[Date Prev][Date Next] [Chronological] [Thread] [Top]

RE: Intermittent problems in connection to slapd


I figure out the problem. On Windows OS, the TCP/IP socket was
implemented differently than Berkley. Berkely states that if the
connection backlog is full, TCP should ignore any new connection
requests thus causing clients to retry after a certain timeout has
occured. However, Microsoft behave differently. It will send a RST
message back to the client. When a client gets a RST message, it will
immediately try again. Since the backlog is not likely to have been
clear so fast, it would fail again. After a total of 3 tries, it would
go up to the application layer saying it failed.

To get around this problem, I recompiled slapd so that the backlog
specified in the listen is much higher. Since than I haven't got the
problem anymore.

Hope this helps. Thanks.


-----Original Message-----
From: Dan Rossi [mailto:daniel@electroteque.org]
Sent: Tuesday, December 10, 2002 4:33 PM
To: Alvin Wong; OpenLDAP Software List
Subject: RE: Intermittent problems in connection to slapd

look i am experiencing similar problems except , the bind has never gone
, i have posted many times and i am pretty upset.

-----Original Message-----
From: owner-openldap-software@OpenLDAP.org
[mailto:owner-openldap-software@OpenLDAP.org]On Behalf Of Alvin Wong
Sent: Wednesday, December 11, 2002 6:38 AM
To: OpenLDAP Software List
Subject: Intermittent problems in connection to slapd

OpenLDAP: 2.0.19
database: ldbm
Server OS: Windows 2000 Server
Client OS: Windows 2000


I need some help in figuring out a problem I'm experiencing.

I've two sites where my C client ldap programs (using OpenLDAP libs)
sometimes cannot bind to OpenLDAP slapd server. I would get "Can't
contact LDAP server". This problem will persist for about 20 minutes and
then the bind attempts will work again. Multiple client PCs trying to
contact the server will experience the problem at the same time.

The authentication used by the clients did not change between attempts.
slapd was running the whole time. I can ping the server computer from
the client PC during the problem period.

Has anyone experience this problem before? Any help, areas to explore
will be greatly appreciated. Thanks.