Re: n_distinct off by a factor of 1000 - Mailing list pgsql-general

From Klaudie Willis
Subject Re: n_distinct off by a factor of 1000
Date
Msg-id -80UM46RdRZs8iKBV88cCySSbiaUvpSPMDWA2hE5ApvQdeV4uOVErw5rJYtakUAjHjYCygWX3SlHhrL9y7tB9klZJCxDBZTCUzi1SnHZ9uc=@protonmail.com
Whole thread Raw
In response to Re: n_distinct off by a factor of 1000  (Klaudie Willis <Klaudie.Willis@protonmail.com>)
List pgsql-general
I am not sure whether I am misinterpreting something, or if it is a bug (probably former) however, when using partitions, are not n_distinct_inherited supposed to propagate to the child partitions? It does not seem to do so. (Yes, I run Analyze after setting the variable)  I had to set the n_distinct separately on all partitions to get the desired planer behavior, but I thought that setting n_distinct_inherited was supposed to prevent manually setting the partitions.

I follow up with a dbfiddle for this problem:
 
best regards
K

pgsql-general by date:

Previous
From: Klaudie Willis
Date:
Subject: Re: n_distinct off by a factor of 1000
Next
From: Adrian Klaver
Date:
Subject: Re: error messages (autovaccum canceled and syntax errors) whileloading a DUMP