Compiled crcmod required for restores using multi-part uploads from google_storage

[yuraymondl]

What happened? When the medusa-restore init container attempts a restore from a GCS backup created using multipart uploads, timeouts and ultimately failure occur. This is likely due to the medusa docker image (k8ssandra/medusa:0.13.4) not having compiled crcmod installed, which was confirmed via error logs in the outputted gsutil manifest file in the medusa-restore init container as well as checking gsutil version -l which yielded compiled crcmod: false.

Did you expect to see something different? Expected a functioning restore similar to the example case in the documentation, which succeeded for me.

How to reproduce it (as minimally and precisely as possible): Restore from a backup that contains files backed up using multipart uploads.

Environment

• K8ssandra Operator version: v1.5.2

  v1.5.2

• Kubernetes version information:

  v1.26.0

• Kubernetes cluster kind:

  kind, via the setup-kind-multicluster.sh script.

┆Issue is synchronized with this Jira Story by Unito

[yuraymondl]

I was able to get around this by building a new docker image based off of k8ssandra/medusa:0.13.4

FROM docker.io/k8ssandra/medusa:0.13.4 USER root RUN apt-get update -y && apt-get install gcc python3-dev python3-pip python3-setuptools -y RUN pip3 install --force-reinstall --no-input --no-cache dir -U crcmod USER cassandra

[adejanovski]

Hi @yuraymondl,

good to know. If you feel like pushing a PR to the thelastpickle/cassandra-medusa repo, it would be fantastic 🙏