Re: searching for the same object twice using the meta backend returns 'no such object' on the second search (ITS#2289)

[ntb@mts.ru]

Greetings.

Pierangelo Masarati wrote:
> 
> > Greetings.
> >
> > I tested it and it now works.
> > until all servers referenced in "uri" are responding.
> >
> > when connection refused by target server, slapd-meta core dumps.
> 
> Can you detail what happens? I tested back-meta on my system,
> and switching off one target does not result in any problem
> in slapd; the "dead" targt is simply ignored.

I'm running openldap HEAD checkout at Feb 03 or Feb 04 (didn't remember
exact day) compiled with gcc 2.95 on solaris 2.8

I use following META setup:

---
database        meta
suffix          "dc=mts,dc=ru"
lastmod         off
uri             "ldapi:///dc=mts,dc=ru"
uri             "ldap://somehost/dc=spb,dc=mts,dc=ru";
---

i.e. one's suffix is inside another's suffix.
It handles searches well until both targets running.
When slapd-meta receives TCP RST as reply to TCP SYN
(attempting to estabilish connection to target)
it core dumps.

Initially I get this error when PIX firewall sent this RST
to indicate prohibited connection;
But then I test this with two local targets, and when
I stop slapd on some target, slapd-meta core dumps on first search
as before.
I didn't test this with switched off or nonexisting target host.
I also noticed that this also applies to openldap 2.1.12
i.e. this seems be irrelevant with this its.

SMTP /Perece/.