Thread: Streaming Replication and Hash Indices
Hi list, today i got bitten by the issue that streaming replication does not replicate "hash" indices as those changes are not writtento WAL. That fact is documented here: http://www.postgresql.org/docs/current/static/indexes-types.html But i could not find that information anywhere else like: https://wiki.postgresql.org/wiki/Streaming_Replication#Restrictions or http://www.postgresql.org/docs/current/static/warm-standby.html I suppose it would be very helpful to add it there, as the usual policy for "corrupt slave data" is to recreate the slavefrom a base_backup which would make you into thinking that the issue was fixed. That is wrong though, as it will reappear after some time when the data in the table and the index diverge :)! Regards, Seb
On Mon, Jan 18, 2016 at 01:15:28PM +0100, Seb Mel wrote: > Hi list, > > today i got bitten by the issue that streaming replication does not replicate "hash" indices as those changes are not writtento WAL. That fact is documented here: > http://www.postgresql.org/docs/current/static/indexes-types.html > > But i could not find that information anywhere else like: > https://wiki.postgresql.org/wiki/Streaming_Replication#Restrictions > or > http://www.postgresql.org/docs/current/static/warm-standby.html > > I suppose it would be very helpful to add it there, as the usual policy for "corrupt slave data" is to recreate the slavefrom a base_backup which would make you into thinking that the issue was fixed. > That is wrong though, as it will reappear after some time when the data in the table and the index diverge :)! We issue a warning on hash index creation in 9.5: Issue a warning during the creation of <link linkend="indexes-types">hash</> indexes because they are not crash-safe (Bruce Momjian) -- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB http://enterprisedb.com + As you are, so once was I. As I am, so you will be. + + Roman grave inscription +