Proxmox VE 9 host freeze on kernel 6.14.11-2-pve (soft lockup, requires hard reset)

[Perfeitor]

I am experiencing full host freezes on Proxmox VE 9 with kernel 6.14.11-2-pve. The system becomes unresponsive (SSH, Web UI all dead), and I need to reboot manually to recover.

Environment​

• CPU: Intel Core i5-12600
• Memory: 64GB DDR4 (no ECC)
• Storage:
  • rpool (NVMe Micron 256GB, firmware 3011P4LN)
  • vmpool (NVMe Samsung 1TB, firmware 5M2QEXF7)
  • lxcpool (Micron 256GB, same as above)
• Proxmox VE: 9.0.10
• Kernel: 6.14.11-2-pve (also installed: 6.8.12-13-pve)
• VMs: mix of Linux and one Windows Server VM (using VirtIO drivers)
• Microcode: intel-microcode 3.20250211.1~deb12u1

SMART / ZFS​

• All ZFS pools ONLINE, no data errors (zpool status).
• Micron NVMe: ~50TB reads, ~27TB writes, 128 unsafe shutdowns, 1197 NVMe error log entries (Invalid Field in Command).
• Samsung NVMe: ~23TB reads, ~28TB writes, 116 unsafe shutdowns, 3653 NVMe error log entries (Invalid Field in Command).
• Both drives still report “SMART PASSED” and healthy temperature.

Logs​

From journalctl -b -1 -e (before crash):

watchdog: BUG: soft lockup - CPU#4 stuck for 30927s! [systemd:1]
watchdog: BUG: soft lockup - CPU#2 stuck for 30927s! [pveproxy worker:353325]
watchdog: BUG: soft lockup - CPU#9 stuck for 30927s! [pveproxy worker:337252]
watchdog: BUG: soft lockup - CPU#11 stuck for 30923s! [kworker/u48:1:468685]
watchdog: BUG: soft lockup - CPU#5 stuck for 20779s! [kworker/5:2:522670]
watchdog: BUG: soft lockup - CPU#6 stuck for 30819s! [rcu_exp_gp_kthr:20]

Call trace snippet:

on_each_cpu_cond_mask
flush_tlb_kernel_range
__purge_vmap_area_lazy
drain_vmap_area_work
process_one_work
worker_thread
kthread

From dmesg after reboot:

x86/split lock detection: #AC: CPU 1/KVM/4885 took a split_lock trap

Has anyone else seen similar soft lockups / RCU stalls on Proxmox VE 9 with kernel 6.14.x?
Would it be recommended to switch back to kernel LTS for stability?
Any advice or confirmation would be greatly appreciated.