SEGFAULT in v1.14.0

[Fricounet]

Hello, I'm seeing segfaults appearing in the node driver starting v1.14.0. It appears when the node plugin tries to mount or unmount volumes after being asked by the kubelet:

  Warning  FailedMount             10s (x8 over 78s)  kubelet                  MountVolume.MountDevice failed for volume "pvc-c22aae37-2017-4430-bcc7-67ca90bbb6ff" : rpc error: code = Internal desc = Error when getting device path: rpc error: code = Internal desc = error verifying GCE PD ("pvc-c22aae37-2017-4430-bcc7-67ca90bbb6ff") is attached: failed to find and re-link disk pvc-c22aae37-2017-4430-bcc7-67ca90bbb6ff with udevadm after retrying for 3s: couldn't get serial number for disk pvc-c22aae37-2017-4430-bcc7-67ca90bbb6ff at device path /dev/sdb: scsi_id failed for device "/dev/sdb" with output : signal: segmentation fault

After trying to exec on the pod to debug, it turned out I couldn't even bash exec (exit code 139 being a segfault):

❯ k exec -it gcp-pd-csi-driver-node-vjvpj -c csi-plugin -- bash
command terminated with exit code 139

I think this might be happening because of #1694 because when I rebuild a new image without this change, I don't have any issue. Maybe the change to the latest debian messed up some of the C libs that are needed by the binaries copied manually in the dockerfile?

[Fricounet]

Actually, turns out that it's because the base image we use to run the driver on has an earlier version of the libc (2.35-0ubuntu3.8) which makes it unable to run the binaries from debian which are now build against (2.36-9+deb12u7).
Closing as this is probably only an issue for us