zones - Solaris application containers
The zones facility in Solaris provides an isolated environment for running applications. Processes running in a zone are prevented from monitoring or interfering with other activity in the system. Access to other processes, network interfaces, file systems, devices, and inter-process communication facilities are restricted to prevent interaction between processes in different zones.
The privileges available within a zone are restricted to prevent operations with system-wide impact. See privileges(5).
You can configure and administer zones with the zoneadm(1M) and zonecfg(1M) utilities. You can specify the configuration details a zone, install file system contents including software packages into the zone, and manage the runtime state of the zone. You can use the zlogin(1) to run commands within an active zone. You can do this without logging in through a network-based login server such as in.rlogind(1M) or sshd(1M).
The autobooting of zones is enabled and disabled by the zones service, identified by the FMRI:
See zoneadm(1M). Note that a zone has an autoboot property, which can be set to true (always autoboot). However, if the zones service is disabled, autoboot will not occur, regardless of the setting of the autoboot property for a given zone. During a clean shutdown of the global zone, this service shuts down, halts, or suspends each zone, depending on its autoshutdown property. See zonecfg(1M).
An alphanumeric name and numeric ID identify each active zone. Alphanumeric names are configured using the zonecfg(1M) utility. Numeric IDs are automatically assigned when the zone is booted. The zonename(1) utility reports the current zone name, and the zoneadm(1M) utility can be used to report the names and IDs of configured zones.
A zone can be in one of several states:
Indicates that the configuration for the zone has been completely specified and committed to stable storage.
Indicates that the zone is in the midst of being installed or uninstalled, or was interrupted in the midst of such a transition.
Indicates that the zone's configuration has been instantiated on the system: packages have been installed under the zone's root path.
Indicates that the “virtual platform” for the zone has been established. For instance, file systems have been mounted, devices have been configured, but no processes associated with the zone have been started.
Indicates that user processes associated with the zone application environment are running.
Indicates that the zone is being halted. The zone can become stuck in one of these states if it is unable to tear down the application environment state (such as mounted file systems) or if some portion of the virtual platform cannot be destroyed. Such cases require operator intervention.
Indicates that the zone has been installed but cannot be booted. A zone enters the unavailable state when the zone's storage is unavailable while svc:/system/zones:default is onlining or when the zone tries to boot; when archive-based installations fail after successful archive extraction; and when the zone's software is incompatible with the global zone's software, such as after an improper forced attach.
Processes running inside a zone (aside from the global zone) have restricted access to other processes. Only processes in the same zone are visible through /proc (see proc(4) or through system call interfaces that take process IDs such as kill(2) and priocntl(2). Attempts to access processes that exist in other zones (including the global zone) fail with the same error code that would be issued if the specified process did not exist.
Processes running within a non-global zone are restricted to a subset of privileges, in order to prevent one zone from being able to perform operations that might affect other zones. The set of privileges limits the capabilities of privileged users (such as the super-user or root user) within the zone. The list of privileges available within a zone can be displayed using the ppriv(1) utility. For more information about privileges, see privileges(5).
The set of devices available within a zone is restricted, to prevent a process in one zone from interfering with processes in other zones. For example, a process in a zone should not be able to modify kernel memory using /dev/kmem, or modify the contents of the root disk. Thus, by default, only a few pseudo devices considered safe for use within a zone are available. Additional devices can be made available within specific zones using the zonecfg(1M) utility.
The device and privilege restrictions have a number of effects on the utilities that can run in a non-global zone. For example, the eeprom(1M), prtdiag(1M), and prtconf(1M) utilities do not work in a zone since they rely on devices that are not normally available.
A zone can be assigned a brand when it is initially created. A branded zone is one whose software does not match that software found in the global zone. The software can include Solaris software configured or laid out differently. The particular collection of software is called a “brand” (see brands(5)). Once installed, a zone's brand can not be changed unless the zone is first uninstalled. The solaris-kz brand provides even more flexibility, altering many of the restrictions mentioned in here.
Each zone has its own section of the file system hierarchy, rooted at a directory known as the zone root. Processes inside the zone can access only files within that part of the hierarchy, that is, files that are located beneath the zone root. This prevents processes in one zone from corrupting or examining file system data associated with another zone. The chroot(1M) utility can be used within a zone, but can only restrict the process to a root path accessible within the zone.
In order to preserve file system space, sections of the file system can be mounted into one or more zones using the read-only option of the lofs(7FS) file system. This allows the same file system data to be shared in multiple zones, while preserving the security guarantees supplied by zones.
NFS and autofs mounts established within a zone are local to that zone; they cannot be accessed from other zones, including the global zone. The mounts are removed when the zone is halted or rebooted.
ZFS datasets that are delegated to a zone are managable within the zone. Within a delegated dataset, child datasets can be created. Datasets that are created within a delegated dataset are themselves delegated. Delegated datasets other than the top level delegated dataset can be destroyed. Most, but not all, properties can be set on delegated datasets. See zfs(1M) for details.
Each zone has a top-level delegated dataset, which in turn contains the ROOT and potentially other datasets such as .../export and .../export/home. Datasets that exist under the ROOT dataset make up the zone's boot environment(s). Boot environment datasets should only be created or destroyed using the zoneadm(1M) or beadm(1M) commands.
A zone has its own port number space for TCP, UDP, and SCTP applications and typically one or more separate IP addresses (but some configurations of Trusted Extensions share IP address(es) between zones).
For the IP layer (IP routing, ARP, IPsec, IP Filter, and so on) a zone can either share the configuration and state with the global zone (a shared-IP zone), or have its distinct IP layer configuration and state (an exclusive- IP zone).
If a zone is to be connected to the same datalink, that is, be on the same IP subnet or subnets as the global zone, then it is appropriate for the zone to use the shared IP instance.
If a zone needs to be isolated at the IP layer on the network, for instance being connected to different VLANs or different LANs than the global zone and other non-global zones, then for isolation reasons the zone should have its exclusive IP.
A shared-IP zone is prevented from doing certain things towards the network (such as changing its IP address or sending spoofed IP or Ethernet packets), but an exclusive-IP zone has more or less the same capabilities towards the network as a separate host that is connected to the same network interface. In particular, the superuser in such a zone can change its IP address and spoof ARP packets.
The shared-IP zones are assigned one or more network interface names and IP addresses in zonecfg(1M). The network interface name(s) must also be configured in the global zone.
The exclusive-IP zones are assigned one or more network interface names in zonecfg(1M). The network interface names must be exclusively assigned to that zone, that is, it (or they) can not be assigned to some other running zone, nor can they be used by the global zone.
The full IP-level functionality in the form of DHCP client, IPsec and IP Filter, is available in exclusive-IP zones and not in shared-IP zones.
A zone is capable of emulating a 32-bit host identifier, which can be configured via zonecfg(1M), for the purpose of system consolidation. If a zone emulates a host identifier, then commands such as hostid(1) and sysdef(1M) as well as C interfaces such as sysinfo(2) and gethostid(3C) that are executed within the context of the zone will display or return the zone's emulated host identifier rather than the host machine's identifier.
The output of the zone's console is logged to /var/log/zones/console.<zonename>. Other runtime information is logged to /var/log/zones/messages.<zonename>. Each log is rotated periodically using logadm(1M).
You can reconfigure a running zone without needing to reboot it. You can also inspect live configuration of a running zone. zonecfg(1M) allows to retrieve and inspect the live configuration, make desired changes, and temporarily apply them to the running zone. zonecfg(1M) allows to reconfigure the running zone persistently based on the saved zone configuration. For more information, see zonecfg(1M).
The zone configuration consists of resources and resource properties defined in the zonecfg(1M) manual page. For the purpose of live zone reconfiguration, only resources and resource properties known to zonecfg(1M), which are also permitted by the associated brand are supported.
See brand specific manual pages for the list of resources and resource properties supported by the live zone reconfiguration for the selected brand. However, there are restrictions which apply to all brands.
The following resources and resource properties are not supported by the live zone reconfiguration for all brands:
brand zonename zonepath ip-type rootzpool
Any changes made to the listed resources and resource properties will cause the live zone reconfiguration fail if they are applied to the running zone.
The resources and resource properties listed below do not affect the running zone directly. For this reason, you can modify them in the persistent configuration at any time. But, any attempts to change them in the live configuration will be refused. This applies to all brands.
admin attr autoboot autoshutdown bootargs suspend
See attributes(5) for descriptions of the following attributes:
hostid(1), zlogin(1), zonename(1), beadm(1M), in.rlogind(1M), logadm(1M), solaris-kz(5), sshd(1M), sysdef(1M), zfs(1M), zoneadm(1M), zonecfg(1M), kill(2), priocntl(2), sysinfo(2), gethostid(3C), getzoneid(3C), ucred_get(3C), proc(4), attributes(5), brands(5), privileges(5), crgetzoneid(9F)