Kubernetes节点之间的ping监控怎么实现

发布时间：2021-11-02 16:32:07 阅读：168 作者：小新栏目：web开发

开发者测试专用服务器限时活动，0元免费领，库存有限，领完即止！点击查看>>

小编给大家分享一下Kubernetes节点之间的ping监控怎么实现，希望大家阅读完这篇文章之后都有所收获，下面让我们一起去探讨吧！

脚本和配置

我们解决方案的主要组件是一个脚本，该脚本监视每个节点的.status.addresses值。如果某个节点的该值已更改(例如添加了新节点)，则我们的脚本使用Helm value方式将节点列表以ConfigMap的形式传递给Helm图表：

apiVersion: v1 kind: ConfigMap metadata: name: ping-exporter-config namespace: d8-system data: nodes.json: > {{ .Values.pingExporter.targets | toJson }}    .Values.pingExporter.targets类似以下：  "cluster_targets":[{"ipAddress":"192.168.191.11","name":"kube-a-3"},{"ipAddress":"192.168.191.12","name":"kube-a-2"},{"ipAddress":"192.168.191.22","name":"kube-a-1"},{"ipAddress":"192.168.191.23","name":"kube-db-1"},{"ipAddress":"192.168.191.9","name":"kube-db-2"},{"ipAddress":"51.75.130.47","name":"kube-a-4"}],"external_targets":[{"host":"8.8.8.8","name":"google-dns"},{"host":"youtube.com"}]}

下面是Python脚本：

#!/usr/bin/env python3 import subprocess import prometheus_client import re import statistics import os import json import glob import better_exchook import datetime better_exchook.install() FPING_CMDLINE = "/usr/sbin/fping -p 1000 -C 30 -B 1 -q -r 1".split(" ") FPING_REGEX = re.compile(r"^(\S*)\s*: (.*)$", re.MULTILINE) CONFIG_PATH = "/config/targets.json" registry = prometheus_client.CollectorRegistry() prometheus_exceptions_counter = \ prometheus_client.Counter('kube_node_ping_exceptions', 'Total number of exceptions', [], registry=registry) prom_metrics_cluster = {"sent": prometheus_client.Counter('kube_node_ping_packets_sent_total',                                               'ICMP packets sent',                                               ['destination_node', 'destination_node_ip_address'],                                               registry=registry),             "received": prometheus_client.Counter('kube_node_ping_packets_received_total',                                                   'ICMP packets received',                                                  ['destination_node', 'destination_node_ip_address'],                                                  registry=registry),             "rtt": prometheus_client.Counter('kube_node_ping_rtt_milliseconds_total',                                              'round-trip time',                                             ['destination_node', 'destination_node_ip_address'],                                             registry=registry),             "min": prometheus_client.Gauge('kube_node_ping_rtt_min', 'minimum round-trip time',                                            ['destination_node', 'destination_node_ip_address'],                                            registry=registry),             "max": prometheus_client.Gauge('kube_node_ping_rtt_max', 'maximum round-trip time',                                            ['destination_node', 'destination_node_ip_address'],                                            registry=registry),             "mdev": prometheus_client.Gauge('kube_node_ping_rtt_mdev',                                             'mean deviation of round-trip times',                                             ['destination_node', 'destination_node_ip_address'],                                             registry=registry)} prom_metrics_external = {"sent": prometheus_client.Counter('external_ping_packets_sent_total',                                               'ICMP packets sent',                                               ['destination_name', 'destination_host'],                                               registry=registry),             "received": prometheus_client.Counter('external_ping_packets_received_total',                                                   'ICMP packets received',                                                  ['destination_name', 'destination_host'],                                                  registry=registry),             "rtt": prometheus_client.Counter('external_ping_rtt_milliseconds_total',                                              'round-trip time',                                             ['destination_name', 'destination_host'],                                             registry=registry),             "min": prometheus_client.Gauge('external_ping_rtt_min', 'minimum round-trip time',                                            ['destination_name', 'destination_host'],                                            registry=registry),             "max": prometheus_client.Gauge('external_ping_rtt_max', 'maximum round-trip time',                                            ['destination_name', 'destination_host'],                                            registry=registry),             "mdev": prometheus_client.Gauge('external_ping_rtt_mdev',                                             'mean deviation of round-trip times',                                             ['destination_name', 'destination_host'],                                             registry=registry)} def validate_envs(): envs = {"MY_NODE_NAME": os.getenv("MY_NODE_NAME"), "PROMETHEUS_TEXTFILE_DIR": os.getenv("PROMETHEUS_TEXTFILE_DIR"),         "PROMETHEUS_TEXTFILE_PREFIX": os.getenv("PROMETHEUS_TEXTFILE_PREFIX")} for k, v in envs.items():     if not v:         raise ValueError("{} environment variable is empty".format(k)) return envs @prometheus_exceptions_counter.count_exceptions() def compute_results(results): computed = {} matches = FPING_REGEX.finditer(results) for match in matches:     host = match.group(1)     ping_results = match.group(2)     if "duplicate" in ping_results:         continue     splitted = ping_results.split(" ")     if len(splitted) != 30:         raise ValueError("ping returned wrong number of results: \"{}\"".format(splitted))     positive_results = [float(x) for x in splitted if x != "-"]     if len(positive_results) > 0:         computed[host] = {"sent": 30, "received": len(positive_results),                         "rtt": sum(positive_results),                         "max": max(positive_results), "min": min(positive_results),                         "mdev": statistics.pstdev(positive_results)}     else:         computed[host] = {"sent": 30, "received": len(positive_results), "rtt": 0,                         "max": 0, "min": 0, "mdev": 0} if not len(computed):     raise ValueError("regex match\"{}\" found nothing in fping output \"{}\"".format(FPING_REGEX, results)) return computed @prometheus_exceptions_counter.count_exceptions() def call_fping(ips): cmdline = FPING_CMDLINE + ips process = subprocess.run(cmdline, stdout=subprocess.PIPE,                          stderr=subprocess.STDOUT, universal_newlines=True) if process.returncode == 3:     raise ValueError("invalid arguments: {}".format(cmdline)) if process.returncode == 4:     raise OSError("fping reported syscall error: {}".format(process.stderr)) return process.stdout envs = validate_envs() files = glob.glob(envs["PROMETHEUS_TEXTFILE_DIR"] + "*") for f in files: os.remove(f) labeled_prom_metrics = {"cluster_targets": [], "external_targets": []} while True: with open(CONFIG_PATH, "r") as f:     config = json.loads(f.read())     config["external_targets"] = [] if config["external_targets"] is None else config["external_targets"]     for target in config["external_targets"]:         target["name"] = target["host"] if "name" not in target.keys() else target["name"] if labeled_prom_metrics["cluster_targets"]:     for metric in labeled_prom_metrics["cluster_targets"]:         if (metric["node_name"], metric["ip"]) not in [(node["name"], node["ipAddress"]) for node in config['cluster_targets']]:             for k, v in prom_metrics_cluster.items():                 v.remove(metric["node_name"], metric["ip"]) if labeled_prom_metrics["external_targets"]:     for metric in labeled_prom_metrics["external_targets"]:         if (metric["target_name"], metric["host"]) not in [(target["name"], target["host"]) for target in config['external_targets']]:             for k, v in prom_metrics_external.items():                 v.remove(metric["target_name"], metric["host"]) labeled_prom_metrics = {"cluster_targets": [], "external_targets": []} for node in config["cluster_targets"]:     metrics = {"node_name": node["name"], "ip": node["ipAddress"], "prom_metrics": {}}     for k, v in prom_metrics_cluster.items():         metrics["prom_metrics"][k] = v.labels(node["name"], node["ipAddress"])     labeled_prom_metrics["cluster_targets"].append(metrics) for target in config["external_targets"]:     metrics = {"target_name": target["name"], "host": target["host"], "prom_metrics": {}}     for k, v in prom_metrics_external.items():         metrics["prom_metrics"][k] = v.labels(target["name"], target["host"])     labeled_prom_metrics["external_targets"].append(metrics) out = call_fping([prom_metric["ip"]   for prom_metric in labeled_prom_metrics["cluster_targets"]] + \                  [prom_metric["host"] for prom_metric in labeled_prom_metrics["external_targets"]]) computed = compute_results(out) for dimension in labeled_prom_metrics["cluster_targets"]:     result = computed[dimension["ip"]]     dimension["prom_metrics"]["sent"].inc(computed[dimension["ip"]]["sent"])     dimension["prom_metrics"]["received"].inc(computed[dimension["ip"]]["received"])     dimension["prom_metrics"]["rtt"].inc(computed[dimension["ip"]]["rtt"])     dimension["prom_metrics"]["min"].set(computed[dimension["ip"]]["min"])     dimension["prom_metrics"]["max"].set(computed[dimension["ip"]]["max"])     dimension["prom_metrics"]["mdev"].set(computed[dimension["ip"]]["mdev"]) for dimension in labeled_prom_metrics["external_targets"]:     result = computed[dimension["host"]]     dimension["prom_metrics"]["sent"].inc(computed[dimension["host"]]["sent"])     dimension["prom_metrics"]["received"].inc(computed[dimension["host"]]["received"])     dimension["prom_metrics"]["rtt"].inc(computed[dimension["host"]]["rtt"])     dimension["prom_metrics"]["min"].set(computed[dimension["host"]]["min"])     dimension["prom_metrics"]["max"].set(computed[dimension["host"]]["max"])     dimension["prom_metrics"]["mdev"].set(computed[dimension["host"]]["mdev"]) prometheus_client.write_to_textfile(     envs["PROMETHEUS_TEXTFILE_DIR"] + envs["PROMETHEUS_TEXTFILE_PREFIX"] + envs["MY_NODE_NAME"] + ".prom", registry)AI代码助手复制代码

该脚本在每个Kubernetes节点上运行，并且每秒两次发送ICMP数据包到Kubernetes集群的所有实例。收集的结果会存储在文本文件中。

该脚本会包含在Docker镜像中：

FROM python:3.6-alpine3.8 COPY rootfs / WORKDIR /app RUN pip3 install --upgrade pip && pip3 install -r requirements.txt && apk add --no-cache fping ENTRYPOINT ["python3", "/app/ping-exporter.py"]

另外，我们还创建了一个ServiceAccount和一个具有唯一权限的对应角色用于获取节点列表(这样我们就可以知道它们的IP地址)：

apiVersion: v1 kind: ServiceAccount metadata: name: ping-exporter namespace: d8-system --- kind: ClusterRole apiVersion: rbac.authorization.k8s.io/v1 metadata: name: d8-system:ping-exporter rules: - apiGroups: [""] resources: ["nodes"] verbs: ["list"] --- kind: ClusterRoleBinding apiVersion: rbac.authorization.k8s.io/v1 metadata: name: d8-system:kube-ping-exporter subjects: - kind: ServiceAccount name: ping-exporter namespace: d8-system roleRef: apiGroup: rbac.authorization.k8s.io kind: ClusterRole name: d8-system:ping-exporter

最后，我们需要DaemonSet来运行在集群中的所有实例：

apiVersion: apps/v1 kind: DaemonSet metadata: name: ping-exporter namespace: d8-system spec: updateStrategy: type: RollingUpdate selector: matchLabels:   name: ping-exporter template: metadata:   labels:     name: ping-exporter spec:   terminationGracePeriodSeconds: 0   tolerations:   - operator: "Exists"   hostNetwork: true   serviceAccountName: ping-exporter   priorityClassName: cluster-low   containers:   - image: private-registry.flant.com/ping-exporter/ping-exporter:v1     name: ping-exporter     env:       - name: MY_NODE_NAME         valueFrom:           fieldRef:             fieldPath: spec.nodeName       - name: PROMETHEUS_TEXTFILE_DIR         value: /node-exporter-textfile/       - name: PROMETHEUS_TEXTFILE_PREFIX         value: ping-exporter_     volumeMounts:       - name: textfile         mountPath: /node-exporter-textfile       - name: config         mountPath: /config   volumes:     - name: textfile       hostPath:         path: /var/run/node-exporter-textfile     - name: config       configMap:         name: ping-exporter-config   imagePullSecrets:   - name: private-registry

该解决方案的最后操作细节是：

Python脚本执行时，其结果(即存储在主机上/var/run/node-exporter-textfile目录中的文本文件)将传递到DaemonSet类型的node-exporter。
node-exporter使用--collector.textfile.directory /host/textfile参数启动，这里的/host/textfile是hostPath目录/var/run/node-exporter-textfile。(你可以点击这里了解关于node-exporter中文本文件收集器的更多信息。)
最后node-exporter读取这些文件，然后Prometheus从node-exporter实例上收集所有数据。

那么结果如何?

现在该来享受期待已久的结果了。指标创建之后，我们可以使用它们，当然也可以对其进行可视化。以下可以看到它们是怎样的。

首先，有一个通用选择器可让我们在其中选择节点以检查其“源”和“目标”连接。你可以获得一个汇总表，用于在Grafana仪表板中指定的时间段内ping选定节点的结果：

Kubernetes节点之间的ping监控怎么实现