LDAP Daemon Aborts

[patrickmoore]

Hi Florian

I am running Centos 4, in a VMware Virtual machine with 1GB RAM. No other LDAP apps running.

Is there anything else I can provide which will help ?

Thanks
Patrick

[jillrae]

Florian,

We are running a fully patched SLES 9, Scalix 11.0.3 (since Thursday) have 1G memory and no other LDAP anything running. We have not gone down since the 11.0.3 update on Thursday. (knock on wood)

jillrae

[florian]

can you paste the contents of your /var/opt/scalix/<instance>/s/sys/pam.d/omslapdeng and .../sys/om_ldap.conf files?

Thx,
Florian.

[eyalm]

omslapdeng:

auth sufficient om_ldap
auth sufficient om_auth
auth required pam_deny
account required om_auth
password required om_auth
session required om_auth

om_ldap.conf:

host=data.int.cardonhealthcare.com
search=subtree
base=dc=cardonhealthcare,dc=com
filter=uid=%s
tls=off

[florian]

Could you run your ldap server in debug mode as per instructions from our dev team:

There isn't enough information here to help me track it down, unfortunately,
Meanwhile, to help track this down, it would be helpful to run the LDAP server
in debug mode. To do this, kill the currently running ldap server (with kill,
not omoff) and run "omslapd -d31 > /tmp/logfile" When the LDAP server
crashes, send the log file to me or attach it to this bug. (The log file will
most likely be very large but it will compress nicely.)

The logfile is large - if anyone has one of these logfiles done at the time of crash, please ping me via PM or eMail and I will provide an upload location for you.

Cheers, Thanks,
Florian.

[jch]

I recommend using this shell script to both keep the ldap server running and to catch enough logging information for me to begin to try to find out what it causing this problem.

Code: Select all

#!/bin/sh

trap 'test -n "$pid" && kill "$pid"; exit 0' 1 2 3 15

# kill the running ldap server if necessary
kill $(omsetsvc -p slapd) >/dev/null 2>&1

log=$(mktemp -u /tmp/omslapd.log.XXXXXX)
n=1
while :; do
  /opt/scalix/bin/omslapd -d31 "$@" >$log.$n 2>&1 &
  pid=$!
  wait
  omshowlog -p2 -l19 -s slapd >>$log.$n 2>&1
  n=$(($n+1))
  # the mapper sometimes gets in a strange state if the
  # ldap server dies, so restart it
  /etc/rc.d/init.d/ldapmapper restart
done


Run the script as root and get in touch with me when you have more than one log file (all but the last log file should be as a result of a failure).

I'd recommend compressing the files with gzip -- I'd guess it will bring them down to 10% of their original size.

When you do get a log file after the bug has been triggered, get in touch with me. There's no need to tell Florian as well as he'd only pass the information on to me.

jch

[netcomrade]

Has there been any progress on this issue?

thanks

[jch]

Many thanks for the log file (sent privately). It showed the same problem as a log from someone else and that's always comforting.

I've added the pertinent last few lines from that to bug 15137 and I'm about to resolve that one with a patch that I outlined in that bug.

The fix will be in the June release, with any luck.

jch