examples/tracing is broken

[RoadRunnr]

Describe the bug

The tracing_otel-collector_1 container from examples/tracing does not start.
A log entry indicates a podman problem:

$ docker logs tracing_backend_1
time="2021-09-15T14:41:07Z" level=error msg="mkdir /rundir/libpod: no such file or directory"

Steps to reproduce

in examples/tracing, do:

docker-compose up -d
docker logs tracing_backend_1

What did you expect to see?

A running instance of the tracing_backend_1 container

What did you see instead?

No instance of the tracing_backend_1 container

What version did you use?

master

What config did you use?

unmodified config from the example

Environment
OS: Ubuntu 21.10

[mx-psi]

cc @owais, related to podman receiver?

[RoadRunnr]

removing the unprivileged user from the Dockerfile solves it, But I'm not sure that this is the correct fix.

[piellick]

Hey @RoadRunnr could you explain your workaround ?

[RoadRunnr]

remove these two lines and rebuild:

ARG USER_UID=10001
USER ${USER_UID}

[gebn]

We ran into this while attempting to upgrade from v0.34.0 to v0.35.0 using the official images. Changing only the tag:

otel/opentelemetry-collector-contrib:0.34.0:

$ kubectl logs otel-collector-b875885fc-2cdqr
2021-09-19T22:33:29.476Z        info    service/collector.go:303        Starting otelcontribcol...      {"Version": "v0.34.0", "NumCPU": 8}
...

otel/opentelemetry-collector-contrib:0.35.0:

$ kubectl logs otel-collector-567676fbbd-mj2fb
time="2021-09-19T22:23:16Z" level=error msg="mkdir /rundir/libpod: permission denied"

In case it's relevant, the pod is running with a custom serviceAccountName to allow k8s_tagger to function.

[githubcdr]

We ran into this while attempting to upgrade from v0.34.0 to v0.35.0 using the official images. Changing only the tag:

confirmed here

[gebn]

This also reproduces with 0.36.0:

time="2021-09-21T23:30:39Z" level=error msg="mkdir /rundir/libpod: permission denied"