postgresql - Slow query due to planer invalid stats - even after analyze -

i have 2 tables:

create table sf.dir_current (     id bigint primary key,     volume_id integer not null,     path varchar not null ); create index dir_volid_path_indx on dir_current (volume_id, path);  create table sf.event (     id bigint,  -- no primary key here!     volume_id integer not null,     parent_path varchar not null,     type bigint,      depth integer ); 

table dir contains ~50 millions of rows , in rows volume_id = 1. table events contains ~20k rows.

i execute following query (in plsql function - vol_id, min_id, max_id , on function params):

select dir.id parent_id, event event_row         sf.event event             left outer join sf.dir_current dir on dir.volume_id = vol_id , parent_path = dir.path         event.volume_id = vol_id             , event.id between min_id , max_id             , (depth_filter null or event.depth = depth_filter)             , (type_filter null or event.type = type_filter)         order event.depth; 

everything works fine when rows in dir table has volume_id = 1. after adding few thousand rows volume_id = 2 (and running analyze) query takes long. here explain of long running query: explain.depesz.com

as visible query planner had no idea there many rows volume_id = 2 , created plan far optimal.

after debugging found out analyze did not find row volume_id = 2. confirmed query:

starfish=# select most_common_vals, n_distinct pg_stats tablename = 'dir_current' , attname = 'volume_id';   most_common_vals | n_distinct  ------------------+------------  {1}              |          1 (1 row) 

after few analyze's finnally finds values vol_id = 2 , query gets normal execution time: explain.depesz.com

question: how prevent extremely long query time? there way force analyze find rows? or maybe manually modify stats column (setting n_distinct vol_id column not help).

i'm using postresql 9.5