On Freitag, 12. September 2008, ralf(a)openldap.org wrote: > Update of /repo/OpenLDAP/pkg/ldap/servers/slapd/back-bdb > > Modified Files: > init.c 1.293 -> 1.294 > > Log Message: > stop and remove checkpoint task in _db_destroy (ITS#5698) Generally I'd think that this would better fit into the bdb_db_close() function, but the _db_close hooks are not called from backend_destroy_one() (which is what config_add_internal() calls when it fails). Is there a special reason for that?

Ralf Haferkamp wrote: > On Freitag, 12. September 2008, ralf(a)openldap.org wrote: >> Update of /repo/OpenLDAP/pkg/ldap/servers/slapd/back-bdb >> >> Modified Files: >> init.c 1.293 -> 1.294 >> >> Log Message: >> stop and remove checkpoint task in _db_destroy (ITS#5698) > > Generally I'd think that this would better fit into the > bdb_db_close() function, but the _db_close hooks are not called > from > backend_destroy_one() As an alternative, perhaps we should restructure the use of ca->cleanup, and move the runqueue submission into bdb_cf_cleanup. Then if the add fails, no task is submitted at all.