Comment 7 for bug 1762450

Revision history for this message
LGB [Gábor Lénárt] (lgb) wrote :

Removing the scsi_mod.use_blk_mq=1 from /etc/default/grub and the mentioned udev rule file, and regenerating initial ramdisk + reboot seems to do the trick, now everything works with default (cfq) I/O scheduler.

Before this comment, and the problem:

# cat /sys/block/sda/queue/scheduler
[bfq] none
# cat /sys/block/sdb/queue/scheduler
[none] bfq

After I removed the BFQ specific stuffs and everything worked again:

# cat /sys/block/sda/queue/scheduler
noop deadline [cfq]
# cat /sys/block/sdb/queue/scheduler
noop deadline [cfq]

So, sorry about the noise, maybe it was my fault ... Interestingly, this setup of mine with BFQ worked in 17.10 on other machines with that kernel parameter and udev rule I mentioned in the previous comment.