Добавил:
Upload Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:
Oracle Solaris 11 Advanced Administration Cookbook.docx
Скачиваний:
3
Добавлен:
01.07.2025
Размер:
3 Мб
Скачать
☆

Setting and getting other zfs properties

Managing ZFS properties is one of the secrets when we are working with the ZFS filesystem, and this is the reason why understanding the inherence concept is very important.

One ZFS property can usually have three origins as source: local (the property value was set locally), default (the property wasn't set either locally or by inheritance), and inherited (the property was inherited from an ancestor). Additionally, two other values are possible: temporary (the value isn't persistent) and none (the property is read-only, and its value was generated by ZFS). Based on these key concepts, the sections are going to present different and interesting properties for daily administration.

Getting ready

This recipe can be followed using two virtual machines (VirtualBox or VMware) with Oracle Solaris 11 installed, 4 GB RAM, and eight disks of at least 4 GB.

How to do it…

Working as a small review, datasets such as pools, filesystems, snapshots, and clones have several properties that administrators are able to list, handle, and configure. Therefore, the following commands will create a pool and three filesystems under this pool. Additionally, we are going to copy some data (a reminder again—we could use any data) into the first filesystem as follows:

root@solaris11-1:~# zpool create prop_pool c8t5d0 root@solaris11-1:~# zfs create prop_pool/zfs_1 root@solaris11-1:~# zfs create prop_pool/zfs_2 root@solaris11-1:~# zfs create prop_pool/zfs_3

root@solaris11-1:~# cp -r socat-2.0.0-b6.tar.gz dtbook_scripts* /prop_ pool/zfs_1

To get all the properties from a pool and filesystem, execute the following command:

root@solaris11-1:~# zpool get all prop_pool

NAME PROPERTY VALUE SOURCE

prop_pool allocated 1.13M -

prop_pool altroot - default

prop_pool autoexpand off default

prop_pool autoreplace off default

prop_pool bootfs - default

prop_pool cachefile - default

ZFS

prop_pool capacity 0% -

prop_pool dedupditto 0 default

prop_pool dedupratio 1.00x -

prop_pool delegation on default

prop_pool failmode wait default

prop_pool free 3.97G -

prop_pool guid 10747479388132741479 - prop_pool health ONLINE -

prop_pool listshares off default

prop_pool listsnapshots off default

prop_pool readonly off -

prop_pool size 3.97G -

NAME

PROPERTY

VALUE

SOURCE

prop_pool/zfs_1

aclinherit

restricted

default

prop_pool/zfs_1

aclmode

discard

default

prop_pool/zfs_1

atime

on

default

prop_pool/zfs_1

available

3.91G

prop_pool version 34 default root@solaris11-1:~# zfs get all prop_pool/zfs_1

(truncated output)

Both commands have a similar syntax, and we've got all the properties from the prop_pool

pool and the prop_pool/zfs_1 filesystem.

In the ZFS shadowing section, we touched the NFS subject, and some filesystems were shared using the share command. Nonetheless, they could have been shared using ZFS properties, such as sharenfs, that have a value equal to off by default (when we use this value, it isn't managed by ZFS and is still using /etc/dfs/dfstab). Let's take the sharenfs property, which will be used to highlight some basic concepts about properties.

As usual, the property listing is too long; it is faster to get only one property's value by executing the following command:

root@solaris11-1:~# zfs get sharenfs prop_pool

NAME PROPERTY VALUE SOURCE

prop_pool share.nfs off default

root@solaris11-1:~# zfs get sharenfs prop_pool/zfs_1

NAME PROPERTY VALUE SOURCE

prop_pool/zfs_1 share.nfs off default

Chapter 2

Moreover, the same property can be got recursively by running the following command:

root@solaris11-1:~# zfs get -r sharenfs prop_pool

NAME PROPERTY VALUE SOURCE

prop_pool share.nfs off default prop_pool/zfs_1 share.nfs off default prop_pool/zfs_2 share.nfs off default prop_pool/zfs_3 share.nfs off default

From the last three outputs, we noticed that the sharenfs property is disabled on the pool and filesystems, and this is the default value set by Oracle Solaris 11.

The sharenfs property can be enabled by executing the following command:

root@solaris11-1:~# zfs set sharenfs=on prop_pool/zfs_1 root@solaris11-1:~# zfs get -r sharenfs prop_pool/zfs_1 NAME PROPERTY VALUE SOURCE

prop_pool/zfs_1 share.nfs on local

prop_pool/zfs_1% share.nfs on inherited from prop_pool/zfs_1

As sharenfs was set to on for prop_pool/zfs_1, the source value has changed to local, indicating that this value wasn't inherited, but it was set locally. Therefore, execute the following command:

root@solaris11-1:~# zfs get -s local all prop_pool/zfs_1

NAME PROPERTY VALUE SOURCE

prop_pool/zfs_1 share.* ... local root@solaris11-1:~# zfs get -r sharenfs prop_pool NAME PROPERTY VALUE SOURCE

prop_pool share.nfs off default prop_pool/zfs_1 share.nfs on local

prop_pool/zfs_1% share.nfs on inherited from prop_pool/zfs_1 prop_pool/zfs_2 share.nfs off default

prop_pool/zfs_3 share.nfs off default

The NFS sharing can be confirmed by running the following command:

root@solaris11-1:~# share

IPC$ smb - Remote IPC

c$ /var/smb/cvol smb - Default Share prop_pool_zfs_1 /prop_pool/zfs_1 nfs sec=sys,rw

ZFS

Creating a new file stem under zfs_1 shows us an interesting characteristic. Execute the following command:

root@solaris11-1:~# zfs create prop_pool/zfs_1/zfs_4 root@solaris11-1:~# zfs get -r sharenfs prop_pool NAME PROPERTY VALUE SOURCE

prop_pool share.nfs off default prop_pool/zfs_1 share.nfs on local

prop_pool/zfs_1% share.nfs on inherited from prop_pool/zfs_1 prop_pool/zfs_1/zfs_4 share.nfs on inherited from prop_pool/zfs_1 prop_pool/zfs_1/zfs_4% share.nfs on inherited from prop_pool/zfs_1 prop_pool/zfs_2 share.nfs off default

prop_pool/zfs_3 share.nfs off default

The new zfs_4 filesystem has the sharenfs property inherited from the upper zfs_1

filesystem; now execute the following command to list all the inherited properties:

root@solaris11-1:~# zfs get -s inherited all prop_pool/zfs_1/zfs_4

NAME PROPERTY VALUE SOURCE

prop_pool/zfs_1/zfs_4 share.* ... inherited root@solaris11-1:~# share

IPC$ smb - Remote IPC

c$ /var/smb/cvol smb - Default Share prop_pool_zfs_1 /prop_pool/zfs_1 nfs sec=sys,rw

prop_pool_zfs_1_zfs_4 /prop_pool/zfs_1/zfs_4 nfs sec=sys,rw

That's great! The new zfs_4 filesystem has inherited the sharenfs property, and it appears

in the share output command.

A good question is whether a filesystem will be able to fill all the space of a pool. Yes, it will be able to! Now, this is the reason for ZFS having several properties related to the amount of space on the disk. The first of them, the quota property, is a well-known property that limits how much space a dataset (filesystem in this case) can fill in a pool. Let's take an example:

root@solaris11-1:~# zfs list -r prop_pool

NAME USED AVAIL REFER MOUNTPOINT

prop_pool 399M 3.52G 391M /prop_pool prop_pool/zfs_1 8.09M 3.52G 8.06M /prop_pool/zfs_1 prop_pool/zfs_1/zfs_4 31K 3.52G 31K /prop_pool/zfs_1/zfs_4 prop_pool/zfs_2 31K 3.52G 31K /prop_pool/zfs_2 prop_pool/zfs_3 31K 3.52G 31K /prop_pool/zfs_3

Chapter 2

All filesystems struggle to use the same space (3.52G), and one of them can fill more space than the other (or all the free space), so it is possible that a filesystem suffered a "run out space" error. A solution would be to limit the space a filesystem can take up by executing the following command:

root@solaris11-1:~# zfs quota=1G prop_pool/zfs_3

root@solaris11-1:~# zfs list -r prop_pool

NAME USED AVAIL REFER MOUNTPOINT

prop_pool 399M 3.52G 391M /prop_pool prop_pool/zfs_1 8.09M 3.52G 8.06M /prop_pool/zfs_1 prop_pool/zfs_1/zfs_4 31K 3.52G 31K /prop_pool/zfs_1/zfs_4 prop_pool/zfs_2 31K 3.52G 31K /prop_pool/zfs_2 prop_pool/zfs_3 31K 1024M 31K /prop_pool/zfs_3

The zfs_3 filesystem space was limited to 1 GB, and it can't exceed this threshold. Nonetheless, there isn't any additional guarantee that it has 1 GB to fill. This is subtle—it can't exceed 1 GB, but there is no guarantee that even 1 GB is enough for doing it. Another serious detail—this quota space is shared by the filesystem and all the descendants such as snapshots and clones. Finally and obviously, it isn't possible to set a quota value lesser than the currently used space of the dataset.

A solution for this apparent problem is the reservation property. When using reservation, the space is guaranteed for the filesystem, and nobody else can take this space. Sure, it isn't possible to make a reservation above the quota or maximum free space, and the same rule is followed—the reservation is for a filesystem and its descendants.

When the reservation property is set to a value, this amount is discounted from the total available pool space, and the used pool space is increased by the same value:

root@solaris11-1:~# zfs list -r prop_pool

NAME USED AVAIL REFER MOUNTPOINT

prop_pool 399M 3.52G 391M /prop_pool prop_pool/zfs_1 8.09M 3.52G 8.06M /prop_pool/zfs_1 prop_pool/zfs_1/zfs_4 31K 3.52G 31K /prop_pool/zfs_1/zfs_4 prop_pool/zfs_2 31K 3.52G 31K /prop_pool/zfs_2 prop_pool/zfs_3 31K 1024M 31K /prop_pool/zfs_3

Each dataset under prop_pool has its reservation property:

root@solaris11-1:~# zfs get -r reservation prop_pool

NAME PROPERTY VALUE SOURCE

prop_pool reservation none default prop_pool/zfs_1 reservation none default prop_pool/zfs_1% reservation - -

ZFS

prop_pool/zfs_1/zfs_4

reservation

none

default

prop_pool/zfs_1/zfs_4%

reservation

-

-

prop_pool/zfs_2

reservation

none

default

prop_pool/zfs_3

reservation

none

default

The reservation property is configured to a specific value (for example, 512 MB), given that this amount is subtracted from the pool's available space and added to its used space. Now, execute the following command:

root@solaris11-1:~# zfs set reservation=512M prop_pool/zfs_3

root@solaris11-1:~# zfs list -r prop_pool

NAME USED AVAIL REFER MOUNTPOINT

prop_pool 911M 3.02G 391M /prop_pool prop_pool/zfs_1 8.09M 3.02G 8.06M /prop_pool/zfs_1 prop_pool/zfs_1/zfs_4 31K 3.02G 31K /prop_pool/zfs_1/zfs_4 prop_pool/zfs_2 31K 3.02G 31K /prop_pool/zfs_2 prop_pool/zfs_3 31K 1024M 31K /prop_pool/zfs_3 root@solaris11-1:~# zfs get -r reservation prop_pool

NAME PROPERTY VALUE SOURCE

prop_pool reservation none default prop_pool/zfs_1 reservation none default prop_pool/zfs_1% reservation - - prop_pool/zfs_1/zfs_4 reservation none default prop_pool/zfs_1/zfs_4% reservation - - prop_pool/zfs_2 reservation none default prop_pool/zfs_3 reservation 512M local

The concern about space is usually focused on a total value for the whole pool, but it's possible to limit the available space for individual users or groups.

Setting the quota for users is done through the userquota property and for groups using the

groupquota property:

root@solaris11-1:~# zfs set userquota@aborges=750M

prop_pool/zfs_3

root@solaris11-1:~# zfs set userquota@alexandre=1.5G prop_pool/zfs_3

root@solaris11-1:~# zfs get userquota@aborges prop_pool/zfs_3

NAME PROPERTY VALUE SOURCE

prop_pool/zfs_3 userquota@aborges 750M local root@solaris11-1:~# zfs get userquota@alexandre prop_pool/zfs_3

Chapter 2

NAME PROPERTY VALUE SOURCE

prop_pool/zfs_3 userquota@alexandre 1.50G local root@solaris11-1:~# zfs set groupquota@staff=1G prop_pool/zfs_3 root@solaris11-1:~# zfs get groupquota@staff prop_pool/zfs_3 NAME PROPERTY VALUE SOURCE

prop_pool/zfs_3 groupquota@staff 1G local

Getting the used and quota space from users and groups is done by executing the following command:

root@solaris11-1:~# zfs userspace prop_pool/zfs_3

TYPE NAME USED QUOTA

POSIX User aborges 0 750M POSIX User alexandre 0 1G POSIX User root 3K none

root@solaris11-1:~# zfs groupspace prop_pool/zfs_3

TYPE NAME USED QUOTA

POSIX Group root 3K none POSIX Group staff 0 1G

Removing all the quota values that were set until now is done through the following sequence:

root@solaris11-1:~# zfs set quota=none prop_pool/zfs_3 root@solaris11-1:~# zfs set userquota@aborges=none prop_pool/zfs_3 root@solaris11-1:~# zfs set userquota@alexandre=none prop_pool/zfs_3 root@solaris11-1:~# zfs set groupquota@staff=none prop_pool/zfs_3 root@solaris11-1:~# zfs userspace prop_pool/zfs_3

TYPE NAME USED QUOTA

POSIX User root 3K none

root@solaris11-1:~# zfs groupspace prop_pool/zfs_3

TYPE NAME USED QUOTA

POSIX Group root 3K none

An overview of the recipe

In this section, you saw some properties such as sharenfs, quota, reservation, userquota, and groupquota. All of the properties alter the behavior of the ZFS pool, filesystems, snapshots, and clones. Moreover, there are other additional properties that can improve the ZFS functionality, and I suggest that readers look for all of them in ZFS Administration Guide.

ZFS

Соседние файлы в предмете [НЕСОРТИРОВАННОЕ]