Segmentation fault when running on GKE

Hi!

I have a problem with jibri-pod-controller when running on Google Cloud k8s cluster - GKE.
On first try pods in the deployment end up in CrashLoopBackOff without any log accessible with kubectl logs.
The only information I had was

    State:          Waiting
      Reason:       CrashLoopBackOff
    Last State:     Terminated
      Reason:       Error
      Exit Code:    139

So I changed the entrypoint of Docker image to sleep infinity, ran the deployment, enter pod's shell and ran the binary manually ending up with

    / $ /usr/local/bin/jibri-pod-controller
    Segmentation fault

I tried in different namespaces (default and my own custom one) without any luck.
Environment variables seems to be passed correctly since I can see them in the pod's description as well as from inside the shell

    Environment:
      RUST_LOG:           info
      PORT:               8080
      JIBRI_HEALTH_PORT:  2222
      JIBRI_BUSY_LABELS:  app=jibri,state=busy
      SWEEP_INTERVAL:     300
      POD_NAME:           jibri-pod-controller-5c644f77f6-kkgqw (v1:metadata.name)
      NAMESPACE:          default (v1:metadata.namespace)
    Mounts:
      /var/run/secrets/kubernetes.io/serviceaccount from kube-api-access-rb6xb (ro)

I also tried to run this image in local Docker and it outputs error logs correctly (no access to k8s)

[jbg]

Hi! Please share details of how you built the image.

I built the image using repository's Dockerfile just like README says, with my tag and it built successfuly
docker build -t <my_custom_tag> .
My Docker version
Docker version 20.10.17, build 100c701
Here is the full build log from my latest build with sleep entrypoint
docker_build_log.txt

Hi @jbg do you have any uptade in this issue maybe?