Ok, I will investigate as soon as I have time. By the way, I am wondering if using BFQ as I/O scheduler can be the problem, ie:
# cat /etc/udev/rules.d/60-schedulers.rules ## set noop scheduler for non-rotating disks ACTION=="add|change", KERNEL=="sd[a-z]", ATTR{queue/rotational}=="0", ATTR{queue/scheduler}="noop" ## set bfq scheduler for rotating disks ACTION=="add|change", KERNEL=="sd[a-z]", ATTR{queue/rotational}=="1", ATTR{queue/scheduler}="bfq"
# cat /proc/cmdline BOOT_IMAGE=/boot/vmlinuz-4.15.0-15-generic root=UUID=d203063e-78c0-425a-bee2-0e3ecb4ea74c ro quiet splash scsi_mod.use_blk_mq=1 vt.handoff=1
I guess, I try to remove BFQ specific stuffs as well to see if this fixes the problem.
Ok, I will investigate as soon as I have time. By the way, I am wondering if using BFQ as I/O scheduler can be the problem, ie:
# cat /etc/udev/ rules.d/ 60-schedulers. rules ="add|change" , KERNEL=="sd[a-z]", ATTR{queue/ rotational} =="0", ATTR{queue/ scheduler} ="noop" ="add|change" , KERNEL=="sd[a-z]", ATTR{queue/ rotational} =="1", ATTR{queue/ scheduler} ="bfq"
## set noop scheduler for non-rotating disks
ACTION=
## set bfq scheduler for rotating disks
ACTION=
# cat /proc/cmdline /boot/vmlinuz- 4.15.0- 15-generic root=UUID= d203063e- 78c0-425a- bee2-0e3ecb4ea7 4c ro quiet splash scsi_mod. use_blk_ mq=1 vt.handoff=1
BOOT_IMAGE=
I guess, I try to remove BFQ specific stuffs as well to see if this fixes the problem.