Package: postgresql-common
Version: 73
Severity: normal
I'm using the postgresql packages in an environment having 20,000+
groups, queried over LDAP on a remote server. In this situation, the
change_uid function in PgCommon.pm is unbearably slow. To wit:
[EMAIL PROTECTED]:~$ time sudo perl -I /usr/share/postgresql-common/ -MPgCommon
-e 'change_ugid(29,29)'
real 36m7.986s
user 0m39.722s
sys 0m18.360s
The following implementation seems to work equivalently, and is orders
of magnitude faster.
------------------------------------------------------------------
sub change_uid {
my ($uid, $gid) = @_;
my $groups = $gid;
$groups .= " $groups"; # first additional group
# collect all auxiliary groups the user is in
my ($uname, undef) = getpwuid($uid);
$groups .= " " . `id -G $uname`;
$) = $groups;
$( = $gid;
$> = $< = $uid;
error 'Could not change user id' if $< != $uid;
error 'Could not change group id' if $( != $gid;
}
------------------------------------------------------------------
Please consider integrating it. The id command is in coreutils, so I
assume there are no dependency problems involved.
Alternatively, if for some reason this would not work (my unix-fu is not
that high) you should consider cacheing the results of the getpw* calls.
Thanks.
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
