Hi,
After switching to LDAP authentication on our system, I noticed that
PostgreSQL start-up (and shut-down) time because ridiculously high.
(The PostgreSQL server isn't used as the back-end for LDAP; the LDAP
servers sits elsewhere).
Searching through google yielded the following post which seems still
unresolved:
http://www.mail-archive.com/debian-user@lists.debian.org/msg483707.html
By looking at the LDAP server logs, I noticed that during the long time
it takes the PostgreSQL server to start, it continuously queries the
LDAP server (there are many hundreds of users and groups in our domain).
Looking back at the start-up code, I narrowed the program down to the
/usr/share/postgresql-common/PgCommon.pm file, and more specifically
the function "change_ugid()": this functions loops through the entire
LDAP database, entry by entry, repeatedly querying the LDAP server and
consuming cpu, network and most of all lost of time ...
This seems like a bug (inefficient implementation) in PostgreSQL.
(it doesn't look like a problem in the way the LDAP client side - or
server - have been setup).
This problem is verified for postgresql-client-common versions 8.1,
and 8.3 on debian unstable, and testing.
I'd appreciate suggestions as how this can be solved.
Thanks,
Oren.
(ps. please CC me as I'm not subscribed to the list)
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
