(ITS#5488) syncrepl received contextCSN not passed on to syncprov consumers

[rein@OpenLDAP.org]

Full_Name: Rein Tollevik
Version: CVS head
OS: 
URL: ftp://ftp.openldap.org/incoming/
Submission from: (NULL) (81.93.160.250)
Submitted by: rein


When syncrepl and syncprov are both used on a glue database, the
contextCSN received from the syncrepl producers are not passed on to the
syncprov consumers when changes in subordinate databases are received.
The reason is that syncrepl queues the CSNs in the glue backend, while
syncprov fetches them from the backend where the changes are made.  As a
consequence, the consumers will be passed a cookie without any csn
value.

My first attempt at fixing this was to change syncprov to fetch the
queued csn values from the glue backend where it was used.  But that
failed as other modules queues the csn values in their own backend when
they changes things.  Instead I changed ctxcsn.c so that it always
queues them in the glue backend where syncprov is used.  But I don't
feel that my understanding of this stuff is good enough to be sure that
this is the optimal solution..

Btw, in syncprov_checkpoint() there is a similar SLAP_GLUE_SUBORDINATE
test, should that have included an overlay_is_inst() clause as well?

Rein Tollevik
Basefarm AS

Index: OpenLDAP/servers/slapd/ctxcsn.c
diff -u OpenLDAP/servers/slapd/ctxcsn.c:1.1.1.11
OpenLDAP/servers/slapd/ctxcsn.c:1.2
--- OpenLDAP/servers/slapd/ctxcsn.c:1.1.1.11	Sat Mar 22 16:47:49 2008
+++ OpenLDAP/servers/slapd/ctxcsn.c	Wed Apr 30 19:44:25 2008
@@ -37,67 +37,78 @@
 )
 {
 	struct slap_csn_entry *csne, *committed_csne = NULL;
+	Backend *be = op->o_bd;
 
 	if ( maxcsn ) {
 		BER_BVZERO( maxcsn );
 	}
 
-	ldap_pvt_thread_mutex_lock( op->o_bd->be_pcl_mutexp );
+	if ( SLAP_GLUE_SUBORDINATE( be ) && !overlay_is_inst( be, "syncprov" ))
+		be = select_backend( &be->be_nsuffix[0], 1 );
+	ldap_pvt_thread_mutex_lock( be->be_pcl_mutexp );
 
-	LDAP_TAILQ_FOREACH( csne, op->o_bd->be_pending_csn_list, ce_csn_link ) {
+	LDAP_TAILQ_FOREACH( csne, be->be_pending_csn_list, ce_csn_link ) {
 		if ( csne->ce_opid == op->o_opid && csne->ce_connid == op->o_connid ) {
 			csne->ce_state = SLAP_CSN_COMMIT;
 			break;
 		}
 	}
 
-	LDAP_TAILQ_FOREACH( csne, op->o_bd->be_pending_csn_list, ce_csn_link ) {
+	LDAP_TAILQ_FOREACH( csne, be->be_pending_csn_list, ce_csn_link ) {
 		if ( csne->ce_state == SLAP_CSN_COMMIT ) committed_csne = csne;
 		if ( csne->ce_state == SLAP_CSN_PENDING ) break;
 	}
 
 	if ( committed_csne && maxcsn ) *maxcsn = committed_csne->ce_csn;
-	ldap_pvt_thread_mutex_unlock( op->o_bd->be_pcl_mutexp );
+	ldap_pvt_thread_mutex_unlock( be->be_pcl_mutexp );
 }
 
 void
 slap_rewind_commit_csn( Operation *op )
 {
 	struct slap_csn_entry *csne;
+	Backend *be = op->o_bd;
 
-	ldap_pvt_thread_mutex_lock( op->o_bd->be_pcl_mutexp );
+	if ( SLAP_GLUE_SUBORDINATE( be ) && !overlay_is_inst( be, "syncprov" ))
+		be = select_backend( &be->be_nsuffix[0], 1 );
+	ldap_pvt_thread_mutex_lock( be->be_pcl_mutexp );
 
-	LDAP_TAILQ_FOREACH( csne, op->o_bd->be_pending_csn_list, ce_csn_link ) {
+	LDAP_TAILQ_FOREACH( csne, be->be_pending_csn_list, ce_csn_link ) {
 		if ( csne->ce_opid == op->o_opid && csne->ce_connid == op->o_connid ) {
 			csne->ce_state = SLAP_CSN_PENDING;
 			break;
 		}
 	}
 
-	ldap_pvt_thread_mutex_unlock( op->o_bd->be_pcl_mutexp );
+	ldap_pvt_thread_mutex_unlock( be->be_pcl_mutexp );
 }
 
 void
 slap_graduate_commit_csn( Operation *op )
 {
 	struct slap_csn_entry *csne;
+	Backend *be;
 
 	if ( op == NULL ) return;
 	if ( op->o_bd == NULL ) return;
 
+	be = op->o_bd;
+	if ( SLAP_GLUE_SUBORDINATE( be ) && !overlay_is_inst( be, "syncprov" ))
+		be = select_backend( &be->be_nsuffix[0], 1 );
+
 #if 0
 	/* it is NULL when we get here from the frontendDB;
 	 * alternate fix: initialize frontendDB like all other backends */
-	assert( op->o_bd->be_pcl_mutexp != NULL );
+	assert( be->be_pcl_mutexp != NULL );
 #endif
 	
-	if ( op->o_bd->be_pcl_mutexp == NULL ) return;
+	if ( be->be_pcl_mutexp == NULL ) return;
 
-	ldap_pvt_thread_mutex_lock( op->o_bd->be_pcl_mutexp );
+	ldap_pvt_thread_mutex_lock( be->be_pcl_mutexp );
 
-	LDAP_TAILQ_FOREACH( csne, op->o_bd->be_pending_csn_list, ce_csn_link ) {
+	LDAP_TAILQ_FOREACH( csne, be->be_pending_csn_list, ce_csn_link ) {
 		if ( csne->ce_opid == op->o_opid && csne->ce_connid == op->o_connid ) {
-			LDAP_TAILQ_REMOVE( op->o_bd->be_pending_csn_list,
+			LDAP_TAILQ_REMOVE( be->be_pending_csn_list,
 				csne, ce_csn_link );
 			Debug( LDAP_DEBUG_SYNC, "slap_graduate_commit_csn: removing %p %s\n",
 				csne->ce_csn.bv_val, csne->ce_csn.bv_val, 0 );
@@ -110,7 +121,7 @@
 		}
 	}
 
-	ldap_pvt_thread_mutex_unlock( op->o_bd->be_pcl_mutexp );
+	ldap_pvt_thread_mutex_unlock( be->be_pcl_mutexp );
 
 	return;
 }
@@ -161,22 +172,25 @@
 	struct berval *csn )
 {
 	struct slap_csn_entry *pending;
+	Backend *be = op->o_bd;
 
 	pending = (struct slap_csn_entry *) ch_calloc( 1,
 			sizeof( struct slap_csn_entry ));
 
 	Debug( LDAP_DEBUG_SYNC, "slap_queue_csn: queing %p %s\n", csn->bv_val,
csn->bv_val, 0 );
 
-	ldap_pvt_thread_mutex_lock( op->o_bd->be_pcl_mutexp );
+	if ( SLAP_GLUE_SUBORDINATE( be ) && !overlay_is_inst( be, "syncprov" ))
+		be = select_backend( &be->be_nsuffix[0], 1 );
+	ldap_pvt_thread_mutex_lock( be->be_pcl_mutexp );
 
 	ber_dupbv( &pending->ce_csn, csn );
 	ber_bvreplace_x( &op->o_csn, &pending->ce_csn, op->o_tmpmemctx );
 	pending->ce_connid = op->o_connid;
 	pending->ce_opid = op->o_opid;
 	pending->ce_state = SLAP_CSN_PENDING;
-	LDAP_TAILQ_INSERT_TAIL( op->o_bd->be_pending_csn_list,
+	LDAP_TAILQ_INSERT_TAIL( be->be_pending_csn_list,
 		pending, ce_csn_link );
-	ldap_pvt_thread_mutex_unlock( op->o_bd->be_pcl_mutexp );
+	ldap_pvt_thread_mutex_unlock( be->be_pcl_mutexp );
 }
 
 int