[Pgpool-hackers] [Pgpool-general] node 0 is not valid backend

[erobles]

Hi!

I have the following problem with pgpool:
I have the backend system replicating between  backend0 and  backend1 ..

then unplug  the network cable from the backend1 and I  wait for  the 
health check.

while i  receive  these messages:

notice_backend_Error: 1 fail over request several times until  the 
failover  run and  starts degeneration mode

then   the new master node was  established  like the  same backend0, 
and immediately sends the message  backend 0 is not valid backend.


If the backend1 has been disconnected, and   the backend0   was 
established  as a master, how  is possible  pgpool tell me that is not 
valid?


this is the detail of messages:

2010-06-01 19:21:18 ERROR: pid 8594: new_connection: create_cp() failed
2010-06-01 19:21:18 ERROR: pid 8470: connect_inet_domain_socket: 
connect() failed: Connection refused
2010-06-01 19:21:18 ERROR: pid 8470: health check failed. 1 th host 
backend1 at port 6543 is down
2010-06-01 19:21:18 LOG:   pid 8470: set
2010-06-01 19:21:18 LOG:   pid 8594: notice_backend_error: 1 fail over 
request from pid 8594
2010-06-01 19:21:18 ERROR: pid 8489: connect_inet_domain_socket: 
connect() failed: Connection refused
2010-06-01 19:21:18 ERROR: pid 8489: connection to backend1(6543) failed 
1 th backend down status
2010-06-01 19:21:18 LOG:   pid 8470:
2010-06-01 19:21:18 ERROR: pid 8489: new_connection: create_cp() failed
2010-06-01 19:21:18 ERROR: pid 8488: connect_inet_domain_socket: 
connect() failed: Connection refused
2010-06-01 19:21:18 ERROR: pid 8488: connection to backend1(6543) failed
2010-06-01 19:21:18 ERROR: pid 8488: new_connection: create_cp() failed
starting degeneration. shutdown host backend1(6543)
2010-06-01 19:21:18 ERROR: pid 8593: connect_inet_domain_socket: 
connect() failed: Connection refused
2010-06-01 19:21:18 LOG:   pid 8470: execute command: 
/usr/local/pgsql83/data/pgpool_failover  1 backend1 6543 
/usr/local/pgsql83/data 0 0

2010-06-01 19:21:18 ERROR: pid 8593: connection to backend1(6543) failed
2010-06-01 19:21:18 ERROR: pid 8593: new_connection: create_cp() failed
Executing pgpool_failover as user root
Failover of node 1 at hostname backend1. New Master is 0. Old master 
node was 0.
2010-06-01 19:21:18 LOG:   pid 8470: failover_handler: set new master 
node: 0
2010-06-01 19:21:18 LOG:   pid 8593: notice_backend_error: node 0 is not 
valid backend.
2010-06-01 19:21:18 LOG:   pid 8470: failover done. shutdown host 
backend1(6543)

2010-06-01 19:21:18 LOG:   pid 8489: notice_backend_error: node 0 is not 
valid backend.
2010-06-01 19:21:18 LOG:   pid 8488: notice_backend_error: node 0 is not 
valid backend.
2010-06-01 19:21:20 LOG:   pid 8470: failover: no backends are degenerated








This is an extract of pgpool.conf:



#set pgpool2 hostname
pgpool2_hostname = 'servidor'


# system DB info
system_db_hostname = 'backend0'
system_db_port = 6543
system_db_dbname = 'pgpool'
system_db_schema = 'pgpool_catalog'
system_db_user = 'postgres'
system_db_password = ''


# backend_hostname, backend_port, backend_weight
# here are examples
backend_hostname0 = 'backend0'
backend_port0 = 6543
backend_weight0 = 1
backend_data_directory0 = '/usr/local/pgsql83/data'

backend_hostname1 = 'backend1'
backend_port1 = 6543
backend_weight1 = 1
backend_data_directory1 = '/usr/local/pgsql83/data'









_______________________________________________
Pgpool-general mailing list
Pgpool-general at pgfoundry.org
http://pgfoundry.org/mailman/listinfo/pgpool-general