Db_stat performance - will it depend on database size ?

Question

Does db_stat have to trawl through the database to get its information or is BerkeleyDB clever about maintaining the information ready for db_stat to just query and return ?
We're running db_stat (from a perl program as it happens) and finding that for large databases - 10G or more - it can take over an hour to run so I guess that would suggest it's trawling the database. But for some of our databases it seems to be almost instantaneous. I can't find any docs anywhere
about how it works internally.

Answer 1

FWIW, using the perl wrapper for BerkeleyDB here http://search.cpan.org/dist/BerkeleyDB/BerkeleyDB.pod you can do
my $stats = $self->{db}->db_stat(DB_FAST_STAT);
though it's not 100% clear from the docs. If you still want a count of unique keys (bt_nkeys), create and open the database with the DB_RECNUM option, as below
$self->{db} = new BerkeleyDB::Btree
-Filename => $self->{dbName},
-Env => $self->{env},
-Flags => DB_CREATE | DB_INIT_LOG,
-Property => DB_RECNUM
or croak "Cannot open main database $self->{dbName}: $! $self->{db}::Error\n";