RE: slapd stopping with no error message

["Douglas B. Jones" <douglas@gpc.edu>]

I attached a gdb to the new process plus the -d 7 output (you know,
that neat thing that unix has called redirection). Hopefully, the
connection I have to the computer will not be lost and I will be
able to get the trace if it fails. Also, not the results below are
just the last part of the output. Let me know if I need more. For
what it is worth, I did the 'he' search shown at the bottom below
and it did not fail.... Thanks!

-----Original Message-----
From: openldap-software-bounces+douglas=gpc.edu@openldap.org
[mailto:openldap-software-bounces+douglas=gpc.edu@openldap.org]On Behalf
Of Douglas B. Jones
Sent: Friday, February 23, 2007 2:22 PM
To: openldap-software@openldap.org
Cc: douglas@gpc.edu
Subject: RE: slapd stopping with no error message



Finally failed, I did not have a gdb on it, but did have -d 7. Below
are the results. Thanks!

do_search
ldap_read: want=8 error=Resource temporarily unavailable
ber_scanf fmt ({miiiib) ber:
>>> dnPrettyNormal: <o=gpc,c=us>
=> ldap_bv2dn(o=gpc,c=us,0)
<= ldap_bv2dn(o=gpc,c=us)=0 
=> ldap_dn2bv(272)
<= ldap_dn2bv(o=gpc,c=us)=0 
=> ldap_dn2bv(272)
<= ldap_dn2bv(o=gpc,c=us)=0 
<<< dnPrettyNormal: <o=gpc,c=us>, <o=gpc,c=us>
SRCH "o=gpc,c=us" 2 0    100 3600 0
ber_scanf fmt ({m) ber:
ber_scanf fmt (m) ber:
ber_scanf fmt ({m) ber:
ber_scanf fmt (m) ber:
ber_scanf fmt ({m) ber:
ber_scanf fmt (m) ber:
    filter: (|(cn=he*)(mail=he*)(sn=he*))
ber_scanf fmt ({M}}) ber:
    attrs: cn mail
*** glibc detected *** corrupted double-linked list: 0x29400048 ***
/etc/init.d/gpcldap-33: line 51: 11051 Aborted                 ${slapd} -d 7 -u $user -h "ldap://:$P/"; $OPTIONS $SLAPD_OPTIONS

-----Original Message-----
From: openldap-software-bounces+douglas=gpc.edu@openldap.org
[mailto:openldap-software-bounces+douglas=gpc.edu@openldap.org]On Behalf
Of Douglas B. Jones
Sent: Wednesday, February 21, 2007 12:38 PM
To: openldap-software@openldap.org
Cc: douglas@gpc.edu
Subject: RE: slapd stopping with no error message



Howard pointed out that I am still asleep - he stated what to
do below '-d 7'. Thanks! 

-----Original Message-----
From: openldap-software-bounces+douglas=gpc.edu@openldap.org [mailto:openldap-software-bounces+douglas=gpc.edu@openldap.org] On
Behalf Of Douglas B. Jones
Sent: Wednesday, February 21, 2007 10:52 AM
To: 'Howard Chu'
Cc: douglas@gpc.edu; openldap-software@openldap.org
Subject: RE: slapd stopping with no error message


Slap/slap/slap - waking up now - sorry not much sleep since last week - sick
child, still should never forget the most basic. I will redirect. Question: what
level of debug would you recommend. Thanks!

-----Original Message-----
From: openldap-software-bounces+douglas=gpc.edu@openldap.org [mailto:openldap-software-bounces+douglas=gpc.edu@openldap.org] On
Behalf Of Howard Chu
Sent: Wednesday, February 21, 2007 10:43 AM
To: Douglas B. Jones
Cc: openldap-software@openldap.org
Subject: Re: slapd stopping with no error message

Douglas B. Jones wrote:
> About three times in the last several days, ldap-2.3.33 (RHEL4) has just stopped.
> No core or error messages in the log files. The last entry in the log file
> is just a search entry, with nothing common in the search from 'crash' to
> 'crash'. I have even tried the searches and they worked fine. I restart it
> notes an unclean shutdown detected and attempts recovery and appears to do so.
> 
> My question is what is the proper way to debug this. If I do -d, it will
> not fork, but since it fails at random times I probably will not be there
> to see the output.

Unix has this cool thing called I/O redirection, perhaps you've heard of it?
There's also this other cool thing called job control...

In [t]csh I would use
	slapd -d 7 >& Errs &
In sh I would use
	nohup slapd -d 7 > Errs 2>&1 &

And if any of that looks unfamiliar to you, then you Really Really need to go 
read Kernighan & Pike's _The_Unix_Programming_Environment_ before you do 
anything else.

> Is this the best way to try to see what is happening?
> If so, what is the recommended debug level? It was built with:
> 
> ./configure \
>         --prefix=/usr/local/openldap-2.3.33 \
>         --includedir=/usr/local/lib/BerkleyDB.4.5.20.NC/include \
>         --enable-crypt \
>         --enable-cleartext \
>         --enable-debug \
>         --enable-meta=yes \
>         --enable-monitor=yes \
>         --enable-relay=yes \
>         --enable-ldap=yes \
>         --enable-rewrite=yes \
>         --enable-overlays=yes \
>         -exec-prefix=/usr/local/openldap-2.3.33
> 
> I see that there is 2.3.34 out and that it can use BDB 4.5. Is the
> best course of action to upgrade?

All OpenLDAP 2.3.x releases work with BDB 4.5 (and 4.6, for that matter). The 
explicit test in configure is just a formality. (If the configure test didn't 
find a specific BDB version that it recognized, it would eventually fallback 
to -ldb4 which would work anyway.)
-- 
   -- Howard Chu
   Chief Architect, Symas Corp.  http://www.symas.com
   Director, Highland Sun        http://highlandsun.com/hyc
   Chief Architect, OpenLDAP     http://www.openldap.org/project/