[HACKERS] nulls in GIN index

2014-08-10 Thread worthy7
http://www.postgresql.org/docs/9.1/static/gin-implementation.html """As of PostgreSQL 9.1, NULL key values can be included in the index. Also, placeholder NULLs are included in the index for indexed items that are NULL or contain no keys according to extractValue. This allows searches that should

Re: [HACKERS] nulls in GIN index

2014-08-11 Thread worthy7
Perhaps I'm missing something Table has 2 columns, text and ftstext text: "how are you" ftstest: (nothing) Because "how" and "are" and "you" are too common to be tsvectored. Which is fine. So if a user searches for "how are you": select * from tbl_lines WHERE ftstext @@ plainto_tsquery('English

[HACKERS] GIST create index very very slow

2014-08-15 Thread worthy7
CREATE INDEX USING GIST(timerange); On 1.3 million rows this took only 30 seconds. on 70 million its already taken over a day. I swear it didn't take this long on version 9.3 Is there some kind of known bug with GIST? CPU is at 4% or less and ram is at 150mbs IO usage is at 100% but most of it i