Re: [GENERAL] update performance of degenerate index

2013-01-28 Thread Filip RembiaƂkowski
Just some questions, which might be helpful. What size is this index? What is underlying table size? Is ANALYZE running regularly (autovacuum or manual)? What are stats for exported_when column (pg_stats)? Did you look at pg_locks during this lengthy update? Do you have many concurrent statements

[GENERAL] update performance of degenerate index

2013-01-28 Thread Scott Ribe
I'm seeing occasional simple-looking updates take way longer than I think they should, and if my theory about it is correct, it's not actually a problem. Consider this index, intended to provide extremely quick access to a small number of items from a much larger table: create index not_expor