[Pgpool-general] pcp_recovery_node
Tatsuo Ishii
ishii at sraoss.co.jp
Fri Jan 21 08:15:32 UTC 2011
This doesn't look good?
25314 2011-01-21 10:00:43 CET ERROR: function pgpool_walrecrunning() does
not exist at character 32
--
Tatsuo Ishii
SRA OSS, Inc. Japan
English: http://www.sraoss.co.jp/index_en.php
Japanese: http://www.sraoss.co.jp
> Ok, this is before starting pcp_recovery_node:
>
> # psql -p 9999 -c "show pool_nodes;" -Upostgres
> node_id | hostname | port | status | lb_weight | state
> ---------+------------------+------+--------+-----------+-------
> 0 | node1 | 5432 | 3 | 0.500000 | S
> 1 | node2 | 5432 | 2 | 0.500000 | P
>
> then, I type:
>
> pcp_recovery_node 20 localhost 9898 postgres <password> 0 -d
>
> (I have disabled creating trigger file in failback for this example)
>
> pgpool.log (I have removed some debug duplicated messages):
>
> 2011-01-21 08:58:18 LOG: pid 4467: starting recovering node 0
>
>
> 2011-01-21 08:58:18 LOG: pid 4467: starting recovery command: "SELECT
> pgpool_recovery('basebackup.sh', 'puma.intera.si', '/pgdata/data90.C/')"
>
> 2011-01-21 08:58:19 DEBUG: pid 4434: starting health checking
>
>
> 2011-01-21 08:58:19 DEBUG: pid 4434: health_check: 0 th DB node status: 3
>
>
> 2011-01-21 08:58:19 DEBUG: pid 4434: health_check: 1 th DB node status: 2
>
>
> 2011-01-21 08:58:56 LOG: pid 4467: 1st stage is done
>
>
> 2011-01-21 08:58:56 LOG: pid 4467: check_postmaster_started: try to
> connect to postmaster on hostname:node1 database:postgres user:postgres
> (retry 0 times)
> 2011-01-21 08:58:56 LOG: pid 4467: check_postmaster_started: failed to
> connect to postmaster on hostname:node1 database:postgres user:postgres
>
> 2011-01-21 08:58:59 LOG: pid 4467: check_postmaster_started: try to
> connect to postmaster on hostname:puma.intera.si database:postgres
> user:postgres (retry 1 times)
> 2011-01-21 08:58:59 LOG: pid 4467: 0 node restarted
>
>
> 2011-01-21 08:58:59 LOG: pid 4467: send_failback_request: fail back 0 th
> node request from pid 4467
>
> 2011-01-21 08:58:59 DEBUG: pid 4434: failover_handler called
>
>
> 2011-01-21 08:58:59 DEBUG: pid 4434: failover_handler: starting to select
> new master node
>
> 2011-01-21 08:58:59 LOG: pid 4434: starting fail back. reconnect host
> node1(5432)
>
> 2011-01-21 08:58:59 LOG: pid 4434: execute command:
> /usr/local/pgpool-II/etc/failback.sh 0 "node1" 5432 /pgdata/data90.C/ 0 1
> "node2" 1
> 2011-01-21 08:58:59 DEBUG: pid 4434: failover_handler: kill 4944
>
>
> 2011-01-21 08:58:59 DEBUG: pid 4944: child received shutdown request signal
> 3
>
> 2011-01-21 08:58:59 DEBUG: pid 4434: failover_handler: kill 4945
>
>
> 2011-01-21 08:58:59 DEBUG: pid 4434: failover_handler: kill 4946
>
>
> 2011-01-21 08:58:59 DEBUG: pid 4434: failover_handler: kill 4947
>
>
> 2011-01-21 08:58:59 DEBUG: pid 4434: failover_handler: kill 4948
> 2011-01-21 08:58:59 DEBUG: pid 4434: failover_handler: kill 4974
>
>
> 2011-01-21 08:58:59 DEBUG: pid 4434: failover_handler: kill 4975
>
>
> 2011-01-21 08:58:59 LOG: pid 4434: failover_handler: set new master node:
> 0
>
> 2011-01-21 08:58:59 DEBUG: pid 5583: I am 5583
>
>
> 2011-01-21 08:58:59 DEBUG: pid 4975: child received shutdown request signal
> 3
>
> 2011-01-21 08:58:59 DEBUG: pid 4971: child received shutdown request signal
> 3
> 2011-01-21 08:58:59 DEBUG: pid 4945: child received shutdown request signal
> 3
>
> 2011-01-21 08:58:59 DEBUG: pid 5613: I am 5613
>
>
> 2011-01-21 08:58:59 LOG: pid 4434: failback done. reconnect host
> node1(5432)
>
> 2011-01-21 08:58:59 DEBUG: pid 5614: I am 5614
>
>
> 2011-01-21 08:58:59 DEBUG: pid 4964: child received shutdown request signal
> 3
>
> 2011-01-21 08:58:59 DEBUG: pid 4434: pool_ssl: SSL requested but SSL support
> is not available
>
> 2011-01-21 08:58:59 DEBUG: pid 4434: s_do_auth: auth kind: 0
>
>
> 2011-01-21 08:58:59 DEBUG: pid 4434: s_do_auth: parameter status data
> received
>
> 2011-01-21 08:58:59 DEBUG: pid 4434: s_do_auth: parameter status data
> received
> 2011-01-21 08:58:59 DEBUG: pid 4434: do_query: kind: E
>
>
> 2011-01-21 08:58:59 DEBUG: pid 4434: do_query: kind: Z
>
>
> 2011-01-21 08:58:59 LOG: pid 4434: find_primary_node: do_query returns no
> rows
>
> 2011-01-21 08:58:59 LOG: pid 4434: find_primary_node: do_query returns no
> data
>
> 2011-01-21 08:58:59 LOG: pid 4434: find_primary_node: primary node id is 0
>
>
> 2011-01-21 08:58:59 DEBUG: pid 4434: reap_handler called
>
>
> 2011-01-21 08:58:59 DEBUG: pid 4434: reap_handler: call wait3
>
>
> 2011-01-21 08:58:59 DEBUG: pid 4434: child 4944 exits with status 0
> 2011-01-21 08:58:59 DEBUG: pid 4434: reap_handler: normally exited
>
>
> 2011-01-21 08:58:59 LOG: pid 4467: recovery done
>
>
> 2011-01-21 08:58:59 DEBUG: pid 4468: do_query: kind: E
>
>
> 2011-01-21 08:58:59 ERROR: pid 4468: pool_read: EOF encountered with backend
>
>
> 2011-01-21 08:58:59 ERROR: pid 4468: do_query: error while reading message
> kind
>
> 2011-01-21 08:58:59 ERROR: pid 4468: check_replication_time_lag: SELECT
> pg_current_xlog_location() failed
>
> 2011-01-21 08:59:09 DEBUG: pid 4434: starting health checking
>
>
> 2011-01-21 08:59:09 DEBUG: pid 4434: health_check: 0 th DB node status: 1
>
>
> 2011-01-21 08:59:09 DEBUG: pid 4434: health_check: 1 th DB node status: 2
>
>
> 2011-01-21 08:59:09 DEBUG: pid 4468: pool_ssl: SSL requested but SSL support
> is not available
>
> 2011-01-21 08:59:09 DEBUG: pid 4468: s_do_auth: auth kind: 0
> 2011-01-21 09:03:40 DEBUG: pid 4468: do_query: kind: T
> 2011-01-21 09:03:40 DEBUG: pid 4468: num_fileds: 1
> 2011-01-21 09:03:40 DEBUG: pid 4468: do_query: kind: E
> 2011-01-21 09:03:40 DEBUG: pid 4468: do_query: kind: Z
> 2011-01-21 09:03:40 ERROR: pid 4468: check_replication_time_lag: SELECT
> pg_current_xlog_location() returns no rows
> 2011-01-21 09:03:49 DEBUG: pid 4434: starting health checking
> 2011-01-21 09:03:49 DEBUG: pid 4434: health_check: 0 th DB node status: 1
> 2011-01-21 09:03:50 DEBUG: pid 4434: health_check: 1 th DB node status: 2
> 2011-01-21 09:03:50 DEBUG: pid 4468: do_query: kind: T
> 2011-01-21 09:03:50 DEBUG: pid 4468: num_fileds: 1
> 2011-01-21 09:03:50 DEBUG: pid 4468: do_query: kind: E
> 2011-01-21 09:03:50 DEBUG: pid 4468: do_query: kind: Z
> 2011-01-21 09:03:50 ERROR: pid 4468: check_replication_time_lag: SELECT
> pg_current_xlog_location() returns no rows
> 2011-01-21 09:04:00 DEBUG: pid 4434: starting health checking
> 2011-01-21 09:04:00 DEBUG: pid 4434: health_check: 0 th DB node status: 1
> 2011-01-21 09:04:00 DEBUG: pid 4434: health_check: 1 th DB node status: 2
> 2011-01-21 09:04:00 DEBUG: pid 4468: do_query: kind: T
> 2011-01-21 09:04:00 DEBUG: pid 4468: num_fileds: 1
>
> postgresql.conf (node1):
>
> 25308 2011-01-21 10:00:40 CET LOG: database system was interrupted; last
> known up at 2011-01-21 09:58:57 CET
>
> 25308 2011-01-21 10:00:40 CET LOG: creating missing WAL directory
> "pg_xlog/archive_status"
>
> 25308 2011-01-21 10:00:40 CET LOG: entering standby mode
>
>
> 25309 2011-01-21 10:00:40 CET LOG: streaming replication successfully
> connected to primary
>
> 25308 2011-01-21 10:00:41 CET LOG: redo starts at 1/E4000020
>
>
> 25308 2011-01-21 10:00:41 CET LOG: consistent recovery state reached at
> 1/E5000000
>
> 25306 2011-01-21 10:00:41 CET LOG: database system is ready to accept read
> only connections
>
> 25314 2011-01-21 10:00:43 CET LOG: statement: SELECT pg_is_in_recovery()
> AND pgpool_walrecrunning()
>
> 25314 2011-01-21 10:00:43 CET ERROR: function pgpool_walrecrunning() does
> not exist at character 32
>
> 25314 2011-01-21 10:00:43 CET HINT: No function matches the given name and
> argument types. You might need to add explicit type casts.
>
> 25314 2011-01-21 10:00:43 CET STATEMENT: SELECT pg_is_in_recovery() AND
> pgpool_walrecrunning()
>
> 25314 2011-01-21 10:00:43 CET LOG: unexpected EOF on client connection
>
>
> 25319 2011-01-21 10:00:53 CET LOG: statement: SELECT
> pg_current_xlog_location()
>
> 25319 2011-01-21 10:00:53 CET ERROR: recovery is in progress
>
>
> 25319 2011-01-21 10:00:53 CET HINT: WAL control functions cannot be
> executed during recovery.
>
> 25319 2011-01-21 10:00:53 CET STATEMENT: SELECT pg_current_xlog_location()
>
>
> 25319 2011-01-21 10:01:03 CET LOG: statement: SELECT
> pg_current_xlog_location()
>
> 25319 2011-01-21 10:01:03 CET ERROR: recovery is in progress
>
>
> 25319 2011-01-21 10:01:03 CET HINT: WAL control functions cannot be
> executed during recovery.
>
> 25319 2011-01-21 10:01:03 CET STATEMENT: SELECT pg_current_xlog_location()
>
>
> 25319 2011-01-21 10:01:13 CET LOG: statement: SELECT
> pg_current_xlog_location()
>
> 25319 2011-01-21 10:01:13 CET ERROR: recovery is in progress
>
>
> 25319 2011-01-21 10:01:13 CET HINT: WAL control functions cannot be
> executed during recovery.
> 25319 2011-01-21 10:01:13 CET STATEMENT: SELECT pg_current_xlog_location()
> 25319 2011-01-21 10:01:23 CET LOG: statement: SELECT
> pg_current_xlog_location()
> 25319 2011-01-21 10:01:23 CET ERROR: recovery is in progress
> 25319 2011-01-21 10:01:23 CET HINT: WAL control functions cannot be
> executed during recovery.
> 25319 2011-01-21 10:01:23 CET STATEMENT: SELECT pg_current_xlog_location()
> 25319 2011-01-21 10:01:33 CET LOG: statement: SELECT
> pg_current_xlog_location()
>
> postgresql.conf (node2):
>
> 5853 2011-01-21 09:58:57 CET LOG: statement: SELECT
> pgpool_recovery('basebackup.sh', node1', '/pgdata/data90.C/')
> 5856 2011-01-21 09:58:57 CET LOG: statement: SELECT
> pg_start_backup('Streaming Replication', true)
> pg_start_backup
> -----------------
> 1/E4000020
> (1 row)
>
> 5887 2011-01-21 09:59:34 CET LOG: statement: SELECT pg_stop_backup()
> NOTICE: pg_stop_backup complete, all required WAL segments have been
> archived
> pg_stop_backup
> ----------------
> 1/E40000D8
> (1 row)
>
> 5853 2011-01-21 09:59:35 CET LOG: statement: SELECT
> pgpool_remote_start('node1', '/pgdata/data90.C/')
> 5893 2011-01-21 09:59:36 CET LOG: replication connection authorized:
> user=postgres host=192.168.10.120 port=45066
>
>
>
> On Fri, Jan 21, 2011 at 8:44 AM, Tatsuo Ishii <ishii at sraoss.co.jp> wrote:
>
>> Ok. Can you show the pgpool log and PostgreSQL log(both node 0 and 1)
>> when you did:
>>
>> >> > pcp_recovery_node 20 localhost 9898 postgres <password> 0 -d
>> --
>> Tatsuo Ishii
>> SRA OSS, Inc. Japan
>> English: http://www.sraoss.co.jp/index_en.php
>> Japanese: http://www.sraoss.co.jp
>>
>> > Actually, yesterday in desperate moments I have done that, I got
>> > cvs version of pgpool.
>> > I think I have same situation.
>> > I have added failback script that creates trigger file on my new primary
>> > node,
>> > but standby is acting like primary..
>> >
>> > On Fri, Jan 21, 2011 at 8:19 AM, Tatsuo Ishii <ishii at sraoss.co.jp>
>> wrote:
>> >
>> >> > I'm having problem with pcp_recovery_node.
>> >> > I have 3 computer architecture. One computer is with installed
>> pg_pool,
>> >> and
>> >> > two other are nodes.
>> >> > Pg_pool version installed is 3.0.1, and postgres is 9.0.2
>> >> > I have streaming replication configured.
>> >> >
>> >> > I'm trying to setup automatic recovery of nodes.
>> >> > I set one server to be master, and other slave (read only connection).
>> >> > When primary fails, slave is being used as a master and this
>> transition
>> >> from
>> >> > one
>> >> > node to another is automatic and everything works great.
>> >> > But, when I execute pcp_recovery_node command:
>> >> >
>> >> > pcp_recovery_node 20 localhost 9898 postgres <password> 0 -d
>> >> >
>> >> > pgpool is trying to make my first node (one that I'm trying to
>> recover)
>> >> > as a master node, and everything went well except crating trigger
>> file.
>> >>
>> >> That is the limitation of pgpool-II 3.0.x. pgpool-II 3.1 will make
>> >> this possible. You can try the to-be-pgpool-II 3.1 source code from
>> >> CVS HEAD or tar ball from
>> >>
>> >>
>> http://pgpool.projects.postgresql.org/contrib_docs/simple_sr_setting/index.html
>> >> --
>> >> Tatsuo Ishii
>> >> SRA OSS, Inc. Japan
>> >> English: http://www.sraoss.co.jp/index_en.php
>> >> Japanese: http://www.sraoss.co.jp
>> >>
>> >> > So, in log file there is no errors, failover.sh is not executed, and
>> >> > node1 (just recovered node) is acting like it is master, but it is not
>> >> > until I manually create trigger file.
>> >> >
>> >> > Why pgpool is trying to make my just recovered node as master, and
>> >> > not just bring it up and set it to act like slave. Can I configure
>> this?
>> >> >
>> >> > How can I know what action is provided by executing pcp_recovery_node?
>> >> >
>> >> > I hope that I descibed my problem well.
>>
>
>
>
> --
> Armin
More information about the Pgpool-general
mailing list