suffixalias gets stuck in 2.1.5

[Rick van Rein <rick@vanrein.org>]

Hello,

I got OpenLDAP 2.1.5 working on FreeBSD, but it consistently gets stuck
with suffixaliases (which I'm trying for the first time).

My bdb database is setup like this (btw, I really missed a HOWTO on
suffixaliases, or just example configs):

database        bdb
directory       /var/db/openldap/openfortress
index           objectClass eq
suffixalias     "dc=openfortress,dc=nl" "o=OpenFortress,c=nl"
suffix          "dc=openfortress,dc=nl"
suffix          "o=OpenFortress,c=nl"
rootdn          "cn=Directory Manager,ou=Contacts,o=OpenFortress,c=nl"
rootpw          wouldntyouliketoknow

Anything under base DN 'o=Openfortress,c=nl' works flawlessly.
Anything under base DN 'dc=openfortress,dc=nl' invariably gets stuck.

When OpenLDAP gets stuck/blocked
 * it only does so for that single thread, other requests are handled okay
 * it will detect, but not correctly count, the client being torn down
 * it will not gently go down because the client is accounted as open
 * The last words of a thread getting stuck are (due to slapd -d 15)
     suffix_alias: converted "dc=openfortress,dc=nl" to "o=OpenFortress,c=nl"
     => bdb_back_search
     bdb_dn2entry_rw("o=OpenFortress,c=nl")
     => bdb_dn2id_matched( "o=OpenFortress,c=nl" )
   Followed by radio silence.  A -1 debug only adds two more lines,
     daemon: select: listen=8 active_threads=1 tvp=NULL
     daemon: select: listen=9 active_threads=1 tvp=NULL
 * when the client tears down the connection, there is an additional
     daemon: activity on 1 descriptors
     daemon: activity on: 13r
     daemon: read activity on 13
     connection_get(13)
     connection_get(13): got connid=0
     connection_read(13): checking for input on id=0
     ber_get_next
     ldap_read: want=9, got=0

     ber_get_next on fd 13 failed errno=0 (Undefined error: 0)
     connection_read(13): input error=-2 id=0, closing.
     connection_closing: readying conn=0 sd=13 for close
     connection_close: deferring conn=0 sd=13
     daemon: select: listen=8 active_threads=1 tvp=NULL
     daemon: select: listen=9 active_threads=1 tvp=NULL
     daemon: activity on 1 descriptors
     daemon: select: listen=8 active_threads=1 tvp=NULL
     daemon: select: listen=9 active_threads=1 tvp=NULL
   That too is at -d -1.

I'm unsure if this is a software help call, or if this is a bug report.

I am not sure if have constructed a woefully wrong config?  I sure looked
up every doc on the topic, and I sure tried many different orders of suffix
and of suffixaliases; I even split out the "core" database
o=OpenFortress,c=nl into a separate backend.  Nothing worked, the stuck
daemon prevailed.

By the way, is 2.1 an alpha/beta version?


Thanks for any help!

-Rick van Rein.