ovn-controller: incremental processing for multicast group changes

This patch handles SB Multicast_Group table changes incrementally.
The Multicast_Group table changes can be triggered by creating/deleting
a lport of a lswitch. It can be also triggered indirectly by an
updating of a port-binding which is referenced by the multicast
group.

This patch together with previous incremental processing engine
related changes supports incremental processing for lflow changes
and port-binding changes of lports on other HVs, which are the most
common scenarios in a cloud where workloads come up and down.

In ovn-scale-test env [1], the total execution time of creating and
binding 10k ports on 1k HVs with 40 lswitches and 8 lrouters
(5 lswitches/lrouter), decreased from 3h40m to 1h50m because of the
less CPU on HVs. The CPU time of ovn-controller for additional 500
lports creating and binding (on top of already existed 10k lports)
decreased 90% comparing with master.

Latency for end-to-end operations of one extra port on top of the
10k lports, start from port-creation until all flows installation
on all related HVs is also improved significantly:

before: 20.6s in total
- lsp-add: 0.4s
- wait-until port up=true: 4.8s
- --wait=hv sync: 15.4s

after: 7.3s in total
- lsp-add: 0.4s
- wait-until port up=true: 4.0s
- --wait=hv sync: 2.9s

[1] https://github.com/openvswitch/ovn-scale-test

Signed-off-by: Han Zhou <[email protected]>
Signed-off-by: Ben Pfaff <[email protected]>

ovn/controller/ovn-controller.c

@@ -1238,6 +1238,56 @@ flow_output_sb_port_binding_handler(struct engine_node *node)
     return true;
 }
 
+static bool
+flow_output_sb_multicast_group_handler(struct engine_node *node)
+{
+    struct ed_type_runtime_data *data =
+        (struct ed_type_runtime_data *)engine_get_input(
+                "runtime_data", node)->data;
+    struct hmap *local_datapaths = &data->local_datapaths;
+    struct simap *ct_zones = &data->ct_zones;
+
+    struct ed_type_mff_ovn_geneve *ed_mff_ovn_geneve =
+        (struct ed_type_mff_ovn_geneve *)engine_get_input(
+            "mff_ovn_geneve", node)->data;
+    enum mf_field_id mff_ovn_geneve = ed_mff_ovn_geneve->mff_ovn_geneve;
+
+    struct ovsrec_open_vswitch_table *ovs_table =
+        (struct ovsrec_open_vswitch_table *)EN_OVSDB_GET(
+            engine_get_input("OVS_open_vswitch", node));
+    struct ovsrec_bridge_table *bridge_table =
+        (struct ovsrec_bridge_table *)EN_OVSDB_GET(
+            engine_get_input("OVS_bridge", node));
+    const struct ovsrec_bridge *br_int = get_br_int(bridge_table, ovs_table);
+    const char *chassis_id = get_chassis_id(ovs_table);
+
+    struct ovsdb_idl_index *sbrec_chassis_by_name =
+        engine_ovsdb_node_get_index(
+                engine_get_input("SB_chassis", node),
+                "name");
+    const struct sbrec_chassis *chassis = NULL;
+    if (chassis_id) {
+        chassis = chassis_lookup_by_name(sbrec_chassis_by_name, chassis_id);
+    }
+    ovs_assert(br_int && chassis);
+
+    struct ed_type_flow_output *fo =
+        (struct ed_type_flow_output *)node->data;
+    struct ovn_desired_flow_table *flow_table = &fo->flow_table;
+
+    struct sbrec_multicast_group_table *multicast_group_table =
+        (struct sbrec_multicast_group_table *)EN_OVSDB_GET(
+            engine_get_input("SB_multicast_group", node));
+
+    physical_handle_mc_group_changes(multicast_group_table,
+            mff_ovn_geneve, chassis, ct_zones, local_datapaths,
+            flow_table);
+
+    node->changed = true;
+    return true;
+
+}
+
 struct ovn_controller_exit_args {
     bool *exiting;
     bool *restart;
@@ -1356,7 +1406,8 @@ main(int argc, char *argv[])
 
     engine_add_input(&en_flow_output, &en_sb_chassis, NULL);
     engine_add_input(&en_flow_output, &en_sb_encap, NULL);
-    engine_add_input(&en_flow_output, &en_sb_multicast_group, NULL);
+    engine_add_input(&en_flow_output, &en_sb_multicast_group,
+                     flow_output_sb_multicast_group_handler);
     engine_add_input(&en_flow_output, &en_sb_port_binding,
                      flow_output_sb_port_binding_handler);
     engine_add_input(&en_flow_output, &en_sb_mac_binding, NULL);

ovn/controller/physical.c

@@ -1009,6 +1009,29 @@ void physical_handle_port_binding_changes(
 
 }
 
+void
+physical_handle_mc_group_changes(
+        const struct sbrec_multicast_group_table *multicast_group_table,
+        enum mf_field_id mff_ovn_geneve,
+        const struct sbrec_chassis *chassis,
+        const struct simap *ct_zones,
+        const struct hmap *local_datapaths,
+        struct ovn_desired_flow_table *flow_table)
+{
+    const struct sbrec_multicast_group *mc;
+    SBREC_MULTICAST_GROUP_TABLE_FOR_EACH_TRACKED (mc, multicast_group_table) {
+        if (sbrec_multicast_group_is_deleted(mc)) {
+            ofctrl_remove_flows(flow_table, &mc->header_.uuid);
+        } else {
+            if (!sbrec_multicast_group_is_new(mc)) {
+                ofctrl_remove_flows(flow_table, &mc->header_.uuid);
+            }
+            consider_mc_group(mff_ovn_geneve, ct_zones, local_datapaths,
+                              chassis, mc, flow_table);
+        }
+    }
+}
+
 void
 physical_run(struct ovsdb_idl_index *sbrec_port_binding_by_name,
              const struct sbrec_multicast_group_table *multicast_group_table,

ovn/controller/physical.h

@@ -64,4 +64,11 @@ void physical_handle_port_binding_changes(
         struct sset *active_tunnels,
         struct ovn_desired_flow_table *);
 
+void physical_handle_mc_group_changes(
+        const struct sbrec_multicast_group_table *,
+        enum mf_field_id mff_ovn_geneve,
+        const struct sbrec_chassis *,
+        const struct simap *ct_zones,
+        const struct hmap *local_datapaths,
+        struct ovn_desired_flow_table *);
 #endif /* ovn/physical.h */