Istio - 安装
参考资料
以 Istio 1.5 版本为例，介绍整个安装过程：
- 直接在 官网下载对于版本：https://github.com/istio/istio/releases/tag/1.5.0 。
[root@iZuf61kbf845xt6tz10abgZ ~]# cd /home/istio/
[root@iZuf61kbf845xt6tz10abgZ istio]# tar -zxvf istio-1.5.0-linux.tar.gz
- 设置环境变量。
vi /etc/profile
加入 export PATH=$PATH:/home/istio/istio-1.5.0/bin
source /etc/profile
- 安装目录介绍：
| 目录 | 包含内容 |
|---|---|
bin | 包含 istioctl 的客户端文件 |
install | 包含 Consul、GCP 和 Kubernetes 平台的 Istio 安装脚本和文件 |
samples | 包含示例应用程序 |
tools | 包含用于性能测试和在本地机器上进行测试的脚本 |
安装过程
尝试集群下安装
执行安装命令：
[root@iZuf61kbf845xt6tz10abgZ ~]# istioctl manifest apply --set profile=demo
结果：
[root@iZuf61kbf845xt6tz10abgZ ~]# istioctl manifest apply --set profile=demo
Detected that your cluster does not support third party JWT authentication. Falling back to less secure first party JWT. See https://istio.io/docs/ops/best-practices/security/#configure-third-party-service-account-tokens for details.
- Applying manifest for component Base...
✔ Finished applying manifest for component Base.
- Applying manifest for component Pilot...
✔ Finished applying manifest for component Pilot.
- Applying manifest for component IngressGateways...
- Applying manifest for component EgressGateways...
- Applying manifest for component AddonComponents...
✔ Finished applying manifest for component EgressGateways.
2022-04-13T23:31:06.344536Z error installer error running kubectl: exit status 1
✘ Finished applying manifest for component AddonComponents.
2022-04-13T23:31:34.877288Z error installer error running kubectl: exit status 1
✘ Finished applying manifest for component IngressGateways.
失败问题，找https://github.com/istio/istio/ 中issues也无法定位，看错误日志是无法连同从节点的10250端口：
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal Scheduled 4m26s default-scheduler Successfully assigned istio-system/istio-tracing-c7b59f68f-dzrz6 to vm-16-6-centos
Normal Killing 3m16s (x2 over 3m56s) kubelet, vm-16-6-centos Container jaeger failed liveness probe, will be restarted
Warning Unhealthy 3m13s (x7 over 4m23s) kubelet, vm-16-6-centos Readiness probe failed: HTTP probe failed with statuscode: 503
Normal Pulled 3m3s (x3 over 4m25s) kubelet, vm-16-6-centos Container image "docker.io/jaegertracing/all-in-one:1.16" already present on machine
Normal Created 3m3s (x3 over 4m25s) kubelet, vm-16-6-centos Created container jaeger
Normal Started 3m3s (x3 over 4m24s) kubelet, vm-16-6-centos Started container jaeger
Warning Unhealthy 2m56s (x7 over 4m16s) kubelet, vm-16-6-centos Liveness probe failed: HTTP probe failed with statuscode: 503
[root@iZuf61kbf845xt6tz10abgZ ~]# kubectl logs -f --tail 200 istio-tracing-c7b59f68f-dzrz6 -n istio-system
Error from server: Get https://10.0.16.6:10250/containerLogs/istio-system/istio-tracing-c7b59f68f-dzrz6/jaeger?follow=true&tailLines=200: dial tcp 10.0.16.6:10250: i/o timeout
在master下安装
在网上查找其他方案，发现可以在不加入node的情况下，去安装istio
- 清理从节点，调整mater节点能部署pod
kubectl delete node vm-16-6-centos
kubectl get no -o yaml | grep taint -A 5
kubectl taint nodes --all node-role.kubernetes.io/master
istioctl manifest apply --set profile=demo
- 执行后发现容器正常一段时间后又异常了,查看污点和容忍度，发现虚机的硬盘不足：
[root@iZuf61kbf845xt6tz10abgZ ~]# kubectl get no -o yaml | grep taint -A 5
taints:
- effect: NoSchedule
key: node.kubernetes.io/disk-pressure
timeAdded: "2022-04-16T00:56:21Z"
status:
addresses:
# 去掉污点
kubectl taint nodes --all node.kubernetes.io/disk-pressure:NoSchedule-
- 清理后，重新执行就正常了：
[root@iZuf61kbf845xt6tz10abgZ ~]# kubectl get pods -n istio-system
NAME READY STATUS RESTARTS AGE
grafana-78bc994d79-p5g42 1/1 Running 0 9m56s
istio-egressgateway-7c9f7d5bd6-h5swx 1/1 Running 0 10m
istio-ingressgateway-f9b47d445-n7crd 1/1 Running 0 10m
istio-tracing-c7b59f68f-kbfsk 1/1 Running 0 9m56s
istiod-5745bd5f6b-2ptpt 1/1 Running 0 10m
kiali-57fb5bb5c6-26wq8 1/1 Running 0 9m55s
prometheus-78f785fc6b-nwpm8 2/2 Running 0 9m55s
清理istio
执行命令:
istioctl manifest generate --set profile=demo | kubectl delete -f -