16 KiB
% OCI(1) OCI-RUNTIME-TOOL User Manuals % OCI Community % APRIL 2016
NAME
oci-runtime-tool-generate - Generate a config.json for an OCI container
SYNOPSIS
oci-runtime-tool generate [OPTIONS]
DESCRIPTION
oci-runtime-tool generate generates configuration JSON for an OCI bundle.
By default, it writes the JSON to stdout, but you can use --output
to direct it to a file. OCI-compatible runtimes like runC expect to
read the configuration from config.json.
OPTIONS
--apparmor=PROFILE Specifies the apparmor profile for the container
--arch=ARCH Architecture used within the container. "amd64"
--args=OPTION Arguments to run within the container. Can be specified multiple times. If you were going to run a command with multiple options, you would need to specify the command and each argument in order.
--args "/usr/bin/httpd" --args "-D" --args "FOREGROUND"
--bind=HOST-DIR:CONTAINER-DIR][:OPTIONS...
Bind mount directories src:dest:(rw,ro) If you specify, --bind /HOST-DIR:/CONTAINER-DIR, runc bind mounts /HOST-DIR in the host
to /CONTAINER-DIR in the OCI container. The OPTIONS are a colon
delimited list and can be any mount option support by the runtime such
as [rw|ro|rbind|bind|...]. The HOST_DIR and CONTAINER-DIR must be
absolute paths such as /src/docs. You can set the ro or rw
options to a bind-mount to mount it read-only or read-write mode,
respectively. By default, bind-mounts are mounted read-write.
--cap-add=[] Add Linux capabilities
--cap-drop=[] Drop Linux capabilities
--cap-drop-alltrue|false Drop all Linux capabilities
--cgroups-path="" Specifies the path to the cgroups relative to the cgroups mount point.
--cwd=PATH Current working directory for the process. The deafult is /.
--device-add=TYPE:MAJOR:MINOR:PATH[:OPTIONS...] Add a device file in container. e.g. --device=c:10:229:/dev/fuse:fileMode=438:uid=0:gid=0 The TYPE, MAJOR, MINOR, PATH are required. TYPE is the device type. The acceptable values are b (block), c (character), u (unbuffered), p (FIFO). MAJOR/MINOR is the major/minor device id. PATH is the device path. The fileMode, uid, gid are optional. fileMode is the file mode of the device file. uid/gid is the user/group id of the device file. This option can be specified multiple times.
--device-remove=PATH Remove a device file in container. This option can be specified multiple times.
--device-remove-all=true|false Remove all devices for linux inside the container. The default is false.
--disable-oom-kill=true|false Whether to disable OOM Killer for the container or not.
--env=[] Set environment variables e.g. key=value. This option allows you to specify arbitrary environment variables that are available for the process that will be launched inside of the container.
--env-file=[] Set environment variables from a file. This option sets environment variables in the container from the contents of a file formatted with key=value pairs, one per line. When specified multiple times, files are loaded in order with duplicate keys overwriting previous ones.
--gid=GID Gid for the process inside of container
--gidmappings=GIDMAPPINGS Add GIDMappings e.g HostID:ContainerID:Size. Implies -user=.
--groups=GROUP Supplementary groups for the processes inside of container
--help Print usage statement
--hostname="" Set the container host name that is available inside the container.
--label=[] Add annotations to the configuration e.g. key=value. Currently, key containing equals sign is not supported.
--linux-cpu-shares=CPUSHARES Specifies a relative share of CPU time available to the tasks in a cgroup.
--linux-cpu-period=CPUPERIOD Specifies a period of time in microseconds for how regularly a cgroup's access to CPU resources should be reallocated (CFS scheduler only).
--linux-cpu-quota=CPUQUOTA Specifies the total amount of time in microseconds for which all tasks in a cgroup can run during one period.
--linux-cpus=CPUS Sets the CPUs to use within the cpuset (default is to use any CPU available).
--linux-mem-kernel-limit=MEMKERNELLIMIT Sets the hard limit of kernel memory in bytes.
--linux-mem-kernel-tcp=MEMKERNELTCP Sets the hard limit of kernel TCP buffer memory in bytes.
--linux-mem-limit=MEMLIMIT Sets the limit of memory usage in bytes.
--linux-mem-reservation=MEMRESERVATION Sets the soft limit of memory usage in bytes.
--linux-mem-swap=MEMSWAP Sets the total memory limit (memory + swap) in bytes.
--linux-mem-swappiness=MEMSWAPPINESS Sets the swappiness of how the kernel will swap memory pages (Range from 0 to 100).
--linux-mems=MEMS Sets the list of memory nodes in the cpuset (default is to use any available memory node).
--linux-namespace-add=NSNAME[:PATH] Adds or replaces the given linux namespace NSNAME with a namespace entry that has a path of PATH. Omitting PATH means that a new namespace will be created by the container.
--linux-namespace-remove=NSNAME Removes a namespace from the set of namespaces configured in the container, so that the host's namespace will be used by the container instead of creating or joining another namespace.
--linux-namespace-remove-all Removes all namespaces from the set of namespaces configured for a container, such that the container will effectively run on the host.
--linux-network-classid=CLASSID Specifies network class identifier which will be tagged by container's network packets.
--linux-network-priorities=[] Specifies network priorities of network traffic, format is NAME:PRIORITY. e.g. --linux-network-priorities=eth0:123 This option can be specified multiple times. If a interface name was specified more than once, the last PRIORITY makes sense. The special PRIORITY -1 removes existing setting for interface NAME.
--linux-pids-limit=PIDSLIMIT Set maximum number of PIDs.
--linux-realtime-period=REALTIMEPERIOD Sets the CPU period to be used for realtime scheduling (in usecs). Same as --linux-cpu-period but applies to realtime scheduler only.
--linux-realtime-runtime=REALTIMERUNTIME Specifies a period of time in microseconds for the longest continuous period in which the tasks in a cgroup have access to CPU resources.
--masked-paths=[] Specifies paths can not be read inside container. e.g. --masked-paths=/proc/kcore This option can be specified multiple times.
--mount-cgroups=[rw|ro|no] Mount cgroups. The default is no.
--mount-label=MOUNTLABEL Mount Label Depending on your SELinux policy, you would specify a label that looks like this: "system_u:object_r:svirt_sandbox_file_t:s0:c1,c2"
Note you would want your ROOTFS directory to be labeled with a context that
this process type can use.
"system_u:object_r:usr_t:s0" might be a good label for a readonly container,
"system_u:system_r:svirt_sandbox_file_t:s0:c1,c2" for a read/write container.
--no-new-privileges=true|false Set no new privileges bit for the container process. Setting this flag will block the container processes from gaining any additional privileges using tools like setuid apps. It is a good idea to run unprivileged containers with this flag.
--oom-score-adj=adj Specifies oom_score_adj for the container.
--os=OS Operating system used within the container.
--output=PATH Instead of writing the configuration JSON to stdout, write it to a file at PATH (overwriting the existing content if a file already exists at PATH).
--poststart=CMD[:ARGS...] Set command to run in poststart hooks. Can be specified multiple times. The multiple commands will be run in order before the container process gets launched but after the container environment and main process has been created.
--poststop=CMD[:ARGS...] Set command to run in poststop hooks. Can be specified multiple times. The multiple commands will be run in order after the container process is stopped.
--prestart=CMD[:ARGS...] Set command to run in prestart hooks. Can be specified multiple times. The multiple commands will be run in order after the container process has been created but before it executes the user-configured code.
--privileged=true|false Give extended privileges to this container. The default is false.
By default, OCI containers are “unprivileged” (=false) and cannot do some of the things a normal root process can do.
When the operator executes oci-runtime-tool generate --privileged, OCI will enable access to all devices on the host as well as disable some of the confinement mechanisms like AppArmor, SELinux, and seccomp from blocking access to privileged processes. This gives the container processes nearly all the same access to the host as processes generating outside of a container on the host.
--readonly-paths=[] Specifies paths readonly inside container. e.g. --readonly-paths=/proc/sys This option can be specified multiple times.
--rootfs-path=ROOTFSPATH
Path to the rootfs, which can be an absolute path or relative to bundle path.
e.g the absolute path of rootfs is /to/bundle/rootfs, bundle path is /to/bundle,
then the value set as ROOTFSPATH should be /to/bundle/rootfs or rootfs. The default is rootfs.
--rootfs-propagation=PROPOGATIONMODE Mount propagation for root filesystem. Values are "shared, rshared, private, rprivate, slave, rslave"
--rootfs-readonly=true|false Mount the container's root filesystem as read only.
By default a container will have its root filesystem writable allowing processes to write files anywhere. By specifying the --rootfs-readonly flag the container will have its root filesystem mounted as read only prohibiting any writes.
--rlimits-add=[] Specifies resource limits, format is RLIMIT:HARD:SOFT. e.g. --rlimits-add=RLIMIT_NOFILE:1024:1024 This option can be specified multiple times. When same RLIMIT specified over once, the last one make sense.
--rlimits-remove=[] Remove the specified resource limits for process inside the container. This option can be specified multiple times.
--rlimits-remove-all=true|false Remove all resource limits for process inside the container. The default is false.
--seccomp-allow=SYSCALL Specifies syscalls to be added to the ALLOW list. See --seccomp-syscalls for setting limits on arguments.
--seccomp-arch=ARCH Specifies Additional architectures permitted to be used for system calls. By default if you turn on seccomp, only the host architecture will be allowed.
--seccomp-default=ACTION Specifies the the default action of Seccomp syscall restrictions and removes existing restrictions with the specified action Values: KILL,ERRNO,TRACE,ALLOW
--seccomp-default-force=ACTION Specifies the the default action of Seccomp syscall restrictions Values: KILL,ERRNO,TRACE,ALLOW
--seccomp-errno=SYSCALL Specifies syscalls to create seccomp rule to respond with ERRNO.
--seccomp-kill=SYSCALL Specifies syscalls to create seccomp rule to respond with KILL.
--seccomp-only Option to only export the seccomp section of output
--seccomp-remove Specifies syscall restrictions to remove from the configuration.
--seccomp-remove-all Option to remove all syscall restrictions.
--seccomp-trace=SYSCALL Specifies syscalls to create seccomp rule to respond with TRACE.
--seccomp-trap=SYSCALL Specifies syscalls to create seccomp rule to respond with TRAP.
--selinux-label=PROCESSLABEL SELinux Label Depending on your SELinux policy, you would specify a label that looks like this: "system_u:system_r:svirt_lxc_net_t:s0:c1,c2"
Note you would want your ROOTFS directory to be labeled with a context that
this process type can use.
"system_u:object_r:usr_t:s0" might be a good label for a readonly container,
"system_u:object_r:svirt_sandbox_file_t:s0:c1,c2" for a read/write container.
--sysctl=SYSCTLSETTING Add sysctl settings e.g net.ipv4.forward=1, only allowed if the syctl is namespaced.
--template=PATH Override the default template with your own. Additional options will only adjust the relevant portions of your template.
--tmpfs=[] Create a tmpfs mount
Mount a temporary filesystem (tmpfs) mount into a container, for example:
$ oci-runtime-tool generate -d --tmpfs /tmp:rw,size=787448k,mode=1777 my_image
This command mounts a `tmpfs` at `/tmp` within the container. The supported mount options are the same as the Linux default `mount` flags. If you do not specify any options, the systems uses the following options:
`rw,noexec,nosuid,nodev,size=65536k`.
--tty=true|false Allocate a new tty for the container process. The default is false.
--uid=UID Sets the UID used within the container.
--uidmappings Add UIDMappings e.g HostUID:ContainerID:Size. Implies --user=.
EXAMPLES
Generating container in read-only mode
During container image development, containers often need to write to the image content. Installing packages into /usr, for example. In production, applications seldom need to write to the image. Container applications write to volumes if they need to write to file systems at all. Applications can be made more secure by generating them in read-only mode using the --rootfs-readonly switch. This protects the containers image from modification. Read only containers may still need to write temporary data. The best way to handle this is to mount tmpfs directories on /generate and /tmp.
$ oci-runtime-tool generate --rootfs-readonly --tmpfs /generate --tmpfs /tmp --tmpfs /run --rootfs-path /var/lib/containers/fedora --args bash
Exposing log messages from the container to the host's log
If you want messages that are logged in your container to show up in the host's syslog/journal then you should bind mount the /dev/log directory as follows.
$ oci-runtime-tool generate --bind /dev/log:/dev/log --rootfs-path /var/lib/containers/fedora --args bash
From inside the container you can test this by sending a message to the log.
(bash)# logger "Hello from my container"
Then exit and check the journal.
# exit
# journalctl -b | grep Hello
This should list the message sent to logger.
Bind Mounting External Volumes
To mount a host directory as a container volume, specify the absolute path to the directory and the absolute path for the container directory separated by a colon:
$ oci-runtime-tool generate --bind /var/db:/data1 --rootfs-path /var/lib/containers/fedora --args bash
Using SELinux
You can use SELinux to add security to the container. You must specify the process label to run the init process inside of the container using the --selinux-label.
$ oci-runtime-tool generate --bind /var/db:/data1 --selinux-label system_u:system_r:svirt_lxc_net_t:s0:c1,c2 --mount-label system_u:object_r:svirt_sandbox_file_t:s0:c1,c2 --rootfs-path /var/lib/containers/fedora --args bash
Not in the above example we used a type of svirt_lxc_net_t and an MCS Label of s0:c1,c2. If you want to guarantee separation between containers, you need to make sure that each container gets launched with a different MCS Label pair.
Also the underlying rootfs must be labeled with a matching label. For the example above, you would execute a command like:
# chcon -R system_u:object_r:svirt_sandbox_file_t:s0:c1,c2 /var/lib/containers/fedora
This will set up the labeling of the rootfs so that the process launched would be able to write to the container. If you wanted to only allow it to read/execute the content in rootfs, you could execute:
# chcon -R system_u:object_r:usr_t:s0 /var/lib/containers/fedora
When using SELinux, be aware that the host has no knowledge of container SELinux
policy. Therefore, in the above example, if SELinux policy is enforced, the
/var/db directory is not writable to the container. A "Permission Denied"
message will occur and an avc: message in the host's syslog.
To work around this, the following command needs to be generate in order for the proper SELinux policy type label to be attached to the host directory:
# chcon -Rt svirt_sandbox_file_t -l s0:c1,c2 /var/db
Now, writing to the /data1 volume in the container will be allowed and the changes will also be reflected on the host in /var/db.
SEE ALSO
runc(1), oci-runtime-tool(1)
HISTORY
April 2016, Originally compiled by Dan Walsh (dwalsh at redhat dot com)