Hello.
There is small table:
----------
create table some_table (
id int UNIQUE,
value int
);
INSERT INTO some_table values(1,0);
....
INSERT INTO some_table values(50,0);
-------------
When I do UPDATE some_table set value=... where id=...,
query execution time raises in arithmetic progression!
After about 50 updates on every row query consumes ~3 sec against 0.3
sec as it was at the beginning.
psql takes ~80% of CPU time (acording to top).
VACUUM helps to restore execution speed, but i think it is not the way out.
Is it BUG or FEATURE?
Postgres: 7.1.3;
System: Debian woody (kernel 2.4.17) on K6/450 with 128Mb RAM.