checkout command 'sh -c "cat /etc/*release | grep ^ID"' fails on containers with musl libc as 'unknown'

[heavyrain266]

Hello, I just wanted to test program compilation on musl libc with voidlinux-musl/alpine containers and checkout always fails with "unknown" while same command locally also in docker works fine like on this screenshot. Containers with glibc passes without any problems.

Here I'm leaving screenshot with errors and link to actions logs

[jkonecny12]

I'm experiencing similar issue on our custom container based on Fedora image. However, in our case we're getting traceback.

Run actions/checkout@v2
/usr/bin/docker exec  fae19cbcbd4ef47358df43ceb3862152e41feb3eb5f730929439a4dc36240736 sh -c "cat /etc/*release | grep ^ID"
/__e/node12/bin/node[16]: ../src/node_platform.cc:63:std::unique_ptr<long unsigned int> node::WorkerThreadsTaskRunner::DelayedTaskScheduler::Start(): Assertion `(0) == (uv_thread_create(t.get(), start_thread, this))' failed.
 1: 0x9da7c0 node::Abort() [/__e/node12/bin/node]
 2: 0x9da847  [/__e/node12/bin/node]
 3: 0xa49c8a node::WorkerThreadsTaskRunner::WorkerThreadsTaskRunner(int) [/__e/node12/bin/node]
 4: 0xa49d4b node::NodePlatform::NodePlatform(int, node::tracing::TracingController*) [/__e/node12/bin/node]
 5: 0x94e3da node::InitializeV8Platform(int) [/__e/node12/bin/node]
 6: 0x9ac4c1 node::InitializeOncePerProcess(int, char**) [/__e/node12/bin/node]
 7: 0x9ac6b3 node::Start(int, char**) [/__e/node12/bin/node]
 8: 0x7f732230d560  [/lib64/libc.so.6]
 9: 0x7f732230d60c __libc_start_main [/lib64/libc.so.6]
10: 0x94a055  [/__e/node12/bin/node]

Link to the run.

[jkonecny12]

Just a note that this doesn't work in official fedora container. If you use

container:
      image: docker.io/fedora:rawhide

then the checkout action will fail with the traceback above. Please fix this issue, it makes our workflows much more complicated than it is necessary.

[sarroutbi]

Hello. We are facing this issue also ... is this going to be addressed by this repo's team? Do you know if there is a workaround?

[jkonecny12]

The workaround is to use the container only for the required steps not the whole job. However, it makes the workflow much less readable :(.

[sarroutbi]

Ok, thanks for the quick update @jkonecny12