HP-UX Containers (SRP) A.03.01 Administrator's Guide

character device (for example, /dev/rdisk/disk10) within the system container as described in

15.4 Managing Devices . Since operations such as creating a file system on a device using the mkfs

command is not allowed within a system container, you must create the file system on a device before

provisioning it in the container.

NFS mount must be done within a container. Accessing the NFS mount points created within a

container from the global view will result in an error because the NFS mounts done within a system

container are exclusive to that container. NFS mounting from the global view to a path under a

container root directory (/var/hpsrp/container_name) is not supported. Mounting the global

view files via NFS is possible from within a system container by targeting the global view host IP

address when specifying the NFS resource, as the loopback IP interface is private to each system

container.

NFS server is supported only in the global view (using the primary system IP address). You must

configure directory shares as an absolute path from the global system root directory, and the

directories must be physically mounted when the shares are exported.

CacheFS is not supported in a system container.

15.1.2.3 Mount table within a system container

A system container is provisioned with its own version of mounted filesystem table mnttab under the

directory /var/hpsrp/container_name/etc. It contains a table of devices mounted by the

mount command for the container both by the global administrator and the administrator within the

container. The users in the global view have visibility to all the mounts created in the system inclusive

of all system container specific mounts.

When a system container is in started state, its mnttab will have at least the following two entries:

• A root mount point (/) associated with the system container of type srp_dummyvfs which is

automatically created when the container is started. This root entry cannot be mounted or

unmounted by a user.

• A mount point (/stand) lofs mounted to the system’s /stand directory, which gets mounted

by the global administrator when the container is started.

All mount point paths within a system container are always displayed as a path relative to the

container root path, while the same mount point paths are displayed as absolute paths in the global

view. For example, a mount point created as /var/hpsrp/container_name/stand from the

global view will be displayed as /var/hpsrp/container_name/stand in the global view and

as just /stand within the system container container_name.

Any mount point created from the global view for the system container, though visible within the

container, cannot be unmounted by the container administrator. In particular, all lofs mounts created

by the global administrator will display both the filesystem and the mounted on directory to be

identical. For the physical filesystem mounts like vxfs or hfs, the original block device path like

/dev/disk/disk48 is displayed as the filesystem even if the disk/device itself is not provisioned

within the system container. As mentioned earlier, NFS mounting from the global view to a path

under a container root is not supported. Even though the system container specific NFS mounts are

still visible in the global view, they still cannot be accessed from the global view.

A sample output from the bdf command executed within the system container is shown below:

# bdf

Filesystem kbytes used avail %used Mounted on

/ 8192000 4146427 3794023 52% /

/stand 1835008 535520 1289376 29% /stand