Talos shutdown stucks on mlx4_shutdown

[paraddise]

Bug Report

Description

We connect pci device SRIOV and use it for network communications (cilium and drbd), i.e. we use IP over InfiniBand.
When I reboot nodes with small load on this devices like control plane or smal ingress nodes it reboots normally.
But when i'm trying to reboot worker node with several drbd resource connected over ib device talos stucks on mlx4_shutdown was called.

Logs

Can't copy as text.

Environment

• Talos version:

Client:
        Tag:         v1.8.2
        SHA:         88f861a0
        Built:       
        Go version:  go1.22.8
        OS/Arch:     darwin/amd64
Server:
        NODE:        10.21.20.11
        Tag:         v1.9.1
        SHA:         348472f9
        Built:       
        Go version:  go1.23.4
        OS/Arch:     linux/amd64
        Enabled:     RBAC

• Kubernetes version: 1.30.3
• Platform: Proxmox 8.2.2

[frezbo]

was this just normal talosctl reboot which does a kexec, you'd need to use talosctl reboot --mode=powercycle some drivers are not designed to properly shutdown on kexec

[paraddise]

Thanks I'll try it next time

[paraddise]

No success, same error