Re: OpenLDAP becomes unresponsive

[John Von Essen <john@essenz.com>]

Some more details...

I was using ldbm.... Version of OpenLDAP is 2.2.15.

After I killed the pid, and restarted, I get:

Sep  9 12:12:31 daniel slapd[17598]: >>> dnNormalize: 
<uid=mswarr,...,o=eastern.edu>
Sep  9 12:12:31 daniel slapd[17598]: <<< dnNormalize: 
<uid=mswarr,...,o=eastern.edu>

For every single entry, then the following loop begins:

Sep  9 12:14:22 daniel slapd[17540]: => id2entry_r( 1113 )
Sep  9 12:14:22 daniel slapd[17540]: ====> cache_find_entry_id( 1113 ) 
"uid=rstephan,..,o=eastern.edu" (found) (1 tries)
Sep  9 12:14:22 daniel slapd[17540]: <= id2entry_r( 1113 ) 0x415d4bb0 
(cache)
Sep  9 12:14:22 daniel slapd[17540]: ldbm_search: candidate entry 1113 
does not match filter
Sep  9 12:14:22 daniel slapd[17540]: ====> cache_return_entry_r( 1113 
): returned (0)

And this whole process starts as soon as I initiate an ldapsearch call.

In an effort to get things back up, I killed slapd and did a slapcat to 
get an ldif. Then I started from scratch with a new filesystem location 
for a bdb database. My slapd.conf looks like:

database        bdb
directory       /MailStore/BDB
index   objectClass             eq
password-hash {CLEARTEXT}
sizelimit -1
cachesize     500000
dbcachesize   50000000

Everything else is basic defaults. I then loaded the ldif I just 
created with slapadd and started slapd. Now everything is working, 
however, whenever I issue an ldapsearch, I get:

Sep  9 12:56:26 daniel slapd[17757]: bdb_search: 1 does not match filter
Sep  9 12:56:26 daniel slapd[17757]: bdb_search: 2 does not match filter
...repeats 5000 times...
Sep  9 12:56:26 daniel slapd[17757]: bdb_search: 5607 does not match 
filter
Sep  9 12:56:26 daniel slapd[17757]: bdb_search: 5608 does not match 
filter
Sep  9 12:56:26 daniel slapd[17757]: send_ldap_result: conn=18 op=1 p=3
Sep  9 12:56:26 daniel slapd[17757]: send_ldap_response: msgid=2 
tag=101 err=0
Sep  9 12:56:26 daniel slapd[17757]: connection_get(14): got connid=18
Sep  9 12:56:26 daniel slapd[17757]: connection_read(14): checking for 
input on id=18
Sep  9 12:56:26 daniel slapd[17757]: ber_get_next on fd 14 failed 
errno=0 (Error 0)
Sep  9 12:56:26 daniel slapd[17757]: connection_read(14): input 
error=-2 id=18, closing.
Sep  9 12:56:26 daniel slapd[17757]: connection_closing: readying 
conn=18 sd=14 for close
Sep  9 12:56:26 daniel slapd[17757]: connection_close: deferring 
conn=18 sd=14
Sep  9 12:56:26 daniel slapd[17757]: do_unbind
Sep  9 12:56:26 daniel slapd[17757]: connection_resched: attempting 
closing conn=18 sd=14
Sep  9 12:56:26 daniel slapd[17757]: connection_close: conn=18 sd=14


This happens for every ldapsearch call. The ldapsearch itself is 
successfull and returns data in a timely manner.

Is the above log data acceptable? Also, now that I am running bdb, what 
is the proper way to kill slapd without effecting the database?

Thanks
John

On Sep 9, 2004, at 12:56 PM, Quanah Gibson-Mount wrote:

> --On Thursday, September 09, 2004 12:09 PM -0400 John Von Essen 
> <john@essenz.com> wrote:
>
> > I had to kill the slapd daemon. When I restart everything, LDAP become
> > unresponsive to all requests.
>
> If you killed slapd, and you are using BDB, you probably need to run 
> db_recover.
>
> You don't say what version of OpenLDAP you are using, or what your 
> underlying database is.
>
> --Quanah
>
>
> --
> Quanah Gibson-Mount
> Principal Software Developer
> ITSS/Shared Services
> Stanford University
> GnuPG Public Key: http://www.stanford.edu/~quanah/pgp.html