No "/bin/timeout" binary in the trace container

[lainra]

HI there,

Thank you very much for making this project.
I tried it in a GKE cluster running with COS nodes and got an error after creating a trace using

kubectl trace run node-name -e 'tracepoint:syscalls:sys_enter_open { printf("%s %s\n", comm,
    str(args->filename)); }'

The container cannot run because of the following error:

Containers:
  kubectl-trace-c59249b4-e5b5-11e9-9f93-f2189825cd80:
    Container ID:  docker://abca6aed039bd81ef98f4ca205c19040ae808477666b3ddfae509767c5e56916
    Image:         quay.io/iovisor/kubectl-trace-bpftrace:latest
    Image ID:      docker-pullable://quay.io/iovisor/kubectl-trace-bpftrace@sha256:fe1683d6782f4ad7690955543c048296b43b6c6a5046ea345a277318c92482d1
    Port:          <none>
    Host Port:     <none>
    Command:
      /bin/timeout
      --preserve-status
      --signal
      INT
      3600
      /bin/trace-runner
      --program=/programs/program.bt
    State:          Terminated
      Reason:       ContainerCannotRun
      Message:      OCI runtime create failed: container_linux.go:345: starting container process caused "exec: \"/bin/timeout\": stat /bin/timeout: no such file or directory": unknown
      Exit Code:    127
      Started:      Thu, 03 Oct 2019 17:14:15 +0900
      Finished:     Thu, 03 Oct 2019 17:14:15 +0900

I checked the /bin/timeout existence in the docker image used:

➜ docker run -it --entrypoint "stat" quay.io/iovisor/kubectl-trace-bpftrace /bin/timeout
stat: can't stat '/bin/timeout': No such file or directory

The PR related to the source code is #88 but there is no trace of adding the binary to the Dockerfile.

Could you check it please?

Cheers,

Raphael

[leodido]

@dalehamel could you please take a look at it? :)

[leogr]

I've the same issue too:

Warning  Failed     37s   kubelet, kind-control-plane  Error: failed to create containerd task: OCI runtime create failed: container_linux.go:345: starting container process caused "exec: \"/bin/timeout\": stat /bin/timeout: no such file or directory": unknown

I've tested with the latest installed from source.

[dalehamel]

This must be that the tag latest isn't updated, it was last updated 2 months so it is missing this binary. This must still be the alpine image, which wouldn't have this binary.

@leogr @lainra can you please try specifying:

--imagename quay.io/iovisor/kubectl-trace-bpftrace:HEAD

And confirm that works?

@leodido @fntlnz we should either update that tag right away (i don't have access to quay.io i don't think, unless it's controlled by github access), or else release a binary that uses the right tag reference.

[leogr]

@dalehamel yes, it works!

[dalehamel]

Sorry about this, the version I QA'd internally had this overridden. We'll get the tag fixed and then there shouldn't be any need to updated the kubectl-trace binary, it should just work.

[dalehamel]

@leogr @lainra the quay.io/iovisor/kubectl-trace-bpftrace:latest tag has been updated, can you please confirm this problem is resolved without needing the --imagename flag?

[leogr]

Hi @dalehamel ,
now yes it works perfectly without --imagename

Image:         quay.io/iovisor/kubectl-trace-bpftrace:latest
Image ID:      sha256:df8ffc4e9f045fa3d8e066d939a8a6b26b5a19c6e68fb76ce64d354b63037a57

😊
Thank you!

[dalehamel]

thanks for confirming @leogr ! I'll close

[lainra]

@dalehamel
I also confirmed it works! Thank you for the quick fix!!!