CONTAP-462993: NVMe-oF keep-alive tasks may stop running after key-management commands
Issue
On systems using NVMe-oF shelves, keep-alive commands may not run for more than 15 seconds after key-management commands. This can cause NVMe-oF connections to drop and, if enough drives are affected, may result in a multidisk failure.
The defect was associated with a race in the keep-alive timer’s shutdown handling. The diagnostic panic string reported for this defect is:
The defect was associated with a race in the keep-alive timer’s shutdown handling. The diagnostic panic string reported for this defect is:
PANIC: mutex peg_nvmeof_ctrlr_keep_alive_timer_lock not owned at ../../../../src/sys/kern/kern_mutex.c:294