coordinator: Fix error to add route when the nic is down

cmd/coordinator/cmd/command_add.go

@@ -120,6 +120,12 @@ func CmdAdd(args *skel.CmdArgs) (err error) {
 	}
 	logger.Sugar().Debugf("Get current host netns: %v", c.hostNs.Path())
 
+	// checking if the nic is in up state
+	if err = c.checkNICState(args.IfName); err != nil {
+		logger.Error("error to check pod's nic state", zap.Error(err))
+		return fmt.Errorf("error to check pod's nic %s state: %v", args.Args, err)
+	}
+
 	// check if it's first time invoke
 	err = c.coordinatorModeAndFirstInvoke(logger, conf.PodDefaultCniNic)
 	if err != nil {
@@ -299,10 +305,10 @@ func CmdAdd(args *skel.CmdArgs) (err error) {
 		return err
 	}
 
-	c.currentRuleTable = c.mustGetRuleNumber(c.podNics)
-	if c.currentRuleTable < 0 {
-		logger.Error("coordinator must be working with spiderpool: no spiderendpoint records found", zap.Strings("spiderNics", c.podNics))
-		return fmt.Errorf("coordinator must be working with spiderpool: no spiderendpoint records found")
+	c.currentRuleTable, err = c.mustGetRuleNumber()
+	if err != nil {
+		logger.Error("error to get rule number for the nic", zap.Error(err))
+		return fmt.Errorf("error to get rule number for the nic %s: %w", args.IfName, err)
 	}
 	logger.Debug("Get currentRuleTable", zap.Int("ruleTable", c.currentRuleTable))
 

cmd/coordinator/cmd/utils.go

@@ -116,21 +116,41 @@ func (c *coordinator) coordinatorModeAndFirstInvoke(logger *zap.Logger, podFirst
 	return fmt.Errorf("unknown tuneMode: %s", c.tuneMode)
 }
 
+func (c *coordinator) checkNICState(iface string) error {
+	return c.netns.Do(func(netNS ns.NetNS) error {
+		link, err := netlink.LinkByName(iface)
+		if err != nil {
+			return err
+		}
+
+		if link.Attrs().Flags != net.FlagUp {
+			return netlink.LinkSetUp(link)
+		}
+		return nil
+	})
+}
+
 // getRuleNumber return the number of rule table corresponding to the previous interface from the given interface.
-// for example:
-// input: net1, output: 100(eth0)
-// input: net2, output: 101(net1)
-func (c *coordinator) mustGetRuleNumber(spiderNics []string) int {
-	if len(spiderNics) == 0 {
-		return -1
-	}
+func (c *coordinator) mustGetRuleNumber() (int, error) {
+	ruleTable := -1
+	err := c.netns.Do(func(netNS ns.NetNS) error {
+		links, err := netlink.LinkList()
+		if err != nil {
+			return fmt.Errorf("error to get link list: %w", err)
+		}
 
-	if c.currentInterface == defaultOverlayVethName {
-		return unix.RT_TABLE_MAIN
-	} else if spiderNics[0] == c.currentInterface {
-		return defaultPodRuleTable
+		// start from 100, and exclude lookback and eth0
+		// example:
+		// lo, eth0, net1 -> 100
+		// lo, eth0, net1, net2 -> 101
+		ruleTable = 99 + len(links) - 2
+		return nil
+	})
+
+	if err != nil {
+		return -1, err
 	}
-	return defaultPodRuleTable + len(spiderNics) - 1
+	return ruleTable, nil
 }
 
 // setupVeth sets up a pair of virtual ethernet devices. move one to the host and other

pkg/networking/networking/route.go

@@ -249,6 +249,7 @@ func moveRouteTable(linkIndex, srcRuleTable, dstRuleTable int, onlyCopyOverlayDe
 				return fmt.Errorf("failed to RouteDel %s in main table: %+v", route.String(), err)
 			}
 			logger.Debug("Del the default route from main successfully", zap.String("Route", route.String()))
+			return nil
 		}
 
 		if onlyCopyOverlayDefaultRoute {
@@ -257,10 +258,11 @@ func moveRouteTable(linkIndex, srcRuleTable, dstRuleTable int, onlyCopyOverlayDe
 		}
 
 		// we need copy the all routes in main table of the podDefaultRouteNic to dstRuleTable.
-		// Otherwise, the reply packet don't know
+		// Otherwise, we don't know how to forward the packet send from the nic
+		route.Table = dstRuleTable
 		if err = netlink.RouteAdd(&route); err != nil && !os.IsExist(err) {
-			logger.Error("failed to RouteAdd in new table ", zap.String("route", route.String()), zap.Error(err))
-			return fmt.Errorf("failed to RouteAdd (%+v) to new table: %+v", route, err)
+			logger.Error("failed to add the route table", zap.String("route", route.String()), zap.Error(err))
+			// return fmt.Errorf("failed to add the route table (%+v): %+v", route, err)
 		}
 		logger.Debug("MoveRoute to new table successfully", zap.String("Route", route.String()))
 		return nil