aws-efs-csi-driver: helm chart v2.3.0 -> EFS CSI v1.4.3 -> missing '/usr/bin/stunnel'
/kind bug
What happened?
On EKS 1.23 I tried to upgrade the EFS CSI helm chart from v2.2.9 to v.2.3.0, but the efs-csi-node Pods started to crash like this:
efs-plugin E1102 11:28:25.736669 1 efs_watch_dog.go:285] Process amazon-efs-mount-watchdog exits exit status 1
efs-plugin E1102 11:28:25.848044 1 efs_watch_dog.go:331] amazon-efs-mount-watchdog[Error]: Traceback (most recent call last):
efs-plugin File "/usr/bin/amazon-efs-mount-watchdog", line 2014, in <module>
efs-plugin main()
efs-plugin File "/usr/bin/amazon-efs-mount-watchdog", line 2004, in main
efs-plugin unmount_count_for_consistency,
efs-plugin File "/usr/bin/amazon-efs-mount-watchdog", line 1029, in check_efs_mounts
efs-plugin restart_tls_tunnel(child_procs, state, state_file_dir, state_file)
efs-plugin File "/usr/bin/amazon-efs-mount-watchdog", line 943, in restart_tls_tunnel
efs-plugin new_tunnel_pid = start_tls_tunnel(child_procs, state_file, state["cmd"])
efs-plugin File "/usr/bin/amazon-efs-mount-watchdog", line 858, in start_tls_tunnel
efs-plugin close_fds=True,
efs-plugin File "/usr/lib64/python3.7/subprocess.py", line 800, in __init__
efs-plugin restore_signals, start_new_session)
efs-plugin File "/usr/lib64/python3.7/subprocess.py", line 1551, in _execute_child
efs-plugin raise child_exception_type(errno_num, err_msg, err_filename)
efs-plugin FileNotFoundError: [Errno 2] No such file or directory: '/usr/bin/stunnel': '/usr/bin/stunnel'
I had to roll back to v2.2.9 version.
What you expected to happen?
Smooth helm upgrade.
How to reproduce it (as minimally and precisely as possible)?
Upgrade EFS CSI helm chart from v2.2.9 to v.2.3.0.
Anything else we need to know?:
Environment
- Kubernetes version (use
kubectl version): v1.23.x - Driver version: v1.4.3
About this issue
- Original URL
- State: closed
- Created 2 years ago
- Reactions: 10
- Comments: 21 (1 by maintainers)
I have the same problem.
EKS 1.22, node OS is Amazon Linux 2.
Looks like the
aws-efs-csi-driver:v1.4.2Docker image still has/usr/bin/stunnel, but inaws-efs-csi-driver:v1.4.3it’s/usr/bin/stunnel5.yum listshows:stunnel.x86_64 4.56-6.amzn2.0.3inaws-efs-csi-driver:v1.4.2stunnel5.x86_64 5.58-1.amzn2.0.1inaws-efs-csi-driver:v1.4.3Did the revert get reverted? We are seeing this in 1.4.5.
I’m kinda amazed how this was released in the first place. This is critical issue that breaks intended driver functionality. Did e2e tests catch this (but results were ignored) or is this a flaw in the test suite?