Support for zone storage devices in various versions of Linux-based operating systems

Distribution kit	Core	Supported technologies
		ZBD	dm-zoned
Leap 15.0	4.12	Yes	Not
Leap 15.1	4.12	Yes	Not
Leap 15.2	5.3	Yes	Yes
Tumbleweed	5.9+	Yes	Yes

Does your operating system support zone storage?

As mentioned above, the Zoned Block Device (ZBD) interface supporting the T10 ZBC and T13 ZAC standards has been added to the Linux kernel since version 4.10, so the first thing to do is check the current version. kernel using the terminal command uname. For example, using Fedora 29, the output from this command will look like this:

uname -r
5.0.13-200.fc29.x86_64

In our case, the fifth version of the kernel is used, in which support for zone storage devices has already been added. However, in order for such drives to be able to interact with the operating system, the kernel must be initially compiled with the CONFIG_BLK_DEV_ZONED parameter active.

To check if this option is enabled, you can directly access the operating system configuration file located in the / boot or / directory, or use the command:

cat /boot/config-`uname -r` | grep CONFIG_BLK_DEV_ZONED

or

cat /lib/modules/`uname -r`/config | grep CONFIG_BLK_DEV_ZONED

For kernels exporting configuration settings via the proc virtual file system, the following commands can be used:

modprobe configs
cat /proc/config.gz | gunzip | grep CONFIG_BLK_DEV_ZONED

or

modprobe configs
zcat /proc/config.gz | grep CONFIG_BLK_DEV_ZONED

If the corresponding option is active, you will see the following output:

CONFIG_BLK_DEV_ZONED=y

If the terminal displays

CONFIG_BLK_DEV_ZONED=n

this means that support for zoned block devices is disabled and the kernel must be recompiled first. By the way, if you try to run the above commands on an operating system that uses a kernel earlier than 4.10, their output will be empty.

The CONFIG_BLK_DEV_ZONED configuration parameter controls support for zone block device emulation at the null_blk driver level, SCSI support for ZBC and ZAC SMR drives, and access to a set of zone namespace commands required to work with NVMe SSDs. As for the control of the order of the write commands, everything turns out to be a little more complicated here.

Initially (in kernels from 4.10 to 4.15 inclusive) this function was implemented at the kernel level. However, starting with version 4.16, control of the order of write commands was transferred to the block-level I / O schedulers deadline (works with a single command queue) and mq-deadline (deadline implementation using blk-mq, supports multiple queues), the use of which has become mandatory for correct operation zone storage devices. Later, with the release of kernel 5.0, deadline support was completely removed and the mq-deadline scheduler became the main solution for working with zoned storage devices.

Thus, if you are using an operating system based on kernel 4.16 or later, you need to make sure that deadline or mq-deadline is selected as the I / O scheduler. To do this, just use the following command:

cat /sys/block/sdb/queue/scheduler

[mq-deadline] kyber bfq none

If a different scheduler is used by default, you can change it to the desired one by entering the command:

echo deadline > /sys/block/sdb/queue/scheduler

or

echo mq-deadline > /sys/block/sdb/queue/scheduler

Now you can be confident that the operating system you are using will be able to communicate adequately with zoned storage devices.