROMOTED (8)          Exiting program because the node has been promoted to master
+* ERR_STR_OVERFLOW (10)     String overflow error
+* ERR_FAILOVER_FAIL (11)    Error encountered during failover (repmgrd only)
+* ERR_BAD_SSH (12)          Error when connecting to remote host via SSH (repmgr only)
+* ERR_SYS_FAILURE (13)      Error when forking (repmgrd only)
+* ERR_BAD_BASEBACKUP (14)   Error when executing pg_basebackup (repmgr only)
+* ERR_MONITORING_FAIL (16)  Unrecoverable error encountered during monitoring (repmgrd only)
+* ERR_BAD_BACKUP_LABEL (17) Corrupt or unreadable backup label encountered (repmgr only)
+* ERR_SWITCHOVER_FAIL (18)  Error encountered during switchover (repmgr only)
+
 
 Support and Assistance
 ----------------------
@@ -1379,5 +1505,6 @@ Thanks from the repmgr core team.
 Further reading
 ---------------
 
+* http://blog.2ndquadrant.com/improvements-in-repmgr-3-1-4/
 * http://blog.2ndquadrant.com/managing-useful-clusters-repmgr/
 * http://blog.2ndquadrant.com/easier_postgresql_90_clusters/

RHEL/repmgr3-93.spec

@@ -1,61 +0,0 @@
-Summary: repmgr
-Name: repmgr
-Version: 3.0
-Release: 1
-License: GPLv3
-Group: System Environment/Daemons
-URL: http://repmgr.org
-Packager: Ian Barwick <ian@2ndquadrant.com>
-Vendor: 2ndQuadrant Limited
-Distribution: centos
-Source0: %{name}-%{version}.tar.gz
-BuildRoot: %{_tmppath}/%{name}-%{version}-%{release}-root
-
-%description
-repmgr is a utility suite which greatly simplifies
-the process of setting up and managing replication
-using streaming replication within a cluster of
-PostgreSQL servers.
-
-%prep
-%setup
-
-%build
-export PATH=$PATH:/usr/pgsql-9.3/bin/
-%{__make} USE_PGXS=1
-
-%install
-[ "%{buildroot}" != "/" ] && %{__rm} -rf %{buildroot}
-
-export PATH=$PATH:/usr/pgsql-9.3/bin/
-%{__make} USE_PGXS=1 install DESTDIR=%{buildroot} INSTALL="install -p"
-%{__make} USE_PGXS=1 install_prog DESTDIR=%{buildroot} INSTALL="install -p"
-%{__make} USE_PGXS=1 install_rhel DESTDIR=%{buildroot} INSTALL="install -p"
-
-
-%clean
-[ "%{buildroot}" != "/" ] && %{__rm} -rf %{buildroot}
-
-
-%files
-%defattr(-,root,root)
-/usr/bin/repmgr
-/usr/bin/repmgrd
-/usr/pgsql-9.3/bin/repmgr
-/usr/pgsql-9.3/bin/repmgrd
-/usr/pgsql-9.3/lib/repmgr_funcs.so
-/usr/pgsql-9.3/share/contrib/repmgr.sql
-/usr/pgsql-9.3/share/contrib/repmgr_funcs.sql
-/usr/pgsql-9.3/share/contrib/uninstall_repmgr.sql
-/usr/pgsql-9.3/share/contrib/uninstall_repmgr_funcs.sql
-%attr(0755,root,root)/etc/init.d/repmgrd
-%attr(0644,root,root)/etc/sysconfig/repmgrd
-%attr(0644,root,root)/etc/repmgr/repmgr.conf.sample
-
-%changelog
-* Tue Mar 10 2015 Ian Barwick ian@2ndquadrant.com>
-- build for repmgr 3.0
-* Thu Jun 05 2014 Nathan Van Overloop <nathan.van.overloop@nexperteam.be> 2.0.2
-- fix witness creation to create db and user if needed
-* Fri Apr 04 2014 Nathan Van Overloop <nathan.van.overloop@nexperteam.be> 2.0.1
-- initial build for RHEL6

RHEL/repmgrd.init

@@ -1,133 +0,0 @@
-#!/bin/sh
-#
-# chkconfig: - 75 16
-# description: Enable repmgrd replication management and monitoring daemon for PostgreSQL
-# processname: repmgrd
-# pidfile="/var/run/${NAME}.pid"
-
-# Source function library.
-INITD=/etc/rc.d/init.d
-. $INITD/functions
-
-# Get function listing for cross-distribution logic.
-TYPESET=`typeset -f|grep "declare"`
-
-# Get network config.
-. /etc/sysconfig/network
-
-DESC="PostgreSQL replication management and monitoring daemon"
-NAME=repmgrd
-
-REPMGRD_ENABLED=no
-REPMGRD_OPTS=
-REPMGRD_USER=postgres
-REPMGRD_BIN=/usr/pgsql-9.3/bin/repmgrd
-REPMGRD_PIDFILE=/var/run/repmgrd.pid
-REPMGRD_LOCK=/var/lock/subsys/${NAME}
-REPMGRD_LOG=/var/lib/pgsql/9.3/data/pg_log/repmgrd.log
-
-# Read configuration variable file if it is present
-[ -r /etc/sysconfig/$NAME ] && . /etc/sysconfig/$NAME
-
-# For SELinux we need to use 'runuser' not 'su'
-if [ -x /sbin/runuser ]
-then
-    SU=runuser
-else
-    SU=su
-fi
-
-test -x $REPMGRD_BIN || exit 0
-
-case "$REPMGRD_ENABLED" in
-    [Yy]*)
-	break
-	;;
-    *)
-	exit 0
-	;;
-esac
-
-
-if [ -z "${REPMGRD_OPTS}" ]
-then
-    echo "Not starting ${NAME}, REPMGRD_OPTS not set in /etc/sysconfig/${NAME}"
-    exit 0
-fi
-
-start()
-{
-    REPMGRD_START=$"Starting ${NAME} service: "
-
-    # Make sure startup-time log file is valid
-    if [ ! -e "${REPMGRD_LOG}" -a ! -h "${REPMGRD_LOG}" ]
-    then
-        touch "${REPMGRD_LOG}" || exit 1
-        chown ${REPMGRD_USER}:postgres "${REPMGRD_LOG}"
-        chmod go-rwx "${REPMGRD_LOG}"
-        [ -x /sbin/restorecon ] && /sbin/restorecon "${REPMGRD_LOG}"
-    fi
-
-    echo -n "${REPMGRD_START}"
-    $SU -l $REPMGRD_USER -c "${REPMGRD_BIN} ${REPMGRD_OPTS} -p ${REPMGRD_PIDFILE} &" >> "${REPMGRD_LOG}" 2>&1 < /dev/null
-    sleep 2
-    pid=`head -n 1 "${REPMGRD_PIDFILE}" 2>/dev/null`
-    if [ "x${pid}" != "x" ]
-    then
-        success "${REPMGRD_START}"
-        touch "${REPMGRD_LOCK}"
-        echo $pid > "${REPMGRD_PIDFILE}"
-        echo
-    else
-        failure "${REPMGRD_START}"
-        echo
-        script_result=1
-    fi
-}
-
-stop()
-{
-    echo -n $"Stopping ${NAME} service: "
-    if [ -e "${REPMGRD_LOCK}" ]
-    then
-        killproc ${NAME}
-        ret=$? 
-        if [ $ret -eq 0 ]
-        then
-            echo_success
-            rm -f "${REPMGRD_PIDFILE}"
-            rm -f "${REPMGRD_LOCK}"
-        else
-            echo_failure
-            script_result=1
-        fi
-    else
-        # not running; per LSB standards this is "ok"   
-        echo_success
-    fi
-    echo
-}
-
-
-# See how we were called.
-case "$1" in
-  start)
-        start
-        ;;
-  stop)
-        stop
-        ;;
-  status)
-        status -p $REPMGRD_PIDFILE $NAME
-        script_result=$?
-        ;;
-  restart)
-        stop
-	start
-        ;;
-  *)
-        echo $"Usage: $0 {start|stop|status|restart}"
-        exit 2
-esac
-
-exit $script_result

RHEL/repmgrd.sysconfig

@@ -1,21 +0,0 @@
-# default settings for repmgrd. This file is source by /bin/sh from
-# /etc/init.d/repmgrd
-
-# disable repmgrd by default so it won't get started upon installation
-# valid values: yes/no
-REPMGRD_ENABLED=no
-
-# Options for repmgrd (required)
-#REPMGRD_OPTS="--verbose -d -f /var/lib/pgsql/repmgr/repmgr.conf"
-
-# User to run repmgrd as
-#REPMGRD_USER=postgres
-
-# repmgrd binary
-#REPMGRD_BIN=/usr/bin/repmgrd
-
-# pid file
-#REPMGRD_PIDFILE=/var/lib/pgsql/repmgr/repmgrd.pid
-
-# log file
-#REPMGRD_LOG=/var/lib/pgsql/repmgr/repmgrd.log

TODO

@@ -53,8 +53,9 @@ Planned feature improvements
   requested, activate the replication slot using pg_receivexlog to negate the
   need to set `wal_keep_segments` just for the initial clone (9.4 and 9.5).
 
-* Take into account the fact that a standby can obtain WAL from an archive,
-  so even if direct streaming replication is interrupted, it may be up-to-date
+* repmgr: enable "standby follow" to point a standby at another standby, not
+  just the replication cluster master (see GitHub #130)
+
 
 Usability improvements
 ======================

config.c

@@ -28,7 +28,7 @@ static void parse_event_notifications_list(t_configuration_options *options, con
 static void tablespace_list_append(t_configuration_options *options, const char *arg);
 static void exit_with_errors(ErrorList *config_errors);
 
-const static char *_progname = '\0';
+const static char *_progname = NULL;
 static char config_file_path[MAXPGPATH];
 static bool config_file_provided = false;
 bool config_file_found = false;
@@ -219,11 +219,15 @@ parse_config(t_configuration_options *options)
 	memset(options->node_name, 0, sizeof(options->node_name));
 	memset(options->promote_command, 0, sizeof(options->promote_command));
 	memset(options->follow_command, 0, sizeof(options->follow_command));
+	memset(options->stop_command, 0, sizeof(options->stop_command));
+	memset(options->start_command, 0, sizeof(options->start_command));
+	memset(options->restart_command, 0, sizeof(options->restart_command));
 	memset(options->rsync_options, 0, sizeof(options->rsync_options));
 	memset(options->ssh_options, 0, sizeof(options->ssh_options));
 	memset(options->pg_bindir, 0, sizeof(options->pg_bindir));
 	memset(options->pg_ctl_options, 0, sizeof(options->pg_ctl_options));
 	memset(options->pg_basebackup_options, 0, sizeof(options->pg_basebackup_options));
+	memset(options->restore_command, 0, sizeof(options->restore_command));
 
 	/* default master_response_timeout is 60 seconds */
 	options->master_response_timeout = 60;
@@ -239,6 +243,8 @@ parse_config(t_configuration_options *options)
 	options->witness_repl_nodes_sync_interval_secs = 30;
 
 	memset(options->event_notification_command, 0, sizeof(options->event_notification_command));
+	options->event_notifications.head = NULL;
+	options->event_notifications.tail = NULL;
 
 	options->tablespace_mapping.head = NULL;
 	options->tablespace_mapping.tail = NULL;
@@ -338,9 +344,16 @@ parse_config(t_configuration_options *options)
 			strncpy(options->promote_command, value, MAXLEN);
 		else if (strcmp(name, "follow_command") == 0)
 			strncpy(options->follow_command, value, MAXLEN);
+		else if (strcmp(name, "stop_command") == 0)
+			strncpy(options->stop_command, value, MAXLEN);
+		else if (strcmp(name, "start_command") == 0)
+			strncpy(options->start_command, value, MAXLEN);
+		else if (strcmp(name, "restart_command") == 0)
+			strncpy(options->restart_command, value, MAXLEN);
 		else if (strcmp(name, "master_response_timeout") == 0)
 			options->master_response_timeout = repmgr_atoi(value, "master_response_timeout", &config_errors, false);
-		/* 'primary_response_timeout' as synonym for 'master_response_timeout' -
+		/*
+		 * 'primary_response_timeout' as synonym for 'master_response_timeout' -
 		 * we'll switch terminology in a future release (3.1?)
 		 */
 		else if (strcmp(name, "primary_response_timeout") == 0)
@@ -372,6 +385,8 @@ parse_config(t_configuration_options *options)
 			parse_event_notifications_list(options, value);
 		else if (strcmp(name, "tablespace_mapping") == 0)
 			tablespace_list_append(options, value);
+		else if (strcmp(name, "restore_command") == 0)
+			strncpy(options->restore_command, value, MAXLEN);
 		else
 		{
 			known_parameter = false;

config.h

@@ -62,6 +62,9 @@ typedef struct
 	char		node_name[MAXLEN];
 	char		promote_command[MAXLEN];
 	char		follow_command[MAXLEN];
+	char		stop_command[MAXLEN];
+	char		start_command[MAXLEN];
+	char		restart_command[MAXLEN];
 	char		loglevel[MAXLEN];
 	char		logfacility[MAXLEN];
 	char		rsync_options[QUERY_STR_LEN];
@@ -72,6 +75,7 @@ typedef struct
 	char		pg_bindir[MAXLEN];
 	char		pg_ctl_options[MAXLEN];
 	char		pg_basebackup_options[MAXLEN];
+	char		restore_command[MAXLEN];
 	char		logfile[MAXLEN];
 	int			monitor_interval_secs;
 	int			retry_promote_interval_secs;
@@ -82,7 +86,11 @@ typedef struct
 	TablespaceList tablespace_mapping;
 }	t_configuration_options;
 
-#define T_CONFIGURATION_OPTIONS_INITIALIZER { "", -1, NO_UPSTREAM_NODE, "", MANUAL_FAILOVER, -1, "", "", "", "", "", "", "", -1, -1, -1, "", "", "", "", 0, 0, 0, 0, "", { NULL, NULL }, {NULL, NULL} }
+/*
+ * The following will initialize the structure with a minimal set of options;
+ * actual defaults are set in parse_config() before parsing the configuration file
+ */
+#define T_CONFIGURATION_OPTIONS_INITIALIZER { "", -1, NO_UPSTREAM_NODE, "", MANUAL_FAILOVER, -1, "", "", "", "", "", "", "", "", "", "", -1, -1, -1, "", "", "", "", "", 0, 0, 0, 0, "", { NULL, NULL }, {NULL, NULL} }
 
 typedef struct ErrorListCell
 {

dbutils.c

@@ -31,9 +31,10 @@
 char repmgr_schema[MAXLEN] = "";
 char repmgr_schema_quoted[MAXLEN] = "";
 
+static int _get_node_record(PGconn *conn, char *cluster, char *sqlquery, t_node_info *node_info);
 
 PGconn *
-_establish_db_connection(const char *conninfo, const bool exit_on_error, const bool log_notice)
+_establish_db_connection(const char *conninfo, const bool exit_on_error, const bool log_notice, const bool verbose_only)
 {
 	/* Make a connection to the database */
 	PGconn	   *conn = NULL;
@@ -49,15 +50,23 @@ _establish_db_connection(const char *conninfo, const bool exit_on_error, const b
 	/* Check to see that the backend connection was successfully made */
 	if ((PQstatus(conn) != CONNECTION_OK))
 	{
-		if (log_notice)
+		bool emit_log = true;
+
+		if (verbose_only == true && verbose_logging == false)
+			emit_log = false;
+
+		if (emit_log)
 		{
-			log_notice(_("connection to database failed: %s\n"),
-					PQerrorMessage(conn));
-		}
-		else
-		{
-			log_err(_("connection to database failed: %s\n"),
-					PQerrorMessage(conn));
+			if (log_notice)
+			{
+				log_notice(_("connection to database failed: %s\n"),
+						   PQerrorMessage(conn));
+			}
+			else
+			{
+				log_err(_("connection to database failed: %s\n"),
+						PQerrorMessage(conn));
+			}
 		}
 
 		if (exit_on_error)
@@ -70,16 +79,35 @@ _establish_db_connection(const char *conninfo, const bool exit_on_error, const b
 	return conn;
 }
 
+
+/*
+ * Establish a database connection, optionally exit on error
+ */
 PGconn *
 establish_db_connection(const char *conninfo, const bool exit_on_error)
 {
-	return _establish_db_connection(conninfo, exit_on_error, false);
+	return _establish_db_connection(conninfo, exit_on_error, false, false);
 }
 
+/*
+ * Attempt to establish a database connection, never exit on error, only
+ * output error messages if --verbose option used
+ */
 PGconn *
-test_db_connection(const char *conninfo, const bool exit_on_error)
+establish_db_connection_quiet(const char *conninfo)
 {
-	return _establish_db_connection(conninfo, exit_on_error, true);
+	return _establish_db_connection(conninfo, false, false, true);
+}
+
+/*
+ * Attempt to establish a database connection, never exit on error,
+ * output connection error messages as NOTICE (useful when connection
+ * failure is expected)
+ */
+PGconn *
+test_db_connection(const char *conninfo)
+{
+	return _establish_db_connection(conninfo, false, true, false);
 }
 
 
@@ -538,7 +566,7 @@ get_conninfo_value(const char *conninfo, const char *keyword, char *output)
 
 	conninfo_options = PQconninfoParse(conninfo, NULL);
 
-	if (conninfo_options == false)
+	if (conninfo_options == NULL)
 	{
 		log_err(_("Unable to parse provided conninfo string \"%s\""), conninfo);
 		return false;
@@ -1681,8 +1709,7 @@ int
 get_node_record(PGconn *conn, char *cluster, int node_id, t_node_info *node_info)
 {
 	char		sqlquery[QUERY_STR_LEN];
-	PGresult   *res;
-	int         ntuples;
+	int		    result;
 
 	sqlquery_snprintf(
 		sqlquery,
@@ -1696,6 +1723,51 @@ get_node_record(PGconn *conn, char *cluster, int node_id, t_node_info *node_info
 
 	log_verbose(LOG_DEBUG, "get_node_record():\n%s\n", sqlquery);
 
+	result = _get_node_record(conn, cluster, sqlquery, node_info);
+
+	if (result == 0)
+	{
+		log_verbose(LOG_DEBUG, "get_node_record(): no record found for node %i\n", node_id);
+	}
+
+	return result;
+}
+
+int
+get_node_record_by_name(PGconn *conn, char *cluster, const char *node_name, t_node_info *node_info)
+{
+	char		sqlquery[QUERY_STR_LEN];
+	int result;
+
+	sqlquery_snprintf(
+		sqlquery,
+		"SELECT id, type, upstream_node_id, name, conninfo, slot_name, priority, active"
+		"  FROM %s.repl_nodes "
+		" WHERE cluster = '%s' "
+		"   AND name = '%s'",
+		get_repmgr_schema_quoted(conn),
+		cluster,
+		node_name);
+
+	log_verbose(LOG_DEBUG, "get_node_record_by_name():\n%s\n", sqlquery);
+
+	result = _get_node_record(conn, cluster, sqlquery, node_info);
+
+	if (result == 0)
+	{
+		log_verbose(LOG_DEBUG, "get_node_record(): no record found for node %s\n", node_name);
+	}
+
+	return result;
+}
+
+
+static int
+_get_node_record(PGconn *conn, char *cluster, char *sqlquery, t_node_info *node_info)
+{
+	int         ntuples;
+	PGresult   *res;
+
 	res = PQexec(conn, sqlquery);
 	if (PQresultStatus(res) != PGRES_TUPLES_OK)
 	{
@@ -1706,7 +1778,6 @@ get_node_record(PGconn *conn, char *cluster, int node_id, t_node_info *node_info
 
 	if (ntuples == 0)
 	{
-		log_verbose(LOG_DEBUG, "get_node_record(): no record found for node %i\n", node_id);
 		return 0;
 	}
 
@@ -1727,6 +1798,9 @@ get_node_record(PGconn *conn, char *cluster, int node_id, t_node_info *node_info
 }
 
 
+
+
+
 int
 get_node_replication_state(PGconn *conn, char *node_name, char *output)
 {

dbutils.h

@@ -52,18 +52,6 @@ typedef struct s_node_info
 }	t_node_info;
 
 
-/*
- * Struct to store replication slot information
- */
-
-typedef struct s_replication_slot
-{
-	char slot_name[MAXLEN];
-    char slot_type[MAXLEN];
-	bool active;
-}   t_replication_slot;
-
-
 #define T_NODE_INFO_INITIALIZER { \
   NODE_NOT_FOUND, \
   NO_UPSTREAM_NODE, \
@@ -78,13 +66,27 @@ typedef struct s_replication_slot
   InvalidXLogRecPtr \
 }
 
+/*
+ * Struct to store replication slot information
+ */
+
+typedef struct s_replication_slot
+{
+	char slot_name[MAXLEN];
+    char slot_type[MAXLEN];
+	bool active;
+}   t_replication_slot;
+
+
+
 PGconn *_establish_db_connection(const char *conninfo,
 								 const bool exit_on_error,
-								 const bool log_notice);
+								 const bool log_notice,
+								 const bool verbose_only);
 PGconn *establish_db_connection(const char *conninfo,
 								const bool exit_on_error);
-PGconn *test_db_connection(const char *conninfo,
-						   const bool exit_on_error);
+PGconn *establish_db_connection_quiet(const char *conninfo);
+PGconn *test_db_connection(const char *conninfo);
 PGconn *establish_db_connection_by_params(const char *keywords[],
 								  const char *values[],
 								  const bool exit_on_error);
@@ -125,6 +127,7 @@ bool		witness_copy_node_records(PGconn *masterconn, PGconn *witnessconn, char *c
 bool		create_node_record(PGconn *conn, char *action, int node, char *type, int upstream_node, char *cluster_name, char *node_name, char *conninfo, int priority, char *slot_name, bool active);
 bool		delete_node_record(PGconn *conn, int node, char *action);
 int			get_node_record(PGconn *conn, char *cluster, int node_id, t_node_info *node_info);
+int			get_node_record_by_name(PGconn *conn, char *cluster, const char *node_name, t_node_info *node_info);
 bool        update_node_record_status(PGconn *conn, char *cluster_name, int this_node_id, char *type, int upstream_node_id, bool active);
 bool        update_node_record_set_upstream(PGconn *conn, char *cluster_name, int this_node_id, int new_upstream_node_id);
 bool        create_event_record(PGconn *conn, t_configuration_options *options, int node_id, char *event, bool successful, char *details);

debian/DEBIAN/control

@@ -1,9 +1,9 @@
 Package: repmgr-auto
-Version: 3.0.1
+Version: 3.1.3
 Section: database
 Priority: optional
 Architecture: all
-Depends: rsync, postgresql-9.3 | postgresql-9.4
+Depends: rsync, postgresql-9.3 | postgresql-9.4 | postgresql-9.5
 Maintainer: Self built package <user@localhost>
 Description: PostgreSQL replication setup, magament and monitoring
  has two main executables

dirmod.c

@@ -0,0 +1,194 @@
+/*
+ *
+ * dirmod.c
+ *	  directory handling functions
+ *
+ * Copyright (C) 2ndQuadrant, 2010-2016
+ *
+ * Portions Copyright (c) 1996-2013, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ *
+ */
+
+#include "postgres_fe.h"
+
+/* Don't modify declarations in system headers */
+
+#include <unistd.h>
+#include <dirent.h>
+#include <sys/stat.h>
+
+/*
+ * pgfnames
+ *
+ * return a list of the names of objects in the argument directory.  Caller
+ * must call pgfnames_cleanup later to free the memory allocated by this
+ * function.
+ */
+char	  **
+pgfnames(const char *path)
+{
+	DIR		   *dir;
+	struct dirent *file;
+	char	  **filenames;
+	int			numnames = 0;
+	int			fnsize = 200;	/* enough for many small dbs */
+
+	dir = opendir(path);
+	if (dir == NULL)
+	{
+		return NULL;
+	}
+
+	filenames = (char **) palloc(fnsize * sizeof(char *));
+
+	while (errno = 0, (file = readdir(dir)) != NULL)
+	{
+		if (strcmp(file->d_name, ".") != 0 && strcmp(file->d_name, "..") != 0)
+		{
+			if (numnames + 1 >= fnsize)
+			{
+				fnsize *= 2;
+				filenames = (char **) repalloc(filenames,
+											   fnsize * sizeof(char *));
+			}
+			filenames[numnames++] = pstrdup(file->d_name);
+		}
+	}
+
+	if (errno)
+	{
+		fprintf(stderr, _("could not read directory \"%s\": %s\n"),
+				path, strerror(errno));
+	}
+
+	filenames[numnames] = NULL;
+
+	if (closedir(dir))
+	{
+		fprintf(stderr, _("could not close directory \"%s\": %s\n"),
+				path, strerror(errno));
+	}
+
+	return filenames;
+}
+
+
+/*
+ *	pgfnames_cleanup
+ *
+ *	deallocate memory used for filenames
+ */
+void
+pgfnames_cleanup(char **filenames)
+{
+	char	  **fn;
+
+	for (fn = filenames; *fn; fn++)
+		pfree(*fn);
+
+	pfree(filenames);
+}
+
+
+/*
+ *	rmtree
+ *
+ *	Delete a directory tree recursively.
+ *	Assumes path points to a valid directory.
+ *	Deletes everything under path.
+ *	If rmtopdir is true deletes the directory too.
+ *	Returns true if successful, false if there was any problem.
+ *	(The details of the problem are reported already, so caller
+ *	doesn't really have to say anything more, but most do.)
+ */
+bool
+rmtree(const char *path, bool rmtopdir)
+{
+	bool		result = true;
+	char		pathbuf[MAXPGPATH];
+	char	  **filenames;
+	char	  **filename;
+	struct stat statbuf;
+
+	/*
+	 * we copy all the names out of the directory before we start modifying
+	 * it.
+	 */
+	filenames = pgfnames(path);
+
+	if (filenames == NULL)
+		return false;
+
+	/* now we have the names we can start removing things */
+	for (filename = filenames; *filename; filename++)
+	{
+		snprintf(pathbuf, MAXPGPATH, "%s/%s", path, *filename);
+
+		/*
+		 * It's ok if the file is not there anymore; we were just about to
+		 * delete it anyway.
+		 *
+		 * This is not an academic possibility. One scenario where this
+		 * happens is when bgwriter has a pending unlink request for a file in
+		 * a database that's being dropped. In dropdb(), we call
+		 * ForgetDatabaseFsyncRequests() to flush out any such pending unlink
+		 * requests, but because that's asynchronous, it's not guaranteed that
+		 * the bgwriter receives the message in time.
+		 */
+		if (lstat(pathbuf, &statbuf) != 0)
+		{
+			if (errno != ENOENT)
+			{
+				result = false;
+			}
+			continue;
+		}
+
+		if (S_ISDIR(statbuf.st_mode))
+		{
+			/* call ourselves recursively for a directory */
+			if (!rmtree(pathbuf, true))
+			{
+				/* we already reported the error */
+				result = false;
+			}
+		}
+		else
+		{
+			if (unlink(pathbuf) != 0)
+			{
+				if (errno != ENOENT)
+				{
+					result = false;
+				}
+			}
+		}
+	}
+
+	if (rmtopdir)
+	{
+		if (rmdir(path) != 0)
+		{
+			result = false;
+		}
+	}
+
+	pgfnames_cleanup(filenames);
+
+	return result;
+}
+

dirmod.h

@@ -0,0 +1,23 @@
+/*
+ * dirmod.h
+ * Copyright (c) 2ndQuadrant, 2010-2016
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ *
+ */
+
+#ifndef _DIRMOD_H_
+#define _DIRMOD_H_
+
+#endif

errcode.h

@@ -29,7 +29,6 @@
 #define ERR_DB_CON 6
 #define ERR_DB_QUERY 7
 #define ERR_PROMOTED 8
-#define ERR_BAD_PASSWORD 9
 #define ERR_STR_OVERFLOW 10
 #define ERR_FAILOVER_FAIL 11
 #define ERR_BAD_SSH 12
@@ -38,5 +37,6 @@
 #define ERR_INTERNAL 15
 #define ERR_MONITORING_FAIL 16
 #define ERR_BAD_BACKUP_LABEL 17
+#define ERR_SWITCHOVER_FAIL 18
 
 #endif   /* _ERRCODE_H_ */

log.c

@@ -142,7 +142,7 @@ log_verbose(int level, const char *fmt, ...)
 
 
 bool
-logger_init(t_configuration_options * opts, const char *ident)
+logger_init(t_configuration_options *opts, const char *ident)
 {
 	char	   *level = opts->loglevel;
 	char	   *facility = opts->logfacility;

log.h

@@ -130,5 +130,7 @@ __attribute__((format(PG_PRINTF_ATTRIBUTE, 2, 3)));
 
 extern int	log_type;
 extern int	log_level;
+extern int	verbose_logging;
+extern int	terse_logging;
 
-#endif
+#endif /* _REPMGR_LOG_H_ */

repmgr.conf.sample

@@ -19,7 +19,7 @@
 
 # Node ID and name
 # (Note: we recommend to avoid naming nodes after their initial
-#  replication funcion, as this will cause confusion when e.g.
+#  replication function, as this will cause confusion when e.g.
 #  "standby2" is promoted to primary)
 #node=2           # a unique integer
 #node_name=node2  # an arbitrary (but unique) string; we recommend using
@@ -28,8 +28,16 @@
 
 # Database connection information as a conninfo string
 # This must be accessible to all servers in the cluster; for details see:
-#   http://www.postgresql.org/docs/current/static/libpq-connect.html#LIBPQ-CONNSTRING
+#
+#   https://www.postgresql.org/docs/current/static/libpq-connect.html#LIBPQ-CONNSTRING
+#
 #conninfo='host=192.168.204.104 dbname=repmgr_db user=repmgr_usr'
+#
+# If repmgrd is in use, consider explicitly setting `connect_timeout` in the
+# conninfo string to determine the length of time which elapses before
+# a network connection attempt is abandoned; for details see:
+#
+#   https://www.postgresql.org/docs/current/static/libpq-connect.html#LIBPQ-CONNECT-CONNECT-TIMEOUT
 
 # Optional configuration items
 # ============================
@@ -93,6 +101,29 @@
 # (if not provided, defaults to system $PATH)
 #pg_bindir=/usr/bin/
 
+# service control commands
+#
+# repmgr provides options to to override the default pg_ctl commands
+# used to stop, start  and restart the PostgreSQL cluster
+#
+# NOTE: These commands must be runnable on remote nodes as well for switchover
+# to function correctly.
+#
+# If you use sudo, the user repmgr runs as (usually 'postgres')  must have
+# passwordless sudo access to execute the command
+#
+# For example, to use systemd, you may use the following configuration:
+#
+#    # this is required when running sudo over ssh without -t:
+#    Defaults:postgres !requiretty
+#    postgres ALL = NOPASSWD: /usr/bin/systemctl stop postgresql-9.5, \
+#       /usr/bin/systemctl start postgresql-9.5, \
+#       /usr/bin/systemctl restart postgresql-9.5
+#
+# start_command = systemctl start postgresql-9.5
+# stop_command = systemctl stop postgresql-9.5
+# restart_command = systemctl restart postgresql-9.5
+
 # external command options
 
 #rsync_options=--archive --checksum --compress --progress --rsh="ssh -o \"StrictHostKeyChecking no\""
@@ -113,6 +144,10 @@
 #
 # tablespace_mapping=/path/to/original/tablespace=/path/to/new/tablespace
 
+# You can specify a restore_command to be used in the recovery.conf that
+# will be placed in the cloned standby
+#
+# restore_command = cp /path/to/archived/wals/%f %p
 
 # Failover settings (repmgrd)
 # ---------------------------
@@ -132,10 +167,18 @@
 #reconnect_interval=10
 
 # Autofailover options
-#failover=manual    # one of 'automatic', 'manual'
-                    # (default: manual)
-#priority=100       # a value of zero or less prevents the node being promoted to primary
+#failover=manual    # one of 'automatic', 'manual' (default: manual)
+                    # defines the action to take in the event of upstream failure
+                    #
+                    # 'automatic': repmgrd will automatically attempt to promote the
+                    #    node or follow the new upstream node
+                    # 'manual': repmgrd will take no action and the mode will require
+                    #    manual attention to reattach it to replication
+
+#priority=100       # indicate a preferred priorty for promoting nodes
+                    # a value of zero or less prevents the node being promoted to primary
                     # (default: 100)
+
 #promote_command='repmgr standby promote -f /path/to/repmgr.conf'
 #follow_command='repmgr standby follow -f /path/to/repmgr.conf -W'
 

repmgr.h

@@ -28,6 +28,7 @@
 #include "dbutils.h"
 #include "errcode.h"
 #include "config.h"
+#include "dirmod.h"
 
 #define MIN_SUPPORTED_VERSION		"9.3"
 #define MIN_SUPPORTED_VERSION_NUM	90300
@@ -46,6 +47,15 @@
 #define NO_UPSTREAM_NODE	-1
 #define UNKNOWN_NODE_ID     -1
 
+#define OPT_HELP                         1
+#define OPT_CHECK_UPSTREAM_CONFIG        2
+#define OPT_RECOVERY_MIN_APPLY_DELAY     3
+#define OPT_IGNORE_EXTERNAL_CONFIG_FILES 4
+#define OPT_CONFIG_ARCHIVE_DIR           5
+#define OPT_PG_REWIND                    6
+#define OPT_PWPROMPT                     7
+#define OPT_CSV                          8
+#define OPT_INITDB_NO_PWPROMPT           9
 
 
 /* Run time options type */
@@ -69,7 +79,7 @@ typedef struct
 	bool		rsync_only;
 	bool		fast_checkpoint;
 	bool		ignore_external_config_files;
-	char		pg_ctl_mode[MAXLEN];
+	bool		csv_mode;
 	char		masterport[MAXLEN];
 	/*
 	 * configuration file parameters which can be overridden on the
@@ -80,6 +90,7 @@ typedef struct
 	/* parameter used by STANDBY SWITCHOVER */
 	char		remote_config_file[MAXLEN];
 	char		pg_rewind[MAXPGPATH];
+	char		pg_ctl_mode[MAXLEN];
 	/* parameter used by STANDBY {ARCHIVE_CONFIG | RESTORE_CONFIG} */
 	char		config_archive_dir[MAXLEN];
 	/* parameter used by CLUSTER CLEANUP */
@@ -90,11 +101,10 @@ typedef struct
 	char		recovery_min_apply_delay[MAXLEN];
 
 	/* deprecated command line options */
-	char		localport[MAXLEN];
-	bool		initdb_no_pwprompt;
+	char            localport[MAXLEN];
 }	t_runtime_options;
 
-#define T_RUNTIME_OPTIONS_INITIALIZER { "", "", "", "", "", "", "", DEFAULT_WAL_KEEP_SEGMENTS, false, false, false, false, false, false, false, false, false, "smart", "", "", "", "", "", 0, "", "", "", false }
+#define T_RUNTIME_OPTIONS_INITIALIZER { "", "", "", "", "", "", "", DEFAULT_WAL_KEEP_SEGMENTS, false, false, false, false, false, false, false, false, false, false, "", "", "", "", "fast", "", 0, "", "", ""}
 
 struct BackupLabel
 {

repmgrd.c

@@ -41,14 +41,17 @@
 #include "access/xlogdefs.h"
 #include "pqexpbuffer.h"
 
+/* Message strings passed in repmgrSharedState->location */
 
+#define PASSIVE_NODE "PASSIVE_NODE"
+#define LSN_QUERY_ERROR "LSN_QUERY_ERROR"
 
 /* Local info */
-t_configuration_options local_options;
+t_configuration_options local_options = T_CONFIGURATION_OPTIONS_INITIALIZER;
 PGconn	   *my_local_conn = NULL;
 
 /* Master info */
-t_configuration_options master_options;
+t_configuration_options master_options = T_CONFIGURATION_OPTIONS_INITIALIZER;
 
 PGconn	   *master_conn = NULL;
 
@@ -59,9 +62,14 @@ t_node_info node_info;
 
 bool		failover_done = false;
 
-char	   *pid_file = NULL;
+/*
+ * when `failover=manual`, and the upstream server has gone away,
+ * this flag is set to indicate we should connect to whatever the
+ * current master is to update monitoring information
+ */
+bool		manual_mode_upstream_disconnected = false;
 
-t_configuration_options config = T_CONFIGURATION_OPTIONS_INITIALIZER;
+char	   *pid_file = NULL;
 
 static void help(void);
 static void usage(void);
@@ -126,7 +134,7 @@ main(int argc, char **argv)
 		{"monitoring-history", no_argument, NULL, 'm'},
 		{"daemonize", no_argument, NULL, 'd'},
 		{"pid-file", required_argument, NULL, 'p'},
-		{"help", no_argument, NULL, '?'},
+		{"help", no_argument, NULL, OPT_HELP},
 		{"version", no_argument, NULL, 'V'},
 		{NULL, 0, NULL, 0}
 	};
@@ -160,6 +168,23 @@ main(int argc, char **argv)
 	{
 		switch (c)
 		{
+			case '?':
+				/* Actual help option given */
+				if (strcmp(argv[optind - 1], "-?") == 0)
+				{
+					help();
+					exit(SUCCESS);
+				}
+				/* unknown option reported by getopt */
+				else
+					goto unknown_option;
+				break;
+			case OPT_HELP:
+				help();
+				exit(SUCCESS);
+			case 'V':
+				printf("%s %s (PostgreSQL %s)\n", progname(), REPMGR_VERSION, PG_VERSION);
+				exit(SUCCESS);
 			case 'f':
 				config_file = optarg;
 				break;
@@ -175,13 +200,9 @@ main(int argc, char **argv)
 			case 'p':
 				pid_file = optarg;
 				break;
-			case '?':
-				help();
-				exit(SUCCESS);
-			case 'V':
-				printf("%s %s (PostgreSQL %s)\n", progname(), REPMGR_VERSION, PG_VERSION);
-				exit(SUCCESS);
+
 			default:
+		unknown_option:
 				usage();
 				exit(ERR_BAD_CONFIG);
 		}
@@ -399,7 +420,7 @@ main(int argc, char **argv)
 			case STANDBY:
 
 				/* We need the node id of the master server as well as a connection to it */
-				log_info(_("connecting to master node '%s'\n"),
+				log_info(_("connecting to master node of cluster '%s'\n"),
 						 local_options.cluster_name);
 
 				master_conn = get_master_connection(my_local_conn,
@@ -435,6 +456,7 @@ main(int argc, char **argv)
 					my_local_conn = establish_db_connection(local_options.conninfo, true);
 					update_registration();
 				}
+
 				/* Log startup event */
 				if (startup_event_logged == false)
 				{
@@ -462,16 +484,16 @@ main(int argc, char **argv)
 
 				do
 				{
-					log_verbose(LOG_DEBUG, "standby check loop...\n");
-
-					if (node_info.type == WITNESS)
-					{
-						witness_monitor();
-					}
-					else if (node_info.type == STANDBY)
+					if (node_info.type == STANDBY)
 					{
+						log_verbose(LOG_DEBUG, "standby check loop...\n");
 						standby_monitor();
 					}
+					else if (node_info.type == WITNESS)
+					{
+						log_verbose(LOG_DEBUG, "witness check loop...\n");
+						witness_monitor();
+					}
 
 					sleep(local_options.monitor_interval_secs);
 
@@ -641,7 +663,7 @@ witness_monitor(void)
 								 local_options.master_response_timeout) != 1)
 		return;
 
-	/* Get local xlog info */
+	/* Get timestamp for monitoring update */
 	sqlquery_snprintf(sqlquery, "SELECT CURRENT_TIMESTAMP");
 
 	res = PQexec(my_local_conn, sqlquery);
@@ -667,7 +689,7 @@ witness_monitor(void)
 					  "            replication_lag, apply_lag )"
 					  "      VALUES(%d, %d, "
 					  "             '%s'::TIMESTAMP WITH TIME ZONE, NULL, "
-					  "             pg_current_xlog_location(), NULL, "
+					  "             pg_catalog.pg_current_xlog_location(), NULL, "
 					  "             0, 0) ",
 					  get_repmgr_schema_quoted(my_local_conn),
 					  master_options.node,
@@ -695,7 +717,7 @@ standby_monitor(void)
 {
 	PGresult   *res;
 	char		monitor_standby_timestamp[MAXLEN];
-	char		last_wal_master_location[MAXLEN];
+	char		last_wal_primary_location[MAXLEN];
 	char		last_xlog_receive_location[MAXLEN];
 	char		last_xlog_replay_location[MAXLEN];
 	char		last_xact_replay_timestamp[MAXLEN];
@@ -706,6 +728,9 @@ standby_monitor(void)
 	XLogRecPtr	lsn_last_xlog_receive_location;
 	XLogRecPtr	lsn_last_xlog_replay_location;
 
+	long long unsigned int replication_lag;
+	long long unsigned int apply_lag;
+
 	int			connection_retries,
 				ret;
 	bool		did_retry = false;
@@ -719,6 +744,8 @@ standby_monitor(void)
 	const char *upstream_node_type = NULL;
 
 	bool		receiving_streamed_wal = true;
+
+
 	/*
 	 * Verify that the local node is still available - if not there's
 	 * no point in doing much else anyway
@@ -740,20 +767,36 @@ standby_monitor(void)
 		goto continue_monitoring_standby;
 	}
 
-	upstream_conn = get_upstream_connection(my_local_conn,
-											local_options.cluster_name,
-											local_options.node,
-											&upstream_node_id,
-											upstream_conninfo);
-
-	upstream_node_type = (upstream_node_id == master_options.node)
-		? "master"
-		: "upstream";
-
-	// ZZZ "5 minutes"?
 	/*
-	 * Check if the upstream node is still available, if after 5 minutes of retries
-	 * we cannot reconnect, try to get a new upstream node.
+	 * Standby has `failover` set to manual and is disconnected from
+	 * replication following a prior upstream node failure - we'll
+	 * find the master to be able to write monitoring information, if
+	 * required
+	 */
+	if (manual_mode_upstream_disconnected == true)
+	{
+		upstream_conn = get_master_connection(my_local_conn,
+												local_options.cluster_name,
+												&upstream_node_id,
+												upstream_conninfo);
+		upstream_node_type = "master";
+	}
+	else
+	{
+		upstream_conn = get_upstream_connection(my_local_conn,
+												local_options.cluster_name,
+												local_options.node,
+												&upstream_node_id,
+												upstream_conninfo);
+
+		upstream_node_type = (upstream_node_id == master_options.node)
+			? "master"
+			: "upstream";
+	}
+
+	/*
+	 * Check that the upstream node is still available
+	 * If not, initiate failover process
 	 */
 
 	check_connection(&upstream_conn, upstream_node_type, upstream_conninfo);
@@ -764,29 +807,52 @@ standby_monitor(void)
 
 	if (PQstatus(upstream_conn) != CONNECTION_OK)
 	{
+		int previous_master_node_id = master_options.node;
+
 		PQfinish(upstream_conn);
 		upstream_conn = NULL;
 
+		/*
+		 * When `failover=manual`, no actual failover will be performed, instead
+		 * the following happens:
+		 *  - find the new master
+		 *  - create an event notification `standby_disconnect_manual`
+		 *  - set a flag to indicate we're disconnected from replication,
+		 */
 		if (local_options.failover == MANUAL_FAILOVER)
 		{
 			log_err(_("Unable to reconnect to %s. Now checking if another node has been promoted.\n"), upstream_node_type);
 
+			/*
+			 * Set the location string in shared memory to indicate to other
+			 * repmgrd instances that we're *not* a promotion candidate and
+			 * that other repmgrd instance should not expect location updates
+			 * from us
+			 */
+
+			update_shared_memory(PASSIVE_NODE);
+
 			for (connection_retries = 0; connection_retries < local_options.reconnect_attempts; connection_retries++)
 			{
 				master_conn = get_master_connection(my_local_conn,
 					local_options.cluster_name, &master_options.node, NULL);
+
 				if (PQstatus(master_conn) == CONNECTION_OK)
 				{
 					/*
 					 * Connected, we can continue the process so break the
 					 * loop
 					 */
-					log_err(_("connected to node %d, continuing monitoring.\n"),
+					log_notice(_("connected to node %d, continuing monitoring.\n"),
 							master_options.node);
 					break;
 				}
 				else
 				{
+					/*
+					 * XXX this is the only place where `retry_promote_interval_secs`
+					 * is used - this parameter should be renamed or possibly be replaced
+					 */
 					log_err(
 					    _("no new master found, waiting %i seconds before retry...\n"),
 					    local_options.retry_promote_interval_secs
@@ -816,30 +882,58 @@ standby_monitor(void)
 
 				terminate(ERR_DB_CON);
 			}
+
+			/*
+			 * connected to a master - is it the same as the former upstream?
+			 * if not:
+			 *  - create event standby_disconnect
+			 *  - set global "disconnected_manual_standby"
+			 */
+
+			if (previous_master_node_id != master_options.node)
+			{
+				PQExpBufferData errmsg;
+				initPQExpBuffer(&errmsg);
+
+				appendPQExpBuffer(&errmsg,
+								  _("node %i is in manual failover mode and is now disconnected from replication"),
+								  local_options.node);
+
+				log_verbose(LOG_DEBUG, "old master: %i; current: %i\n", previous_master_node_id, master_options.node);
+
+				manual_mode_upstream_disconnected = true;
+
+				create_event_record(master_conn,
+									&local_options,
+									local_options.node,
+									"standby_disconnect_manual",
+									/* here "true" indicates the action has occurred as expected */
+									true,
+									errmsg.data);
+
+			}
 		}
 		else if (local_options.failover == AUTOMATIC_FAILOVER)
 		{
 			/*
-			 * When we returns from this function we will have a new master
+			 * When we return from this function we will have a new master
 			 * and a new master_conn
-			 */
-
-			/*
+			 *
 			 * Failover handling is handled differently depending on whether
 			 * the failed node is the master or a cascading standby
 			 */
 			upstream_node = get_node_info(my_local_conn, local_options.cluster_name, upstream_node_id);
 
-            if (upstream_node.type == MASTER)
-            {
-                log_debug(_("failure detected on master node (%i); attempting to promote a standby\n"),
-                          node_info.upstream_node_id);
-                do_master_failover();
-            }
-            else
-            {
-                log_debug(_("failure detected on upstream node %i; attempting to reconnect to new upstream node\n"),
-                          node_info.upstream_node_id);
+			if (upstream_node.type == MASTER)
+			{
+				log_debug(_("failure detected on master node (%i); attempting to promote a standby\n"),
+						  node_info.upstream_node_id);
+				do_master_failover();
+			}
+			else
+			{
+				log_debug(_("failure detected on upstream node %i; attempting to reconnect to new upstream node\n"),
+						  node_info.upstream_node_id);
 
 				if (!do_upstream_standby_failover(upstream_node))
 				{
@@ -847,20 +941,20 @@ standby_monitor(void)
 					initPQExpBuffer(&errmsg);
 
 					appendPQExpBuffer(&errmsg,
-									  _("unable to reconnect to new upstream node, terminating..."));
+							  _("unable to reconnect to new upstream node, terminating..."));
 
 					log_err("%s\n", errmsg.data);
 
 					create_event_record(master_conn,
-										&local_options,
-										local_options.node,
-										"repmgrd_shutdown",
-										false,
-										errmsg.data);
+							    &local_options,
+							    local_options.node,
+							    "repmgrd_shutdown",
+							    false,
+							    errmsg.data);
 
 					terminate(ERR_DB_CON);
 				}
-            }
+			}
 			return;
 		}
 	}
@@ -918,8 +1012,8 @@ standby_monitor(void)
 		 * the stream. If we set the local standby node as failed and it's now running
 		 * and receiving replication data, we should activate it again.
 		 */
-	        set_local_node_status();
-	        log_info(_("standby connection recovered!\n"));
+		set_local_node_status();
+		log_info(_("standby connection recovered!\n"));
 	}
 
 	/* Fast path for the case where no history is requested */
@@ -931,6 +1025,7 @@ standby_monitor(void)
 	 * from the upstream node to write monitoring information
 	 */
 
+	/* XXX not used? */
 	upstream_node = get_node_info(my_local_conn, local_options.cluster_name, upstream_node_id);
 
 	sprintf(sqlquery,
@@ -963,7 +1058,7 @@ standby_monitor(void)
 
 	if (active_master_id != master_options.node)
 	{
-		log_notice(_("connecting to active master (node %i)...\n"), active_master_id); \
+		log_notice(_("connecting to active master (node %i)...\n"), active_master_id);
 		if (master_conn != NULL)
 		{
 			PQfinish(master_conn);
@@ -985,10 +1080,19 @@ standby_monitor(void)
 		return;
 
 	/* Get local xlog info */
+
 	sqlquery_snprintf(sqlquery,
-					  "SELECT CURRENT_TIMESTAMP, pg_last_xlog_receive_location(), "
-					  "pg_last_xlog_replay_location(), pg_last_xact_replay_timestamp(), "
-					  "pg_last_xlog_receive_location() >= pg_last_xlog_replay_location()");
+					  " SELECT ts, "
+					  "        receive_location, "
+					  "        replay_location, "
+					  "        replay_timestamp, "
+					  "        receive_location >= replay_location "
+					  "   FROM (SELECT CURRENT_TIMESTAMP AS ts, "
+					  "         pg_catalog.pg_last_xlog_receive_location() AS receive_location, "
+					  "         pg_catalog.pg_last_xlog_replay_location()  AS replay_location, "
+					  "         pg_catalog.pg_last_xact_replay_timestamp() AS replay_timestamp "
+					  "        ) q ");
+
 
 	res = PQexec(my_local_conn, sqlquery);
 	if (PQresultStatus(res) != PGRES_TUPLES_OK)
@@ -1038,7 +1142,12 @@ standby_monitor(void)
 					"Replayed WAL newer than received WAL - is this standby connected to its upstream?\n");
 	}
 
-	/* Get master xlog info */
+	/*
+	 * Get master xlog position
+	 *
+	 * TODO: investigate whether pg_current_xlog_insert_location() would be a better
+	 * choice; see: https://github.com/2ndQuadrant/repmgr/issues/189
+	 */
 	sqlquery_snprintf(sqlquery, "SELECT pg_catalog.pg_current_xlog_location()");
 
 	res = PQexec(master_conn, sqlquery);
@@ -1049,21 +1158,43 @@ standby_monitor(void)
 		return;
 	}
 
-	strncpy(last_wal_master_location, PQgetvalue(res, 0, 0), MAXLEN);
+	strncpy(last_wal_primary_location, PQgetvalue(res, 0, 0), MAXLEN);
 	PQclear(res);
 
-	/* Calculate the lag */
-	lsn_master_current_xlog_location = lsn_to_xlogrecptr(last_wal_master_location, NULL);
-
+	lsn_master_current_xlog_location = lsn_to_xlogrecptr(last_wal_primary_location, NULL);
 	lsn_last_xlog_replay_location = lsn_to_xlogrecptr(last_xlog_replay_location, NULL);
 
+	/* Calculate apply lag */
 	if (last_xlog_receive_location_gte_replayed == false)
 	{
-		lsn_last_xlog_receive_location = lsn_last_xlog_replay_location;
+		/*
+		 * We're not receiving streaming WAL - in this case the receive location
+		 * equals the last replayed location
+		 */
+		apply_lag = 0;
+		strncpy(last_xlog_receive_location, last_xlog_replay_location, MAXLEN);
+		lsn_last_xlog_receive_location = lsn_to_xlogrecptr(last_xlog_replay_location, NULL);
 	}
 	else
 	{
 		lsn_last_xlog_receive_location = lsn_to_xlogrecptr(last_xlog_receive_location, NULL);
+
+		apply_lag = (long long unsigned int)lsn_last_xlog_receive_location - lsn_last_xlog_replay_location;
+	}
+
+
+	/* Calculate replication lag */
+	if (lsn_master_current_xlog_location >= lsn_last_xlog_receive_location)
+	{
+		replication_lag = (long long unsigned int)(lsn_master_current_xlog_location - lsn_last_xlog_receive_location);
+	}
+	else
+	{
+		/* This should never happen, but in case it does set lag to zero */
+		log_warning("Master xlog (%s) location appears less than standby receive location (%s)\n",
+					last_wal_primary_location,
+					last_xlog_receive_location);
+		replication_lag = 0;
 	}
 
 	/*
@@ -1092,11 +1223,10 @@ standby_monitor(void)
 					  local_options.node,
 					  monitor_standby_timestamp,
 					  last_xact_replay_timestamp,
-					  last_wal_master_location,
+					  last_wal_primary_location,
 					  last_xlog_receive_location,
-					  (long long unsigned int)(lsn_master_current_xlog_location - lsn_last_xlog_receive_location),
-					  (long long unsigned int)(lsn_last_xlog_receive_location - lsn_last_xlog_replay_location));
-
+					  replication_lag,
+					  apply_lag);
 	/*
 	 * Execute the query asynchronously, but don't check for a result. We will
 	 * check the result next time we pause for a monitor step.
@@ -1133,8 +1263,6 @@ do_master_failover(void)
 	XLogRecPtr	xlog_recptr;
 	bool		lsn_format_ok;
 
-	char		last_xlog_replay_location[MAXLEN];
-
 	PGconn	   *node_conn = NULL;
 
 	/*
@@ -1143,8 +1271,8 @@ do_master_failover(void)
 	 */
 	t_node_info nodes[FAILOVER_NODES_MAX_CHECK];
 
-    /* Store details of the failed node here */
-    t_node_info failed_master = T_NODE_INFO_INITIALIZER;
+	/* Store details of the failed node here */
+	t_node_info failed_master = T_NODE_INFO_INITIALIZER;
 
 	/* Store details of the best candidate for promotion to master here */
 	t_node_info best_candidate = T_NODE_INFO_INITIALIZER;
@@ -1154,7 +1282,7 @@ do_master_failover(void)
 			"SELECT id, conninfo, type, upstream_node_id "
 			"  FROM %s.repl_nodes "
 			" WHERE cluster = '%s' "
-            "   AND active IS TRUE "
+		        "   AND active IS TRUE "
 			"   AND priority > 0 "
 			" ORDER BY priority DESC, id "
 			" LIMIT %i ",
@@ -1167,7 +1295,6 @@ do_master_failover(void)
 	{
 		log_err(_("unable to retrieve node records: %s\n"), PQerrorMessage(my_local_conn));
 		PQclear(res);
-		PQfinish(my_local_conn);
 		terminate(ERR_DB_QUERY);
 	}
 
@@ -1316,8 +1443,8 @@ do_master_failover(void)
 				  " considered as new master and exit.\n"),
 				PQerrorMessage(my_local_conn));
 		PQclear(res);
-		sprintf(last_xlog_replay_location, "'%X/%X'", 0, 0);
-		update_shared_memory(last_xlog_replay_location);
+
+		update_shared_memory(LSN_QUERY_ERROR);
 		terminate(ERR_DB_QUERY);
 	}
 	/* write last location in shared memory */
@@ -1367,6 +1494,7 @@ do_master_failover(void)
 
 		while (!nodes[i].is_ready)
 		{
+			char location_value[MAXLEN];
 
 			sqlquery_snprintf(sqlquery,
 							  "SELECT %s.repmgr_get_last_standby_location()",
@@ -1382,7 +1510,11 @@ do_master_failover(void)
 				terminate(ERR_DB_QUERY);
 			}
 
-			xlog_recptr = lsn_to_xlogrecptr(PQgetvalue(res, 0, 0), &lsn_format_ok);
+			/* Copy the returned value as we'll need to reference it a few times */
+			strncpy(location_value, PQgetvalue(res, 0, 0), MAXLEN);
+			PQclear(res);
+
+			xlog_recptr = lsn_to_xlogrecptr(location_value, &lsn_format_ok);
 
 			/* If position reported as "invalid", check for format error or
 			 * empty string; otherwise position is 0/0 and we need to continue
@@ -1390,10 +1522,36 @@ do_master_failover(void)
 			 */
 			if (xlog_recptr == InvalidXLogRecPtr)
 			{
+				bool continue_loop = true;
+
 				if (lsn_format_ok == false)
 				{
+
+					/*
+					 * The node is indicating it is not a promotion candidate -
+					 * in this case we can store its invalid LSN to ensure it
+					 * can't be a promotion candidate when comparing locations
+					 */
+					if (strcmp(location_value, PASSIVE_NODE) == 0)
+					{
+						log_debug("node %i is passive mode\n", nodes[i].node_id);
+						log_info(_("node %i will not be considered for promotion\n"), nodes[i].node_id);
+						nodes[i].xlog_location = InvalidXLogRecPtr;
+						continue_loop = false;
+					}
+					/*
+					 * This should probably never happen but if it does, rule the
+					 * node out as a promotion candidate
+					 */
+					else if (strcmp(location_value, LSN_QUERY_ERROR) == 0)
+					{
+						log_warning(_("node %i is unable to update its shared memory and will not be considered for promotion\n"), nodes[i].node_id);
+						nodes[i].xlog_location = InvalidXLogRecPtr;
+						continue_loop = false;
+					}
+
 					/* Unable to parse value returned by `repmgr_get_last_standby_location()` */
-					if (*PQgetvalue(res, 0, 0) == '\0')
+					else if (*location_value == '\0')
 					{
 						log_crit(
 							_("unable to obtain LSN from node %i"), nodes[i].node_id
@@ -1402,8 +1560,8 @@ do_master_failover(void)
 							_("please check that 'shared_preload_libraries=repmgr_funcs' is set in postgresql.conf\n")
 							);
 
-						PQclear(res);
 						PQfinish(node_conn);
+						/* XXX shouldn't we just ignore this node? */
 						exit(ERR_BAD_CONFIG);
 					}
 
@@ -1411,25 +1569,29 @@ do_master_failover(void)
 					 * Very unlikely to happen; in the absence of any better
 					 * strategy keep checking
 					 */
-					log_warning(_("unable to parse LSN \"%s\"\n"),
-								PQgetvalue(res, 0, 0));
+					else {
+						log_warning(_("unable to parse LSN \"%s\"\n"),
+									location_value);
+					}
 				}
 				else
 				{
 					log_debug(
 						_("invalid LSN returned from node %i: '%s'\n"),
 						nodes[i].node_id,
-						PQgetvalue(res, 0, 0)
-						);
+						location_value);
 				}
 
-				PQclear(res);
-
-				/* If position is 0/0, keep checking */
-				/* XXX we should add a timeout here to prevent infinite looping
+				/*
+				 * If the node is still reporting an InvalidXLogRecPtr, it means
+				 * its repmgrd hasn't yet had time to update it (either with a valid
+				 * XLogRecPtr or a message) so we continue looping.
+				 *
+				 * XXX we should add a timeout here to prevent infinite looping
 				 * if the other node's repmgrd is not up
 				 */
-				continue;
+				if (continue_loop == true)
+					continue;
 			}
 
 			if (nodes[i].xlog_location < xlog_recptr)
@@ -1437,8 +1599,7 @@ do_master_failover(void)
 				nodes[i].xlog_location = xlog_recptr;
 			}
 
-			log_debug(_("LSN of node %i is: %s\n"), nodes[i].node_id, PQgetvalue(res, 0, 0));
-			PQclear(res);
+			log_debug(_("LSN of node %i is: %s\n"), nodes[i].node_id, location_value);
 
 			ready_nodes++;
 			nodes[i].is_ready = true;
@@ -1541,12 +1702,12 @@ do_master_failover(void)
 					log_notice(_("Original master reappeared before this standby was promoted - no action taken\n"));
 
 					PQfinish(master_conn);
+					master_conn = NULL;
+
 					/* no failover occurred but we'll want to restart connections */
 					failover_done = true;
 					return;
 				}
-
-				PQfinish(my_local_conn);
 			}
 
 			log_err(_("promote command failed. You could check and try it manually.\n"));
@@ -1901,7 +2062,7 @@ check_connection(PGconn **conn, const char *type, const char *conninfo)
 static bool
 set_local_node_status(void)
 {
-        PGresult       *res;
+	PGresult       *res;
 	char		sqlquery[QUERY_STR_LEN];
 	int		active_master_node_id = NODE_NOT_FOUND;
 	char		master_conninfo[MAXLEN];
@@ -1994,10 +2155,12 @@ check_cluster_configuration(PGconn *conn)
 	log_info(_("checking cluster configuration with schema '%s'\n"), get_repmgr_schema());
 
 	sqlquery_snprintf(sqlquery,
-					  "SELECT oid FROM pg_class "
+					  "SELECT oid FROM pg_catalog.pg_class "
 					  " WHERE oid = '%s.repl_nodes'::regclass ",
-			                  get_repmgr_schema_quoted(master_conn));
+					  get_repmgr_schema_quoted(master_conn));
+
 	res = PQexec(conn, sqlquery);
+
 	if (PQresultStatus(res) != PGRES_TUPLES_OK)
 	{
 		log_err(_("PQexec failed: %s\n"), PQerrorMessage(conn));
@@ -2112,7 +2275,7 @@ lsn_to_xlogrecptr(char *lsn, bool *format_ok)
 	{
 		if (format_ok != NULL)
 			*format_ok = false;
-		log_err(_("incorrect log location format: %s\n"), lsn);
+		log_warning(_("incorrect log location format: %s\n"), lsn);
 		return 0;
 	}
 
@@ -2416,6 +2579,8 @@ get_node_info(PGconn *conn, char *cluster, int node_id)
 							errmsg.data);
 
 		PQfinish(conn);
+		conn = NULL;
+
 		terminate(ERR_DB_QUERY);
 	}
 

sql/repmgr2_repmgr3.sql

@@ -63,6 +63,15 @@ UPDATE repl_nodes SET type = 'master' WHERE id = $master_id;
 
 -- UPDATE repl_nodes SET active = FALSE WHERE id IN (...);
 
+/* There's also an event table which we need to create */
+CREATE TABLE repl_events (
+  node_id          INTEGER NOT NULL,
+  event            TEXT NOT NULL,
+  successful       BOOLEAN NOT NULL DEFAULT TRUE,
+  event_timestamp  TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT CURRENT_TIMESTAMP,
+  details          TEXT NULL
+);
+
 /* When you're sure of your changes, commit them */
 
 -- COMMIT;

sql/repmgr3.1.1_repmgr3.1.2.sql

@@ -27,5 +27,6 @@
 
 BEGIN;
 
-ALTER TABLE repl_nodes ALTER CONSTRAINT repl_nodes_upstream_node_id_fkey DEFERRABLE;
+ALTER TABLE repl_nodes DROP CONSTRAINT repl_nodes_upstream_node_id_fkey,
+      ADD CONSTRAINT repl_nodes_upstream_node_id_fkey FOREIGN KEY (upstream_node_id) REFERENCES repl_nodes(id) DEFERRABLE;
 COMMIT;

version.h

@@ -1,6 +1,6 @@
 #ifndef _VERSION_H_
 #define _VERSION_H_
 
-#define REPMGR_VERSION "3.1.3"
+#define REPMGR_VERSION "3.1.5"
 
 #endif