Root disk layout, Zfs within a zone – Sun Microsystems SOLARIS 10 User Manual
Page 46
Version 3.1-en
Solaris 10 Container Guide - 3.1 4. Best Practices
Effective: 30/11/2009
4.1.6.4. Root disk layout
[dd] Depending on availability requirements, root disks within a system are mirrored via internal disks
or made available through a variety of controllers and external storage devices. Even nowadays, the
entire OS installation is installed in / in many cases, that is, without any further partitioning into
/usr or /opt. Only /var must be set up as a separate file system in order to avoid flooding / by
e.g. big log files. Furthermore, additional space must be provided for a live upgrade scheduled for
later.
Once the number of required slices per disk is depleted, /var/crash can be translocated
into /var, or the use of soft partitions can be considered.
Frequently, /usr/local is needed as a writable directory in a local zone. If for example a sparse-
root zone is used and /usr is an inherit-pkg-dir, then it is not possible to write into
/usr/local in the local zone. In such cases, it is advisable to set up a soft link in the global zone,
e.g. /usr/local -> ../../opt/local. Since /opt is normally separate for each zone,
writing permission exists for each of these zones. It is important to use relative links so that even an
access from the global zone to e.g. /zones/zone-z/root/usr/local ends up in the
correct directory.
The companion DVD software is installed as /opt/sfw. If it is to be installed centrally, /opt/sfw
would furthermore need to be specified as inherit-pkg-dir. StarOffice for example installs as
/opt/staroffice.
File system sizes vary depending on the type and scope of the installations.
An example root disk layout is shown here (
).
4.1.6.5. ZFS within a zone
[ug] Solaris 10 6/06 makes the new ZFS file system available for the first time. Since ZFS is not
based on a special mounted device, a mechanism is implemented in the zonecfg command by
which a ZFS file system can be transferred to a zone. This is achieved through the zonecfg
subcommand add dataset . The ZFS file system then becomes visible in the zone and can be
used there and administered (with limitations).
With ZFS, file system attributes can be specified that are activated immediately. Here, the attributes
quota (maximum allocatable disk space) and reservation (guaranteed disk space) are of
interest for the file system. If a file system is transferred to a zone through add dataset, the
administrator of the global zone can specify the attributes quota and reservation. The
administrator of the local zone cannot modify these parameters in order to access the resources
available in other zones.
ZFS allows file systems to be hierarchically structured. This is also true for the ZFS file systems
transferred to a zone, which can be further subdivided by the zone administrator. If the attributes
quota and reservation are specified for the transferred file system, the limitation will also
continue to apply within the zone. Consequently, the resources available within the zone are limited.
User-defined attributes for ZFS file systems were introduced with Solaris 10 5/08. These are
particularly useful, especially in the case of ZFS file systems transferred to zones, for documenting
ZFS file system use (
User attributes for ZFS within a zone
).
39