LDAP CRITICAL irregular nagios message

[Patrick Willam <p.willam@gmx.de>]

Hi folks!

I've set up a Skolelinux server some three weeks ago, and since
then Nagios keeps messaging root irregularly every to every second
day that the ldap-server isn't responding ("LDAP CRITICAL ...").
Sometimes it takes just some minutes, but sometimes many hours(!),
until Nagios says ldap's back ("LDAP OK ...").

I'm looking forward to get some hints to hunt down the problem.
If you have some idea, please let me hear.

Here's the configuration I've used...
(Don't let you bother by the custom network-configuration;
its apparently not part of the problem. Listed just FYI.)

 o  Skolelinux 1.0r1+diRC2
 o  main-server profile _combined_ with terminal-server profile !
 o  full apt-get upgrade
 o  it is on purpose, that the clients (w2k-pro workstations)
    _can't_ reach the internet gateway direct, so they have to use
    the squid proxy (which filters "sex.com" etc.) on the server:
     *  gateway (10.0.2.1) attached at eth0 ('workstation NIC')
     *  16x W2k-pro clients attached at eth1 ('thin-client NIC')
     *  deactivated NAT 192.168.0/24->all in /etc/default/enable-nat
     *  added NAT 192.168.0/24->10.0.2/23 in /etc/network/interfaces
 o  the clients are to have fixed ipnrs+names instead of ltspNNN:
     *  changed MACs and names of clients in /etc/dhcp3/dhcpd.conf
     *  changed names of clients in /etc/bind/debian-edu/db.192.168
     *  changed ipnrs of clients in /etc/bind/debian-edu/db.intern
 o  configured the w2k clients for DHCP; joined domain "SKOLELINUX"
 o  set up the user accounts*) -- configured them with proxy=tjener

*) There were some problems with that, but that's something for
bugzilla and not for this mail. Worked it out, anyway.

It's not only Nagios complaining that ldap service is unreachable,
which is on the very same machine. Furthermore the users can't login
at the times reported by Nagios. At the other times everything is nice.
(So it's apparently not a problem with the custom network setup or
some kind of broken cable or something.)

I've set "loglevel" of slapd now to 256 (as supposed in #debian-edu).
But I really don't know whats behind the loglevel numbers mentioned in
the slapd.conf(5) man-page.
Any hints?

Maybe its a problem with DNS and "ldap" doesn't get resolved.
Nagios monitors DNS as well (in /etc/nagios/debian-edu/services.cfg
there's a "check_command check_dns"), doesn't it?

The next thing I would try is to add "10.0.2.2 ldap" to /etc/hosts.
But can bind9 really be a problem?
If so, hopefully someone can give some logging options for bind9.

Really looking forward for help/hints. If I don't get this in a
state of smoothly-working soon, I'll eventually be forced to remove
Skolelinux from this school again. :-(

Best regards,
  Patrick

[german version posted to german user list]

--
nice things to slip into your budget: tab-completion @ password prompt
    -- Patrick Willam