syncrepl consumer locks up (ITS#3263)

[kuenne@rentec.com]

Full_Name: Karsten Kuenne
Version: 2.2.15
OS: Solaris 9
URL: ftp://ftp.openldap.org/incoming/
Submission from: (NULL) (65.213.85.251)


If I restart the syncrepl consumer I managed a couple of times to lock it up so
that I couldn't look up the syncrepl<rid> entry anymore. It was still responding
to other queries. Following is the output from "db_stat -Co":

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Locks grouped by object
Locker   Mode      Count Status  ----------------- Object ---------------
       7 READ          1 HELD    0x33080 len:   5 data: 0x000x00"0xd30x00

80000006 WRITE         1 WAIT    0x33080 len:   5 data: 0x000x00"0xd30x00


       1 READ          1 HELD    id2entry.bdb             handle        0
       8 READ          1 HELD    id2entry.bdb             handle        0

       3 READ          1 HELD    dn2id.bdb                handle        0
       a READ          1 HELD    dn2id.bdb                handle        0


If I try to stop the consumer it tells me that it's still waiting for 2 threads
and starts looping so that I have to kill -9 it. I than ran "db_recover" but
after starting the consumer again the same thing happened. I had to recreate the
database in order to get it back to work. The backend in this case was bdb but
it also happened with hdb.