Unable to filter traffic based on container ID and pod name.

[liuhaibin0528]

Describe the bug
使用kyanos 对容器id和pod名称进行watch时，无法正常执行，提示没有该容器id和pod名称

To Reproduce
使用最新版本的即可重新

Expected behavior
正常对容器id和pod进行watch抓取流量

Screenshots

Information (please complete the following information):

• OS: openEuler 22.03 LTS
  内核：5.10.0-60.139.0.166.oe2203.aarch64

[dosubot]

@liuhaibin0528 你好！我是Dosu 🤖

When Kyanos fails to find a specified container or pod, it typically generates error messages such as "can not find any running container by id %s" or "can not find any running pod by name %s in namespace %s" ^[1][2]. Common causes for these errors include:

1. Incorrect Container ID or Pod Name: Ensure that the container ID or pod name you are using matches exactly with the running instances. Even a small typo can cause Kyanos to fail in identifying the correct resource.

2. Namespace Mismatch: Verify that the namespace specified in your command matches the namespace where the pod is running. In Kubernetes, pods are often organized into different namespaces, and specifying the wrong one will lead to errors.

3. Multiple Containers with the Same Name: If there are multiple containers with the same name, Kyanos might not be able to identify the correct one. Ensure that the container name is unique or use the container ID for more precise identification.

4. CRI Endpoint Issues: Kyanos interacts with container runtimes via CRI endpoints. Ensure that the CRI endpoints are correctly configured and accessible. If there are issues with these endpoints, Kyanos might not be able to retrieve the necessary information.

Given that you are using openEuler 22.03 LTS, ensure that there are no compatibility issues with the kernel version you are using, although no specific issues have been documented ^[3]. Additionally, make sure that Kyanos is running with the necessary privileges, especially if it is being executed within a container or Kubernetes pod ^[3].

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Other}

[hengyoush]

Are there any logs similar to kyanos_xxx.log under the /tmp path? If so, please provide them. @liuhaibin0528

[liuhaibin0528]

kyanos.log content：

only exec watch container printing

[hengyoush]

@liuhaibin0528
There two problems here:

1. You execute kubectl get pods -n devops, then use kyanos to filter traffic by the pod name ***-agent-***, after that kyanos tell you that the pod with the name can't be found.

The reason is that you must ensure the machine running Kyanos is on the same host as the specified pod.

2. The second issue is that filtering by container-id is not working.

Currently, Kyanos supports filtering by either the full untruncated container-id or the 12-character short id. The container-id you're using is actually a 13-character id truncated by crictl, so Kyanos fails to filter. Before we fix this(#204, #205 ), you can use the first 12 characters for filtering.