Sorry, I should have included the index definition, its a normal btree index on a bigint column:
CREATE INDEX user_pictures_picture_dhash_idx ON user_pictures USING btree (picture_dhash);
And the table itself: CREATE TABLE user_pictures (picture_dhash bigint)
(and ~10 other columns not relevant for this I think)
/Victor
On Thu, Dec 17, 2015 at 12:22 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
Victor Blomqvist <vb@viblo.se> writes: >> From time to time I get this and similar errors in my Postgres log file: > < 2015-12-17 07:45:05.976 CST >ERROR: index > "user_pictures_picture_dhash_idx" contains unexpected zero page at block > 123780
Hm, can't tell for sure from the error message text, but the index name suggests that this is a hash index? > The server is a read slave, set up with streaming replication. We run > PostgreSQL 9.3.5.
Hash indexes are not WAL-logged, which means their contents do not propagate to slave servers, which basically means you cannot use them in replication setups. > Will it be fixed with a newer version of Postgres?
Adding WAL-logging to hash indexes has been on the to-do list for a long time; but it's never gotten done, in part because there has never been any clear evidence that hash indexes are better than btree indexes for any real-world purpose. I'm curious why you chose this index type in the first place.