<POD NETWORK DELAY> : cmd exec failed, err: RTNETLINK answers: No such file or directory exit status 2

[ddd1123]

Issue Description

Type: bug report

Describe what happened (or what feature you want)

1. 在 chaosblade-box 中通过 agent 获取 K8s 集群信息，进行 POD NETWORK DELAY 演练；
2. 报错内容为：原因: /opt/chaosblade/bin/nsexec -t 77143 -p -n -- /bin/sh -c tc qdisc add dev eth0 root netem delay 100ms 10ms: cmd exec failed, err: RTNETLINK answers: No such file or directory exit status 2
3. 我尝试在对应的 Node 主机上输入 /opt/chaosblade/bin/nsexec，同样报 -bash: /opt/chaosblade/bin/nsexec: No such file or directory

Describe what you expected to happen

希望可以提供相关解决方法or解决思路，thanks！

How to reproduce it (as minimally and precisely as possible)

Tell us your environment

K8s：v1.18.18
chaosblade-box：v1.0.1
chaos-agent：v1.0.0
chaos-operator：v1.6.0
chaos-tool：v1.6.0

Anything else we need to know?

机器执行信息：
{

"response": {

"code": 54000,
"error": "unexpected status, expected status: `create`, but the real status: `Error`, please wait!",
"result": {
  "error": "`/opt/chaosblade/bin/nsexec -t 77143 -p -n -- /bin/sh -c tc qdisc add dev eth0 root netem delay 100ms 10ms`: cmd exec failed, err: RTNETLINK answers: No such file or directory\n exit status 2",
  "statuses": [
    {
      "error": "`/opt/chaosblade/bin/nsexec -t 77143 -p -n -- /bin/sh -c tc qdisc add dev eth0 root netem delay 100ms 10ms`: cmd exec failed, err: RTNETLINK answers: No such file or directory\n exit status 2",
      "kind": "pod",
      "state": "Error",
      "success": false
    }
  ],
  "success": false,
  "uid": "2c29793a6c8da651"
},
"success": false

}
}

[ddd1123]

但是我做了一个pod-process-kill的演练：

1. 由于填错了signal所以报错，通过报错信息Reason: /opt/chaosblade/bin/nsexec -t 5165 -p -m -- /bin/sh -c kill -128 128: cmd exec failed, err: /bin/sh: line 0: kill: 128: invalid signal specification exit status 1。我发现也是用的/opt/chaosblade/bin/nsexec
2. 后续我更改了正确的signal后，该pod-process-kill演练成功了。

由此我感觉上述的问题是不是不在于说在node主机上输入/opt/chaosblade/bin/nsexec的报错呢？
那问题出现在哪里呢。。

[ddd1123]

补充：在进行pod-network-delay的演练时：

1. 由于一直报错，我曾尝试进入对应的pod内输入tc qdisc add dev eth0 root netem delay 100ms 10ms，仍报错：Error: Specified qdisc not found.
2. 通过调研发现可能是缺少kernel-modules-extra包，由于种种原因还未成功安装此包
3. 是和这个有关系嘛，如果此包安装成功是否上述的问题则不复存在。

[ddd1123]

operator日志：
演练进行时节选：
time="2022-08-11T01:41:38Z" level=info msg="experiment identifiers: [{{ docker 7420e6a5bff3 centos-tc centos-tc-demo-88d8ff5f8-vg278 192.168.0.4 centos-tc-demo} /opt/chaosblade/blade create cri network delay --timeout=125 --time=100 --interface=eth0 --offset=10 --container-id 7420e6a5bff3 --container-runtime docker 0 chaosblade-tool-ljpcv chaosblade chaosblade-tool}]" experiment=c6c75e78f1a37877
time="2022-08-11T01:41:38Z" level=info msg="execute identifier: {ContainerObjectMeta:{Id: ContainerRuntime:docker ContainerId:7420e6a5bff3 ContainerName:centos-tc PodName:centos-tc-demo-88d8ff5f8-vg278 NodeName:192.168.0.4 Namespace:centos-tc-demo} Command:/opt/chaosblade/blade create cri network delay --timeout=125 --time=100 --interface=eth0 --offset=10 --container-id 7420e6a5bff3 --container-runtime docker Error: Code:0 ChaosBladePodName:chaosblade-tool-ljpcv ChaosBladeNamespace:chaosblade ChaosBladeContainerName:chaosblade-tool}" experiment=c6c75e78f1a37877
time="2022-08-11T01:41:38Z" level=info msg="Exec command in pod" command="[/opt/chaosblade/blade create cri network delay --timeout=125 --time=100 --interface=eth0 --offset=10 --container-id 7420e6a5bff3 --container-runtime docker]" container=chaosblade-tool podName=chaosblade-tool-ljpcv podNamespace=chaosblade
time="2022-08-11T01:41:38Z" level=info msg="get err message" command="[/opt/chaosblade/blade create cri network delay --timeout=125 --time=100 --interface=eth0 --offset=10 --container-id 7420e6a5bff3 --container-runtime docker]" container=chaosblade-tool err="{"code":63063,"success":false,"error":"/opt/chaosblade/bin/nsexec -t 5165 -p -n -- /bin/sh -c tc qdisc add dev eth0 root netem delay 100ms 10ms: cmd exec failed, err: RTNETLINK answers: No such file or directory\n exit status 2"}" out= podName=chaosblade-tool-ljpcv podNamespace=chaosblade
time="2022-08-11T01:41:38Z" level=error msg="pods/exec: k8s exec failed, err: {"code":63063,"success":false,"error":"/opt/chaosblade/bin/nsexec -t 5165 -p -n -- /bin/sh -c tc qdisc add dev eth0 root netem delay 100ms 10ms: cmd exec failed, err: RTNETLINK answers: No such file or directory\n exit status 2"}\n" location=github.com/chaosblade-io/chaosblade-spec-go/util.Errorf uid=

[Icesource]

2. kernel-modules-extra

yum install -y kernel-modules-extra 可以安装该模块，问题似乎是由于 pod 内关于 linux 内核流控工具 tc 引起的相关问题，由于内核默认缺少 netem 流控队列，所以会报错 Error: Specified qdisc not found.

但安装该模块不一定能解决 RTNETLINK answers: No such file or directory exit status 2 该问题，可以先尝试安装kernel-modules-extra该模块看是否能解决问题

[tiny-x]

补充：在进行pod-network-delay的演练时：

1. 由于一直报错，我曾尝试进入对应的pod内输入tc qdisc add dev eth0 root netem delay 100ms 10ms，仍报错：Error: Specified qdisc not found.
2. 通过调研发现可能是缺少kernel-modules-extra包，由于种种原因还未成功安装此包
3. 是和这个有关系嘛，如果此包安装成功是否上述的问题则不复存在。

你系统什么版本 centos 8.x 吗，这个包 8.x 需要安装，并且安装后还要重启机器

[ddd1123]

2. kernel-modules-extra

yum install -y kernel-modules-extra 可以安装该模块，问题似乎是由于 pod 内关于 linux 内核流控工具 tc 引起的相关问题，由于内核默认缺少 netem 流控队列，所以会报错 Error: Specified qdisc not found.

但安装该模块不一定能解决 RTNETLINK answers: No such file or directory exit status 2 该问题，可以先尝试安装kernel-modules-extra该模块看是否能解决问题

好嘞，我再尝试安装下此模块，但是的确之前尝试安装遇到了问题，我尝试的yum源均提示没有此包可以安装。。
另外我在其他问题上看到有说在1.6.x以后不会用到pod内的tc了，这是真的嘛

[tiny-x]

是的，你先确认下你内核版本和发行版本吧

[ddd1123]

补充：在进行pod-network-delay的演练时：

1. 由于一直报错，我曾尝试进入对应的pod内输入tc qdisc add dev eth0 root netem delay 100ms 10ms，仍报错：Error: Specified qdisc not found.
2. 通过调研发现可能是缺少kernel-modules-extra包，由于种种原因还未成功安装此包
3. 是和这个有关系嘛，如果此包安装成功是否上述的问题则不复存在。

你系统什么版本 centos 8.x 吗，这个包 8.x 需要安装，并且安装后还要重启机器

我的系统是CentOS Linux release 7.6.1810 (Core)
我是自己拉取的一个docker镜像，然后通过yum -y install iproute装上了tc命令

[ddd1123]

4.18.0-193.el8.x86_64

[ddd1123]

当前进展：

1、我的系统今天进行了一次变更。现在的版本是CentOS Linux release 8.2.2004 (Core)
2、随即成功安装了kernel-modules-extra包
3、演练成功了！，但是恢复阶段报错，报错信息如下

信息：
{
"response": {
"code": 54000,
"error": "unexpected status, expected status: destroy, but the real status: Destroying, please wait!",
"result": {
"error": "pods/exec: k8s exec failed, err: command terminated with exit code 126",
"statuses": [
{
"error": "pods/exec: k8s exec failed, err: command terminated with exit code 126",
"id": "fff09b30e7e8f4a2",
"kind": "pod",
"state": "Error",
"success": false
}
],
"success": false,
"uid": "f9e03fa41bc7c31f"
},
"success": false
}
}
错误：原因: pods/exec: k8s exec failed, err: command terminated with exit code 126
排查：场景状态不匹配，请稍后再试

日志节选：
time="2022-08-17T07:16:20Z" level=info msg="Exec command in pod" command="[/opt/chaosblade/blade status fff09b30e7e8f4a2]" container=chaosblade-tool podName=chaosblade-tool-ljpcv podNamespace=chaosblade
time="2022-08-17T07:16:20Z" level=info msg="get output message" command="[/opt/chaosblade/blade status fff09b30e7e8f4a2]" container=chaosblade-tool err= out="{"code":200,"success":true,"result":{"Uid":"fff09b30e7e8f4a2","Command":"cri","SubCommand":"network delay","Flag":" --offset=5 --container-id=3e1db8dce103 --timeout=125 --container-runtime=docker --time=60 --interface=eth0","Status":"Success","Error":"","CreateTime":"2022-08-17T07:13:45.153329761Z","UpdateTime":"2022-08-17T07:13:45.181644002Z"}}" podName=chaosblade-tool-ljpcv podNamespace=chaosblade
time="2022-08-17T07:16:20Z" level=error msg="pods/exec: k8s exec failed, err: {"code":200,"success":true,"result":{"Uid":"fff09b30e7e8f4a2","Command":"cri","SubCommand":"network delay","Flag":" --offset=5 --container-id=3e1db8dce103 --timeout=125 --container-runtime=docker --time=60 --interface=eth0","Status":"Success","Error":"","CreateTime":"2022-08-17T07:13:45.153329761Z","UpdateTime":"2022-08-17T07:13:45.181644002Z"}}\n" location=github.com/chaosblade-io/chaosblade-spec-go/util.Errorf uid=

[ddd1123]

恢复阶段报错的operator日志：

time="2022-08-17T07:14:08Z" level=info msg="execute identifier: {ContainerObjectMeta:{Id:fff09b30e7e8f4a2 ContainerRuntime:docker ContainerId:3e1db8dce103 ContainerName:centos-tc-done PodName:centos-tc-done-6b584445b9-g5hnw NodeName:192.168.0.4 Namespace:centos-tc-done} Command: --container-label-selector io.kubernetes.pod.name=centos-tc-done-6b584445b9-g5hnw,io.kubernetes.pod.namespace=centos-tc-done,io.kubernetes.docker.type=podsandbox --container-runtime docker Error: Code:0 ChaosBladePodName:chaosblade-tool-ljpcv ChaosBladeNamespace:chaosblade ChaosBladeContainerName:chaosblade-tool}" experiment=f9e03fa41bc7c31f
time="2022-08-17T07:14:08Z" level=info msg="Exec command in pod" command="[ --container-label-selector io.kubernetes.pod.name=centos-tc-done-6b584445b9-g5hnw,io.kubernetes.pod.namespace=centos-tc-done,io.kubernetes.docker.type=podsandbox --container-runtime docker]" container=chaosblade-tool podName=chaosblade-tool-ljpcv podNamespace=chaosblade
time="2022-08-17T07:14:08Z" level=error msg="Invoke exec command error" command="[ --container-label-selector io.kubernetes.pod.name=centos-tc-done-6b584445b9-g5hnw,io.kubernetes.pod.namespace=centos-tc-done,io.kubernetes.docker.type=podsandbox --container-runtime docker]" container=chaosblade-tool err= error="command terminated with exit code 126" out="OCI runtime exec failed: exec failed: container_linux.go:380: starting container process caused: exec: "": executable file not found in $PATH: unknown" podName=chaosblade-tool-ljpcv podNamespace=chaosblade
time="2022-08-17T07:14:08Z" level=error msg="pods/exec: k8s exec failed, err: command terminated with exit code 126" location=github.com/chaosblade-io/chaosblade-spec-go/util.Errorf uid=fff09b30e7e8f4a2

[Icesource]

这可能是平台侧设置的轮训时间太短导致的异常，实际上实验不久后就被正常销毁了，你可以通过观察现象判断 实验是否被正常销毁

[ddd1123]

这可能是平台侧设置的轮训时间太短导致的异常，实际上实验不久后就被正常销毁了，你可以通过观察现象判断 实验是否被正常销毁

感谢回复
尝试了几次并进行观察现象，均没能销毁实验。通过tc qdisc show查看仍存在tc qdisc add ... 添加的实验内容

通过查看报错信息"error": "pods/exec: k8s exec failed, err: command terminated with exit code 126",考虑是因为恢复时并没有成功进入对应的pod，故障注入是能够成功进入的，而恢复不能进入pod就有点问题

[Icesource]

在pod内看看chaosblade的执行日志呢？ 日志一般在/opt/chaosblade下

[ddd1123]

在pod内看看chaosblade的执行日志呢？ 日志一般在/opt/chaosblade下
感谢回复

日志如下，其中10:43为成功执行，10:45为恢复日志
time="2022-08-23 10:43:28.128243385 UTC" level=info msg="create uid: 72e94f5b8b62644b, target: network, scope: pod, action: delay"
time="2022-08-23 10:43:28.142013125 UTC" level=error msg="chaosblade result: []" location=github.com/chaosblade-io/chaosblade/exec/kubernetes.QueryStatus uid=72e94f5b8b62644b
time="2022-08-23 10:45:39.431496547 UTC" level=info msg="destroy by 72e94f5b8b62644b uid, force-remove: false, target: "
time="2022-08-23 10:45:39.65012422 UTC" level=error msg="unexpected status, expected status: destroyed, but the real status: Running, please wait!" location=github.com/chaosblade-io/chaosblade/exec/kubernetes.QueryStatus uid=72e94f5b8b62644b
time="2022-08-23 10:45:43.434151464 UTC" level=error msg="chaosblade result: [{pod network delay false Success see resStatus for the error details [{fd98461695b31b38 Error 0 pods/exec: k8s exec failed, err: command terminated with exit code 126 false pod centos-tc/192.168.0.3/centos-tc-5bc68ff56f-f46fl/centos-tc-done/46d20d1c607c/docker}]}]" location=github.com/chaosblade-io/chaosblade/exec/kubernetes.QueryStatus uid=72e94f5b8b62644b

[zshmmm]

pod network delay 实验时，销毁实验失败：
/opt/chaosblade/bin/nsexec -t 11077 -p -n -- /bin/sh -c tc qdisc del dev eth0 root`: cmd exec failed, err: RTNETLINK answers: No such file or directory exit status 2

在响应node节点的 chaosblade-tool 容器中执行 /opt/chaosblade/bin/nsexec -t 11077 -p -n -- /bin/sh -c tc qdisc del dev eth0 root 同样报错，需要将执行名字加“引号”，然后再执行就可以了。
像下面这样：
/opt/chaosblade/bin/nsexec -t 11077 -p -n -- "/bin/sh -c tc qdisc del dev eth0 root"

是否是因为演练工具的 exec 模块执行命令的格式不对。