On Nov 24, 2006, at 01:37 , Jerry Sievers wrote:
Have a look at the query_start field in pg_stat_activity for the
process holding locks that's causing backlog.
Doesn't this require stats_command_string = on? I don't have that
enabled on production servers.
Alexander.
My application's connections against PostgreSQL 8.1.4 seem to get
stuck in deletion operations.
Some sample ps output:
postgres 18198 10.5 20.7 1072088 863040 ? S11:59 14:23
postgres: [...] DELETE waiting
postgres 18204 11.5 20.8 1072692 867708 ? S11:59 15:43
postgres:
Alexander Staubo [EMAIL PROTECTED] writes:
My application's connections against PostgreSQL 8.1.4 seem to get
stuck in deletion operations.
...
# select * from pg_locks where pid in (18198, 18204, 18208, 18214,
18216);
You really need to show all of pg_locks, not a subset. In this
On Nov 23, 2006, at 16:27 , Tom Lane wrote:
Alexander Staubo [EMAIL PROTECTED] writes:
My application's connections against PostgreSQL 8.1.4 seem to get
stuck in deletion operations.
...
# select * from pg_locks where pid in (18198, 18204, 18208, 18214,
18216);
You really need to show all of
Alexander Staubo [EMAIL PROTECTED] writes:
See below. Seems 18204 is waiting for a process that is idle in
transaction; makes me wish that PostgreSQL could export more
information about the age of in-progress transactions. I am turning
on logging so I can determine what a future hanging
Alexander Staubo [EMAIL PROTECTED] writes:
On Nov 23, 2006, at 16:27 , Tom Lane wrote:
Alexander Staubo [EMAIL PROTECTED] writes:
My application's connections against PostgreSQL 8.1.4 seem to get
stuck in deletion operations.
...
# select * from pg_locks where pid in (18198, 18204,