netmaker/queue/logic.go

package queue

import (
	"time"

	"github.com/gravitl/netmaker/logger"
	"github.com/gravitl/netmaker/logic"
	"github.com/gravitl/netmaker/models"
	"github.com/gravitl/netmaker/netclient/ncutils"
)

func updateNodeMetrics(currentNode *models.Node, newMetrics *models.Metrics) bool {
	if newMetrics.FailoverPeers == nil {
		newMetrics.FailoverPeers = make(map[string]string)
	}
	oldMetrics, err := logic.GetMetrics(currentNode.ID.String())
	if err != nil {
		logger.Log(1, "error finding old metrics for node", currentNode.ID.String())
		return false
	}
	if oldMetrics.FailoverPeers == nil {
		oldMetrics.FailoverPeers = make(map[string]string)
	}

	var attachedClients []models.ExtClient
	if currentNode.IsIngressGateway {
		clients, err := logic.GetExtClientsByID(currentNode.ID.String(), currentNode.Network)
		if err == nil {
			attachedClients = clients
		}
	}
	if len(attachedClients) > 0 {
		// associate ext clients with IDs
		for i := range attachedClients {
			extMetric := newMetrics.Connectivity[attachedClients[i].PublicKey]
			if len(extMetric.NodeName) == 0 &&
				len(newMetrics.Connectivity[attachedClients[i].ClientID].NodeName) > 0 { // cover server clients
				extMetric = newMetrics.Connectivity[attachedClients[i].ClientID]
				if extMetric.TotalReceived > 0 && extMetric.TotalSent > 0 {
					extMetric.Connected = true
				}
			}
			extMetric.NodeName = attachedClients[i].ClientID
			delete(newMetrics.Connectivity, attachedClients[i].PublicKey)
			newMetrics.Connectivity[attachedClients[i].ClientID] = extMetric
		}
	}

	// run through metrics for each peer
	for k := range newMetrics.Connectivity {
		currMetric := newMetrics.Connectivity[k]
		oldMetric := oldMetrics.Connectivity[k]
		currMetric.TotalTime += oldMetric.TotalTime
		currMetric.Uptime += oldMetric.Uptime // get the total uptime for this connection
		if currMetric.Uptime == 0 || currMetric.TotalTime == 0 {
			currMetric.PercentUp = 0
		} else {
			currMetric.PercentUp = 100.0 * (float64(currMetric.Uptime) / float64(currMetric.TotalTime))
		}
		totalUpMinutes := currMetric.Uptime * ncutils.CheckInInterval
		currMetric.ActualUptime = time.Duration(totalUpMinutes) * time.Minute
		delete(oldMetrics.Connectivity, k) // remove from old data
		newMetrics.Connectivity[k] = currMetric

	}

	// add nodes that need failover
	nodes, err := logic.GetNetworkNodes(currentNode.Network)
	if err != nil {
		logger.Log(0, "failed to retrieve nodes while updating metrics")
		return false
	}
	for _, node := range nodes {
		if !newMetrics.Connectivity[node.ID.String()].Connected &&
			len(newMetrics.Connectivity[node.ID.String()].NodeName) > 0 &&
			node.Connected &&
			len(node.FailoverNode) > 0 &&
			!node.Failover {
			newMetrics.FailoverPeers[node.ID.String()] = node.FailoverNode.String()
		}
	}
	shouldUpdate := len(oldMetrics.FailoverPeers) == 0 && len(newMetrics.FailoverPeers) > 0
	for k, v := range oldMetrics.FailoverPeers {
		if len(newMetrics.FailoverPeers[k]) > 0 && len(v) == 0 {
			shouldUpdate = true
		}

		if len(v) > 0 && len(newMetrics.FailoverPeers[k]) == 0 {
			newMetrics.FailoverPeers[k] = v
		}
	}

	for k := range oldMetrics.Connectivity { // cleanup any left over data, self healing
		delete(newMetrics.Connectivity, k)
	}
	return shouldUpdate
}