<div dir="ltr">Hi,<div><br></div><div>I have an environment with 4 pgpool agents managing 5 postgres DBs. All 4 pgpools are not installed on the DBs but on other machines (web servers). After simulating disconnection to 2 of the 5 DBs, pgpool doesn&#39;t detach the DB nodes even after <b>health_check_max_retries </b>is reached.</div><div>please notice that <b>fail_over_on_backend_error</b> is off and <b>backend_flag </b>is set to &#39;ALLOW_TO_FAILOVER&#39; on all DB nodes.</div><div><br></div><div>Can you please advise what could be the reason for that?</div><div><br></div><div><br></div><div>Thanks,</div><div>Nir</div><div><br></div><div><br></div><div><br></div><div>attached the pgpool.conf file:</div><div><br></div><div><div># ----------------------------</div><div># pgPool-II configuration file</div><div># ----------------------------</div><div>#</div><div># This file consists of lines of the form:</div><div>#</div><div>#   name = value</div><div>#</div><div># Whitespace may be used.  Comments are introduced with &quot;#&quot; anywhere on a line.</div><div># The complete list of parameter names and allowed values can be found in the</div><div># pgPool-II documentation.</div><div>#</div><div># This file is read on server startup and when the server receives a SIGHUP</div><div># signal.  If you edit the file on a running system, you have to SIGHUP the</div><div># server for the changes to take effect, or use &quot;pgpool reload&quot;.  Some</div><div># parameters, which are marked below, require a server shutdown and restart to</div><div># take effect.</div><div>#</div><div><br></div><div><br></div><div>#------------------------------------------------------------------------------</div><div># CONNECTIONS</div><div>#------------------------------------------------------------------------------</div><div><br></div><div># - pgpool Connection Settings -</div><div><br></div><div>listen_addresses = &#39;*&#39;</div><div>                                   # Host name or IP address to listen on:</div><div>                                   # &#39;*&#39; for all, &#39;&#39; for no TCP/IP connections</div><div>                                   # (change requires restart)</div><div>port = 9999</div><div>                                   # Port number</div><div>                                   # (change requires restart)</div><div>socket_dir = &#39;/tmp&#39;</div><div>                                   # Unix domain socket path</div><div>                                   # The Debian package defaults to</div><div>                                   # /var/run/postgresql</div><div>                                   # (change requires restart)</div><div>listen_backlog_multiplier = 2</div><div>                                   # Set the backlog parameter of listen(2) to</div><div><span class="" style="white-space:pre">                                                                </span>   # num_init_children * listen_backlog_multiplier.</div><div>                                   # (change requires restart)</div><div>serialize_accept = off</div><div>                                   # whether to serialize accept() call to avoid thundering herd problem</div><div>                                   # (change requires restart)</div><div><br></div><div># - pgpool Communication Manager Connection Settings -</div><div><br></div><div>pcp_listen_addresses = &#39;*&#39;</div><div>                                   # Host name or IP address for pcp process to listen on:</div><div>                                   # &#39;*&#39; for all, &#39;&#39; for no TCP/IP connections</div><div>                                   # (change requires restart)</div><div>pcp_port = 9898</div><div>                                   # Port number for pcp</div><div>                                   # (change requires restart)</div><div>pcp_socket_dir = &#39;/tmp&#39;</div><div>                                   # Unix domain socket path for pcp</div><div>                                   # The Debian package defaults to</div><div>                                   # /var/run/postgresql</div><div>                                   # (change requires restart)</div><div><br></div><div># - Backend Connection Settings -</div><div><br></div><div>backend_hostname0 = &#39;199.203.248.249&#39;</div><div>                                   # Host name or IP address to connect to for backend 0</div><div>backend_port0 = 5432</div><div>                                   # Port number for backend 0</div><div>backend_weight0 = 0</div><div>                                   # Weight for backend 0 (only in load balancing mode)</div><div>backend_data_directory0 = &#39;/var/lib/pgsql/9.5/data&#39;</div><div>                                   # Data directory for backend 0</div><div>backend_flag0 = &#39;ALLOW_TO_FAILOVER&#39;</div><div>                                   # Controls various backend behavior</div><div>                                   # ALLOW_TO_FAILOVER or DISALLOW_TO_FAILOVER</div><div>backend_hostname1 = &#39;199.203.248.219&#39;</div><div>backend_port1 = 5432</div><div>backend_weight1 = 1</div><div>backend_data_directory1 = &#39;/var/lib/pgsql/9.5/data&#39;</div><div>backend_flag1 = &#39;ALLOW_TO_FAILOVER&#39;</div><div>backend_flag2 = &#39;ALLOW_TO_FAILOVER&#39;</div><div>backend_flag3 = &#39;ALLOW_TO_FAILOVER&#39;</div><div>backend_flag4 = &#39;ALLOW_TO_FAILOVER&#39;</div><div><br></div><div><br></div><div><br></div><div># - Authentication -</div><div><br></div><div>enable_pool_hba = off</div><div>                                   # Use pool_hba.conf for client authentication</div><div>pool_passwd = &#39;pool_passwd&#39;</div><div>                                   # File name of pool_passwd for md5 authentication.</div><div>                                   # &quot;&quot; disables pool_passwd.</div><div>                                   # (change requires restart)</div><div>authentication_timeout = 60</div><div>                                   # Delay in seconds to complete client authentication</div><div>                                   # 0 means no timeout.</div><div><br></div><div># - SSL Connections -</div><div><br></div><div>ssl = off</div><div>                                   # Enable SSL support</div><div>                                   # (change requires restart)</div><div>#ssl_key = &#39;./server.key&#39;</div><div>                                   # Path to the SSL private key file</div><div>                                   # (change requires restart)</div><div>#ssl_cert = &#39;./server.cert&#39;</div><div>                                   # Path to the SSL public certificate file</div><div>                                   # (change requires restart)</div><div>#ssl_ca_cert = &#39;&#39;</div><div>                                   # Path to a single PEM format file</div><div>                                   # containing CA root certificate(s)</div><div>                                   # (change requires restart)</div><div>#ssl_ca_cert_dir = &#39;&#39;</div><div>                                   # Directory containing CA root certificate(s)</div><div>                                   # (change requires restart)</div><div><br></div><div><br></div><div>#------------------------------------------------------------------------------</div><div># POOLS</div><div>#------------------------------------------------------------------------------</div><div><br></div><div># - Pool size -</div><div><br></div><div>num_init_children = 32</div><div>                                   # Number of pools</div><div>                                   # (change requires restart)</div><div>max_pool = 4</div><div>                                   # Number of connections per pool</div><div>                                   # (change requires restart)</div><div><br></div><div># - Life time -</div><div><br></div><div>child_life_time = 300</div><div>                                   # Pool exits after being idle for this many seconds</div><div>child_max_connections = 0</div><div>                                   # Pool exits after receiving that many connections</div><div>                                   # 0 means no exit</div><div>connection_life_time = 0</div><div>                                   # Connection to backend closes after being idle for this many seconds</div><div>                                   # 0 means no close</div><div>client_idle_limit = 0</div><div>                                   # Client is disconnected after being idle for that many seconds</div><div>                                   # (even inside an explicit transactions!)</div><div>                                   # 0 means no disconnection</div><div><br></div><div><br></div><div>#------------------------------------------------------------------------------</div><div># LOGS</div><div>#------------------------------------------------------------------------------</div><div><br></div><div># - Where to log -</div><div><br></div><div>log_destination = &#39;stderr&#39;</div><div>                                   # Where to log</div><div>                                   # Valid values are combinations of stderr,</div><div>                                   # and syslog. Default to stderr.</div><div><br></div><div># - What to log -</div><div><br></div><div>log_line_prefix = &#39;%t: pid %p: &#39;   # printf-style string to output at beginning of each log line.</div><div><br></div><div>log_connections = on</div><div>                                   # Log connections</div><div>log_hostname = on</div><div>                                   # Hostname will be shown in ps status</div><div>                                   # and in logs if connections are logged</div><div>log_statement = on</div><div>                                   # Log all statements</div><div>log_per_node_statement = on</div><div>                                   # Log all statements</div><div>                                   # with node and backend informations</div><div>log_standby_delay = &#39;none&#39;</div><div>                                   # Log standby delay</div><div>                                   # Valid values are combinations of always,</div><div>                                   # if_over_threshold, none</div><div><br></div><div># - Syslog specific -</div><div><br></div><div>syslog_facility = &#39;LOCAL0&#39;</div><div>                                   # Syslog local facility. Default to LOCAL0</div><div>syslog_ident = &#39;pgpool&#39;</div><div>                                   # Syslog program identification string</div><div>                                   # Default to &#39;pgpool&#39;</div><div><br></div><div># - Debug -</div><div><br></div><div>debug_level = 99</div><div>                                   # Debug message verbosity level</div><div>                                   # 0 means no message, 1 or more mean verbose</div><div><br></div><div>log_error_verbosity = verbose# terse, default, or verbose messages</div><div><br></div><div>client_min_messages = warning# values in order of decreasing detail:</div><div>                                        #   debug5</div><div>                                        #   debug4</div><div>                                        #   debug3</div><div>                                        #   debug2</div><div>                                        #   debug1</div><div>                                        #   log</div><div>                                        #   notice</div><div>                                        #   warning</div><div>                                        #   error</div><div><br></div><div>log_min_messages = debug5# values in order of decreasing detail:</div><div>                                        #   debug5</div><div>                                        #   debug4</div><div>                                        #   debug3</div><div>                                        #   debug2</div><div>                                        #   debug1</div><div>                                        #   info</div><div>                                        #   notice</div><div>                                        #   warning</div><div>                                        #   error</div><div>                                        #   log</div><div>                                        #   fatal</div><div>                                        #   panic</div><div><br></div><div>#------------------------------------------------------------------------------</div><div># FILE LOCATIONS</div><div>#------------------------------------------------------------------------------</div><div><br></div><div>pid_file_name = &#39;/var/run/pgpool/pgpool.pid&#39;</div><div>                                   # PID file name</div><div>                                   # (change requires restart)</div><div>logdir = &#39;/var/log/pgpool&#39;</div><div>                                   # Directory of pgPool status file</div><div>                                   # (change requires restart)</div><div><br></div><div><br></div><div>#------------------------------------------------------------------------------</div><div># CONNECTION POOLING</div><div>#------------------------------------------------------------------------------</div><div><br></div><div>connection_cache = off</div><div>                                   # Activate connection pools</div><div>                                   # (change requires restart)</div><div><br></div><div>                                   # Semicolon separated list of queries</div><div>                                   # to be issued at the end of a session</div><div>                                   # The default is for 8.3 and later</div><div>reset_query_list = &#39;ABORT; DISCARD ALL&#39;</div><div>                                   # The following one is for 8.2 and before</div><div>#reset_query_list = &#39;ABORT; RESET ALL; SET SESSION AUTHORIZATION DEFAULT&#39;</div><div><br></div><div><br></div><div>#------------------------------------------------------------------------------</div><div># REPLICATION MODE</div><div>#------------------------------------------------------------------------------</div><div><br></div><div>replication_mode = off</div><div>                                   # Activate replication mode</div><div>                                   # (change requires restart)</div><div>replicate_select = off</div><div>                                   # Replicate SELECT statements</div><div>                                   # when in replication mode</div><div>                                   # replicate_select is higher priority than</div><div>                                   # load_balance_mode.</div><div><br></div><div>insert_lock = on</div><div>                                   # Automatically locks a dummy row or a table</div><div>                                   # with INSERT statements to keep SERIAL data</div><div>                                   # consistency</div><div>                                   # Without SERIAL, no lock will be issued</div><div>lobj_lock_table = &#39;&#39;</div><div>                                   # When rewriting lo_creat command in</div><div>                                   # replication mode, specify table name to</div><div>                                   # lock</div><div><br></div><div># - Degenerate handling -</div><div><br></div><div>replication_stop_on_mismatch = off</div><div>                                   # On disagreement with the packet kind</div><div>                                   # sent from backend, degenerate the node</div><div>                                   # which is most likely &quot;minority&quot;</div><div>                                   # If off, just force to exit this session</div><div><br></div><div>failover_if_affected_tuples_mismatch = off</div><div>                                   # On disagreement with the number of affected</div><div>                                   # tuples in UPDATE/DELETE queries, then</div><div>                                   # degenerate the node which is most likely</div><div>                                   # &quot;minority&quot;.</div><div>                                   # If off, just abort the transaction to</div><div>                                   # keep the consistency</div><div><br></div><div><br></div><div>#------------------------------------------------------------------------------</div><div># LOAD BALANCING MODE</div><div>#------------------------------------------------------------------------------</div><div><br></div><div>load_balance_mode = on</div><div>                                   # Activate load balancing mode</div><div>                                   # (change requires restart)</div><div>ignore_leading_white_space = on</div><div>                                   # Ignore leading white spaces of each query</div><div>white_function_list = &#39;ro_func.*&#39;</div><div>                                   # Comma separated list of function names</div><div>                                   # that don&#39;t write to database</div><div>                                   # Regexp are accepted</div><div>black_function_list = &#39;nextval,setval,nextval,setval&#39;</div><div>                                   # Comma separated list of function names</div><div>                                   # that write to database</div><div>                                   # Regexp are accepted</div><div><br></div><div>database_redirect_preference_list = &#39;&#39;</div><div><span class="" style="white-space:pre">                                                                </span>   # comma separated list of pairs of database and node id.</div><div><span class="" style="white-space:pre">                                                                </span>   # example: postgres:primary,mydb[0-4]:1,mydb[5-9]:2&#39;</div><div><span class="" style="white-space:pre">                                                                </span>   # valid for streaming replicaton mode only.</div><div><br></div><div>app_name_redirect_preference_list = &#39;&#39;</div><div><span class="" style="white-space:pre">                                                                </span>   # comma separated list of pairs of app name and node id.</div><div><span class="" style="white-space:pre">                                                                </span>   # example: &#39;psql:primary,myapp[0-4]:1,myapp[5-9]:standby&#39;</div><div><span class="" style="white-space:pre">                                                                </span>   # valid for streaming replicaton mode only.</div><div>allow_sql_comments = off</div><div><span class="" style="white-space:pre">                                                                </span>   # if on, ignore SQL comments when judging if load balance or</div><div><span class="" style="white-space:pre">                                                                </span>   # query cache is possible.</div><div><span class="" style="white-space:pre">                                                                </span>   # If off, SQL comments effectively prevent the judgment</div><div><span class="" style="white-space:pre">                                                                </span>   # (pre 3.4 behavior).</div><div><br></div><div>#------------------------------------------------------------------------------</div><div># MASTER/SLAVE MODE</div><div>#------------------------------------------------------------------------------</div><div><br></div><div>master_slave_mode = on</div><div>                                   # Activate master/slave mode</div><div>                                   # (change requires restart)</div><div>master_slave_sub_mode = &#39;stream&#39;</div><div>                                   # Master/slave sub mode</div><div>                                   # Valid values are combinations slony or</div><div>                                   # stream. Default is slony.</div><div>                                   # (change requires restart)</div><div><br></div><div># - Streaming -</div><div><br></div><div>sr_check_period = 10</div><div>                                   # Streaming replication check period</div><div>                                   # Disabled (0) by default</div><div>sr_check_user = &#39;webcloud&#39;</div><div>                                   # Streaming replication check user</div><div>                                   # This is necessary even if you disable</div><div>                                   # streaming replication delay check with</div><div>                                   # sr_check_period = 0</div><div>sr_check_password = &#39;&#39;</div><div>                                   # Password for streaming replication check user</div><div>sr_check_database = &#39;postgres&#39;</div><div>                                   # Database name for streaming replication check</div><div>delay_threshold = 10000</div><div>                                   # Threshold before not dispatching query to standby node</div><div>                                   # Unit is in bytes</div><div>                                   # Disabled (0) by default</div><div><br></div><div># - Special commands -</div><div><br></div><div>follow_master_command = &#39;&#39;</div><div>                                   # Executes this command after master failover</div><div>                                   # Special values:</div><div>                                   #   %d = node id</div><div>                                   #   %h = host name</div><div>                                   #   %p = port number</div><div>                                   #   %D = database cluster path</div><div>                                   #   %m = new master node id</div><div>                                   #   %H = hostname of the new master node</div><div>                                   #   %M = old master node id</div><div>                                   #   %P = old primary node id</div><div>                                   #   %r = new master port number</div><div>                                   #   %R = new master database cluster path</div><div>                                   #   %% = &#39;%&#39; character</div><div><br></div><div>#------------------------------------------------------------------------------</div><div># HEALTH CHECK</div><div>#------------------------------------------------------------------------------</div><div><br></div><div>health_check_period = 10</div><div>                                   # Health check period</div><div>                                   # Disabled (0) by default</div><div>health_check_timeout = 90</div><div>                                   # Health check timeout</div><div>                                   # 0 means no timeout</div><div>health_check_user = &#39;webcloud&#39;</div><div>                                   # Health check user</div><div>health_check_password = &#39;&#39;</div><div>                                   # Password for health check user</div><div>health_check_database = &#39;postgres&#39;</div><div>                                   # Database name for health check. If &#39;&#39;, tries &#39;postgres&#39; frist, then &#39;template1&#39;</div><div><br></div><div>health_check_max_retries = 5</div><div>                                   # Maximum number of times to retry a failed health check before giving up.</div><div>health_check_retry_delay = 5</div><div>                                   # Amount of time to wait (in seconds) between retries.</div><div>connect_timeout = 10000</div><div>                                   # Timeout value in milliseconds before giving up to connect to backend.</div><div><span class="" style="white-space:pre">                                                                </span>   # Default is 10000 ms (10 second). Flaky network user may want to increase</div><div><span class="" style="white-space:pre">                                                                </span>   # the value. 0 means no timeout.</div><div><span class="" style="white-space:pre">                                                                </span>   # Note that this value is not only used for health check,</div><div><span class="" style="white-space:pre">                                                                </span>   # but also for ordinary conection to backend.</div><div><br></div><div>#------------------------------------------------------------------------------</div><div># FAILOVER AND FAILBACK</div><div>#------------------------------------------------------------------------------</div><div><br></div><div>failover_command = &#39;/etc/pgpool-II/failover.sh %d %h %p %D %m %H %M %P %r %R&#39;</div><div>                                   # Executes this command at failover</div><div>                                   # Special values:</div><div>                                   #   %d = node id</div><div>                                   #   %h = host name</div><div>                                   #   %p = port number</div><div>                                   #   %D = database cluster path</div><div>                                   #   %m = new master node id</div><div>                                   #   %H = hostname of the new master node</div><div>                                   #   %M = old master node id</div><div>                                   #   %P = old primary node id</div><div>                                   #   %r = new master port number</div><div>                                   #   %R = new master database cluster path</div><div>                                   #   %% = &#39;%&#39; character</div><div>failback_command = &#39;/etc/pgpool-II/failback.sh&#39;</div><div>                                   # Executes this command at failback.</div><div>                                   # Special values:</div><div>                                   #   %d = node id</div><div>                                   #   %h = host name</div><div>                                   #   %p = port number</div><div>                                   #   %D = database cluster path</div><div>                                   #   %m = new master node id</div><div>                                   #   %H = hostname of the new master node</div><div>                                   #   %M = old master node id</div><div>                                   #   %P = old primary node id</div><div>                                   #   %r = new master port number</div><div>                                   #   %R = new master database cluster path</div><div>                                   #   %% = &#39;%&#39; character</div><div><br></div><div>fail_over_on_backend_error = off</div><div>                                   # Initiates failover when reading/writing to the</div><div>                                   # backend communication socket fails</div><div>                                   # If set to off, pgpool will report an</div><div>                                   # error and disconnect the session.</div><div><br></div><div>search_primary_node_timeout = 10</div><div>                                   # Timeout in seconds to search for the</div><div>                                   # primary node when a failover occurs.</div><div>                                   # 0 means no timeout, keep searching</div><div>                                   # for a primary node forever.</div><div><br></div><div>#------------------------------------------------------------------------------</div><div># ONLINE RECOVERY</div><div>#------------------------------------------------------------------------------</div><div><br></div><div>recovery_user = &#39;nobody&#39;</div><div>                                   # Online recovery user</div><div>recovery_password = &#39;&#39;</div><div>                                   # Online recovery password</div><div>recovery_1st_stage_command = &#39;&#39;</div><div>                                   # Executes a command in first stage</div><div>recovery_2nd_stage_command = &#39;&#39;</div><div>                                   # Executes a command in second stage</div><div>recovery_timeout = 90</div><div>                                   # Timeout in seconds to wait for the</div><div>                                   # recovering node&#39;s postmaster to start up</div><div>                                   # 0 means no wait</div><div>client_idle_limit_in_recovery = 0</div><div>                                   # Client is disconnected after being idle</div><div>                                   # for that many seconds in the second stage</div><div>                                   # of online recovery</div><div>                                   # 0 means no disconnection</div><div>                                   # -1 means immediate disconnection</div><div><br></div><div><br></div><div>#------------------------------------------------------------------------------</div><div># WATCHDOG</div><div>#------------------------------------------------------------------------------</div><div><br></div><div># - Enabling -</div><div><br></div><div>use_watchdog = on</div><div>                                    # Activates watchdog</div><div>                                    # (change requires restart)</div><div><br></div><div># -Connection to up stream servers -</div><div><br></div><div>trusted_servers = &#39;&#39;</div><div>                                    # trusted server list which are used</div><div>                                    # to confirm network connection</div><div>                                    # (hostA,hostB,hostC,...)</div><div>                                    # (change requires restart)</div><div>ping_path = &#39;/bin&#39;</div><div>                                    # ping command path</div><div>                                    # (change requires restart)</div><div><br></div><div># - Watchdog communication Settings -</div><div><br></div><div>wd_hostname = &#39;199.203.248.219&#39;</div><div>                                    # Host name or IP address of this watchdog</div><div>                                    # (change requires restart)</div><div>wd_port = 9000</div><div>                                    # port number for watchdog service</div><div>                                    # (change requires restart)</div><div>wd_priority  = 1</div><div><span class="" style="white-space:pre">                                                                        </span># priority of this watchdog in leader election</div><div><span class="" style="white-space:pre">                                                                        </span># (change requires restart)</div><div><br></div><div>wd_authkey = &#39;&#39;</div><div>                                    # Authentication key for watchdog communication</div><div>                                    # (change requires restart)</div><div><br></div><div>wd_ipc_socket_dir = &#39;/tmp&#39;</div><div><span class="" style="white-space:pre">                                                                        </span># Unix domain socket path for watchdog IPC socket</div><div><span class="" style="white-space:pre">                                                                        </span># The Debian package defaults to</div><div><span class="" style="white-space:pre">                                                                        </span># /var/run/postgresql</div><div><span class="" style="white-space:pre">                                                                        </span># (change requires restart)</div><div><br></div><div><br></div><div># - Virtual IP control Setting -</div><div><br></div><div>delegate_IP = &#39;&#39;</div><div>                                    # delegate IP address</div><div>                                    # If this is empty, virtual IP never bring up.</div><div>                                    # (change requires restart)</div><div>if_cmd_path = &#39;/sbin&#39;</div><div>                                    # path to the directory where if_up/down_cmd exists </div><div>                                    # (change requires restart)</div><div>if_up_cmd = &#39;ip addr add $_IP_$/24 dev eth0 label eth0:0&#39;</div><div>                                    # startup delegate IP command</div><div>                                    # (change requires restart)</div><div>if_down_cmd = &#39;ip addr del $_IP_$/24 dev eth0&#39;</div><div>                                    # shutdown delegate IP command</div><div>                                    # (change requires restart)</div><div>arping_path = &#39;/usr/sbin&#39;</div><div>                                    # arping command path</div><div>                                    # (change requires restart)</div><div>arping_cmd = &#39;arping -U $_IP_$ -w 1&#39;</div><div>                                    # arping command</div><div>                                    # (change requires restart)</div><div><br></div><div># - Behaivor on escalation Setting -</div><div><br></div><div>clear_memqcache_on_escalation = on</div><div>                                    # Clear all the query cache on shared memory</div><div>                                    # when standby pgpool escalate to active pgpool</div><div>                                    # (= virtual IP holder).</div><div>                                    # This should be off if client connects to pgpool</div><div>                                    # not using virtual IP.</div><div>                                    # (change requires restart)</div><div>wd_escalation_command = &#39;&#39;</div><div>                                    # Executes this command at escalation on new active pgpool.</div><div>                                    # (change requires restart)</div><div>wd_de_escalation_command = &#39;&#39;</div><div><span class="" style="white-space:pre">                                                                        </span># Executes this command when master pgpool resigns from being master.</div><div><span class="" style="white-space:pre">                                                                        </span># (change requires restart)</div><div><br></div><div># - Lifecheck Setting -</div><div><br></div><div># -- common --</div><div><br></div><div>wd_monitoring_interfaces_list = &#39;&#39;  # Comma separated list of interfaces names to monitor.</div><div><span class="" style="white-space:pre">                                                                        </span># if any interface from the list is active the watchdog will</div><div><span class="" style="white-space:pre">                                                                        </span># consider the network is fine</div><div><span class="" style="white-space:pre">                                                                        </span># &#39;any&#39; to enable monitoring on all interfaces except loopback</div><div><span class="" style="white-space:pre">                                                                        </span># &#39;&#39; to disable monitoring</div><div><br></div><div><br></div><div>wd_lifecheck_method = &#39;heartbeat&#39;</div><div>                                    # Method of watchdog lifecheck (&#39;heartbeat&#39; or &#39;query&#39; or &#39;external&#39;)</div><div>                                    # (change requires restart)</div><div>wd_interval = 10</div><div>                                    # lifecheck interval (sec) &gt; 0</div><div>                                    # (change requires restart)</div><div><br></div><div># -- heartbeat mode --</div><div><br></div><div>wd_heartbeat_port = 9694</div><div>                                    # Port number for receiving heartbeat signal</div><div>                                    # (change requires restart)</div><div>wd_heartbeat_keepalive = 2</div><div>                                    # Interval time of sending heartbeat signal (sec)</div><div>                                    # (change requires restart)</div><div>wd_heartbeat_deadtime = 30</div><div>                                    # Deadtime interval for heartbeat signal (sec)</div><div>                                    # (change requires restart)</div><div>heartbeat_destination0 = &#39;199.203.248.243&#39;</div><div>                                    # Host name or IP address of destination 0</div><div>                                    # for sending heartbeat signal.</div><div>                                    # (change requires restart)</div><div>heartbeat_destination_port0 = 9694</div><div>                                    # Port number of destination 0 for sending</div><div>                                    # heartbeat signal. Usually this is the</div><div>                                    # same as wd_heartbeat_port.</div><div>                                    # (change requires restart)</div><div>heartbeat_device0 = &#39;eth0&#39;</div><div>                                    # Name of NIC device (such like &#39;eth0&#39;)</div><div>                                    # used for sending/receiving heartbeat</div><div>                                    # signal to/from destination 0.</div><div>                                    # This works only when this is not empty</div><div>                                    # and pgpool has root privilege.</div><div>                                    # (change requires restart)</div><div><br></div><div>heartbeat_destination1 = &#39;199.203.248.216&#39;</div><div>heartbeat_destination_port1 = 9694</div><div>heartbeat_device1 = &#39;eth0&#39;</div><div><br></div><div># -- query mode --</div><div><br></div><div>wd_life_point = 3</div><div>                                    # lifecheck retry times</div><div>                                    # (change requires restart)</div><div>wd_lifecheck_query = &#39;SELECT 1&#39;</div><div>                                    # lifecheck query to pgpool from watchdog</div><div>                                    # (change requires restart)</div><div>wd_lifecheck_dbname = &#39;template1&#39;</div><div>                                    # Database name connected for lifecheck</div><div>                                    # (change requires restart)</div><div>wd_lifecheck_user = &#39;nobody&#39;</div><div>                                    # watchdog user monitoring pgpools in lifecheck</div><div>                                    # (change requires restart)</div><div>wd_lifecheck_password = &#39;&#39;</div><div>                                    # Password for watchdog user in lifecheck</div><div>                                    # (change requires restart)</div><div><br></div><div># - Other pgpool Connection Settings -</div><div><br></div><div>other_pgpool_hostname0 = &#39;199.203.248.243&#39;</div><div>                                    # Host name or IP address to connect to for other pgpool 0</div><div>                                    # (change requires restart)</div><div>other_pgpool_port0 = 9999</div><div>                                    # Port number for othet pgpool 0</div><div>                                    # (change requires restart)</div><div>other_wd_port0 = 9000</div><div>                                    # Port number for othet watchdog 0</div><div>                                    # (change requires restart)</div><div>other_pgpool_hostname1 = &#39;199.203.248.216&#39;</div><div>other_pgpool_port1 = 9999</div><div>other_wd_port1 = 9000</div><div><br></div><div><br></div><div>#------------------------------------------------------------------------------</div><div># OTHERS</div><div>#------------------------------------------------------------------------------</div><div>relcache_expire = 0</div><div>                                   # Life time of relation cache in seconds.</div><div>                                   # 0 means no cache expiration(the default).</div><div>                                   # The relation cache is used for cache the</div><div>                                   # query result against PostgreSQL system</div><div>                                   # catalog to obtain various information</div><div>                                   # including table structures or if it&#39;s a</div><div>                                   # temporary table or not. The cache is</div><div>                                   # maintained in a pgpool child local memory</div><div>                                   # and being kept as long as it survives.</div><div>                                   # If someone modify the table by using</div><div>                                   # ALTER TABLE or some such, the relcache is</div><div>                                   # not consistent anymore.</div><div>                                   # For this purpose, cache_expiration</div><div>                                   # controls the life time of the cache.</div><div><br></div><div>relcache_size = 256</div><div>                                   # Number of relation cache</div><div>                                   # entry. If you see frequently:</div><div>                                   # &quot;pool_search_relcache: cache replacement happend&quot;</div><div>                                   # in the pgpool log, you might want to increate this number.</div><div><br></div><div>check_temp_table = on</div><div>                                   # If on, enable temporary table check in SELECT statements.</div><div>                                   # This initiates queries against system catalog of primary/master</div><div>                                   # thus increases load of master.</div><div>                                   # If you are absolutely sure that your system never uses temporary tables</div><div>                                   # and you want to save access to primary/master, you could turn this off.</div><div>                                   # Default is on.</div><div><br></div><div>check_unlogged_table = on</div><div>                                   # If on, enable unlogged table check in SELECT statements.</div><div>                                   # This initiates queries against system catalog of primary/master</div><div>                                   # thus increases load of master.</div><div>                                   # If you are absolutely sure that your system never uses unlogged tables</div><div>                                   # and you want to save access to primary/master, you could turn this off.</div><div>                                   # Default is on.</div><div><br></div><div>#------------------------------------------------------------------------------</div><div># IN MEMORY QUERY MEMORY CACHE</div><div>#------------------------------------------------------------------------------</div><div>memory_cache_enabled = off</div><div><span class="" style="white-space:pre">                                                                </span>   # If on, use the memory cache functionality, off by default</div><div>memqcache_method = &#39;shmem&#39;</div><div><span class="" style="white-space:pre">                                                                </span>   # Cache storage method. either &#39;shmem&#39;(shared memory) or</div><div><span class="" style="white-space:pre">                                                                </span>   # &#39;memcached&#39;. &#39;shmem&#39; by default</div><div>                                   # (change requires restart)</div><div>memqcache_memcached_host = &#39;localhost&#39;</div><div><span class="" style="white-space:pre">                                                                </span>   # Memcached host name or IP address. Mandatory if</div><div><span class="" style="white-space:pre">                                                                </span>   # memqcache_method = &#39;memcached&#39;.</div><div><span class="" style="white-space:pre">                                                                </span>   # Defaults to localhost.</div><div>                                   # (change requires restart)</div><div>memqcache_memcached_port = 11211</div><div><span class="" style="white-space:pre">                                                                </span>   # Memcached port number. Mondatory if memqcache_method = &#39;memcached&#39;.</div><div><span class="" style="white-space:pre">                                                                </span>   # Defaults to 11211.</div><div>                                   # (change requires restart)</div><div>memqcache_total_size = 67108864</div><div><span class="" style="white-space:pre">                                                                </span>   # Total memory size in bytes for storing memory cache.</div><div><span class="" style="white-space:pre">                                                                </span>   # Mandatory if memqcache_method = &#39;shmem&#39;.</div><div><span class="" style="white-space:pre">                                                                </span>   # Defaults to 64MB.</div><div>                                   # (change requires restart)</div><div>memqcache_max_num_cache = 1000000</div><div><span class="" style="white-space:pre">                                                                </span>   # Total number of cache entries. Mandatory</div><div><span class="" style="white-space:pre">                                                                </span>   # if memqcache_method = &#39;shmem&#39;.</div><div><span class="" style="white-space:pre">                                                                </span>   # Each cache entry consumes 48 bytes on shared memory.</div><div><span class="" style="white-space:pre">                                                                </span>   # Defaults to 1,000,000(45.8MB).</div><div>                                   # (change requires restart)</div><div>memqcache_expire = 0</div><div><span class="" style="white-space:pre">                                                                </span>   # Memory cache entry life time specified in seconds.</div><div><span class="" style="white-space:pre">                                                                </span>   # 0 means infinite life time. 0 by default.</div><div>                                   # (change requires restart)</div><div>memqcache_auto_cache_invalidation = on</div><div><span class="" style="white-space:pre">                                                                </span>   # If on, invalidation of query cache is triggered by corresponding</div><div><span class="" style="white-space:pre">                                                                </span>   # DDL/DML/DCL(and memqcache_expire).  If off, it is only triggered</div><div><span class="" style="white-space:pre">                                                                </span>   # by memqcache_expire.  on by default.</div><div>                                   # (change requires restart)</div><div>memqcache_maxcache = 409600</div><div><span class="" style="white-space:pre">                                                                </span>   # Maximum SELECT result size in bytes.</div><div><span class="" style="white-space:pre">                                                                </span>   # Must be smaller than memqcache_cache_block_size. Defaults to 400KB.</div><div>                                   # (change requires restart)</div><div>memqcache_cache_block_size = 1048576</div><div><span class="" style="white-space:pre">                                                                </span>   # Cache block size in bytes. Mandatory if memqcache_method = &#39;shmem&#39;.</div><div><span class="" style="white-space:pre">                                                                </span>   # Defaults to 1MB.</div><div>                                   # (change requires restart)</div><div>memqcache_oiddir = &#39;/var/log/pgpool/oiddir&#39;</div><div><span class="" style="white-space:pre">                                </span>   <span class="" style="white-space:pre">                                </span>   # Temporary work directory to record table oids</div><div>                                   # (change requires restart)</div><div>white_memqcache_table_list = &#39;&#39;</div><div>                                   # Comma separated list of table names to memcache</div><div>                                   # that don&#39;t write to database</div><div>                                   # Regexp are accepted</div><div>black_memqcache_table_list = &#39;&#39;</div><div>                                   # Comma separated list of table names not to memcache</div><div>                                   # that don&#39;t write to database</div><div>                                   # Regexp are accepted</div><div>backend_hostname2 = &#39;199.203.248.243&#39;</div><div>backend_port2 = 5432</div><div>backend_weight2 = 1</div><div>backend_data_directory2 = &#39;/var/lib/pgsql/9.5/data&#39;</div><div>backend_hostname3 = &#39;199.203.248.216&#39;</div><div>backend_port3 = 5432</div><div>backend_weight3 = 0</div><div>backend_data_directory3 = &#39;/var/lib/pgsql/9.5/data&#39;</div><div>backend_hostname4 = &#39;199.203.248.218&#39;</div><div>backend_port4 = 5432</div><div>backend_weight4 = 0</div><div>backend_data_directory4 = &#39;/var/lib/pgsql/9.5/data&#39;</div><div>other_pgpool_hostname2 = &#39;199.203.248.218&#39;</div><div>other_pgpool_port2 = 9999</div><div>other_wd_port2 = 9000</div><div>heartbeat_destination2 = &#39;199.203.248.218&#39;</div><div>heartbeat_device2 = &#39;eth0&#39;</div><div>heartbeat_destination_port2 = 9694</div></div><div><br></div></div>