Hola estamos trabajando sobre un servidor de postgresql 8.1 que está en
cluster con pgpool II en la versión 2.4. Recientemente nos dimos cuenta
de que cuando realizamos un insert con un select para uno de los campos,
el nodo 2 se cae.
-¿Hay alguna manera de solucionar ese problema?
Aqui está el log en el momento de realizar la operación:
/2009-09-07 19:34:20 LOG: pid 15563: statement: INSERT INTO
ventas_pago( id_venta, id_corte_caja, id_sucursal, numero_caja,
id_forma_pago, monto) VALUES
( (select max(id_venta) from
ventas where numero_caja=1 and id_sucursal=59 and id_corte_caja=2),2,
59, 1, 1, 6.50);
2009-09-07 19:34:20 DEBUG: pid 15563: do_command: Query: BEGIN
2009-09-07 19:34:20 DEBUG: pid 15563: detect_error: kind: C
2009-09-07 19:34:20 DEBUG: pid 15563: do_command: kind: C
2009-09-07 19:34:20 DEBUG: pid 15563: do_command: kind: Z
2009-09-07 19:34:20 DEBUG: pid 15563: do_command: transaction state: T
2009-09-07 19:34:20 DEBUG: pid 15563: do_command: Query: BEGIN
2009-09-07 19:34:20 DEBUG: pid 15563: detect_error: kind: C
2009-09-07 19:34:20 DEBUG: pid 15563: do_command: kind: C
2009-09-07 19:34:20 DEBUG: pid 15563: do_command: kind: Z
2009-09-07 19:34:20 DEBUG: pid 15563: do_command: transaction state: T
2009-09-07 19:34:20 DEBUG: pid 15563: get_insert_command_table_name:
extracted table name: "ventas_pago"
2009-09-07 19:34:20 DEBUG: pid 15563: do_query: kind: T
2009-09-07 19:34:20 DEBUG: pid 15563: num_fileds: 1
2009-09-07 19:34:20 DEBUG: pid 15563: do_query: kind: D
2009-09-07 19:34:20 DEBUG: pid 15563: do_query: kind: C
2009-09-07 19:34:20 DEBUG: pid 15563: do_query: kind: Z
2009-09-07 19:34:20 DEBUG: pid 15563: get_insert_command_table_name:
extracted table name: "ventas_pago"
2009-09-07 19:34:20 DEBUG: pid 15563: do_command: Query: LOCK TABLE
"ventas_pago" IN SHARE ROW EXCLUSIVE MODE
2009-09-07 19:34:20 DEBUG: pid 15563: detect_error: kind: C
2009-09-07 19:34:20 DEBUG: pid 15563: do_command: kind: C
2009-09-07 19:34:20 DEBUG: pid 15563: do_command: kind: Z
2009-09-07 19:34:20 DEBUG: pid 15563: do_command: transaction state: T
2009-09-07 19:34:20 DEBUG: pid 15563: do_command: Query: LOCK TABLE
"ventas_pago" IN SHARE ROW EXCLUSIVE MODE
2009-09-07 19:34:20 DEBUG: pid 15563: detect_error: kind: C
2009-09-07 19:34:20 DEBUG: pid 15563: do_command: kind: C
2009-09-07 19:34:20 DEBUG: pid 15563: do_command: kind: Z
2009-09-07 19:34:20 DEBUG: pid 15563: do_command: transaction state: T
2009-09-07 19:34:20 DEBUG: pid 15563: wait_for_query_response: waiting
for backend 0 completing the query
2009-09-07 19:34:20 DEBUG: pid 15563: detect_error: kind: C
2009-09-07 19:34:20 DEBUG: pid 15563: detect_error: kind: C
2009-09-07 19:34:20 DEBUG: pid 15563: detect_error: kind: C
2009-09-07 19:34:20 DEBUG: pid 15563: detect_error: kind: C
2009-09-07 19:34:20 DEBUG: pid 15563: wait_for_query_response: waiting
for backend 1 completing the query
2009-09-07 19:34:20 DEBUG: pid 15563: read_kind_from_backend: read kind
from 0 th backend C NUM_BACKENDS: 2
2009-09-07 19:34:20 DEBUG: pid 15563: read_kind_from_backend: read kind
from 1 th backend E NUM_BACKENDS: 2
2009-09-07 19:34:20 ERROR: pid 15563: read_kind_from_backend: 1 th kind
E does not match with master or majority connection kind C
2009-09-07 19:34:20 ERROR: pid 15563: kind mismatch among backends.
Possible last query was: "INSERT INTO ventas_pago( id_venta,
id_corte_caja, id_sucursal, numero_caja, id_forma_pago, monto) VALUES
( (select max(id_venta) from
ventas where numero_caja=1 and id_sucursal=59 and id_corte_caja=2),2,
59, 1, 1, 6.50);" kind details are: 0[C] 1[E]
2009-09-07 19:34:20 LOG: pid 15563: notice_backend_error: 1 fail over
request from pid 15563
2009-09-07 19:34:20 DEBUG: pid 14949: failover_handler called
2009-09-07 19:34:20 DEBUG: pid 14949: failover_handler: starting to
select new master node
2009-09-07 19:34:20 LOG: pid 14949: starting degeneration. shutdown
host node2(5432)/
Saludos
Atentamente
Manuel Alejandro Estevez Fernandez
