Docker Driver

Name: docker

The docker driver provides a first-class Docker workflow on Nomad. The Docker driver handles downloading containers, mapping ports, and starting, watching, and cleaning up after containers.

Task Configuration

task "webservice" {  driver = "docker"   config {    image = "redis:7"    labels {      group = "webservice-cache"    }  }}

The docker driver supports the following configuration in the job spec. Only image is required.

• image - The Docker image to run. The image may include a tag or custom URL and should include https:// if required. By default it will be fetched from Docker Hub. If the tag is omitted or equal to latest the driver will always try to pull the image. If the image to be pulled exists in a registry that requires authentication credentials must be provided to Nomad. Please see the Authentication section.

  config {  image = "https://hub.docker.internal/redis:7"}

• image_pull_timeout - (Optional) A time duration that controls how long Nomad will wait before cancelling an in-progress pull of the Docker image as specified in image. Defaults to "5m".

• args - (Optional) A list of arguments to the optional command. If no command is specified, the arguments are passed directly to the container. References to environment variables or any interpretable Nomad variables will be interpreted before launching the task. For example:

  config {  args = [    "-bind", "${NOMAD_PORT_http}",    "${nomad.datacenter}",    "${MY_ENV}",    "${meta.foo}",  ]}

• auth - (Optional) Provide authentication for a private registry (see below).

• auth_soft_fail (bool: false) - Don't fail the task on an auth failure. Attempt to continue without auth. If the Nomad client configuration has an auth.helper block, the helper will be tried for all images, including public images. If you mix private and public images, you will need to include auth_soft_fail=true in every job using a public image.

• command - (Optional) The command to run when starting the container.

  config {  command = "my-command"}

• container_exists_attempts - (Optional) A number of attempts to be made to purge a container if during task creation Nomad encounters an existing one in non-running state for the same task. Defaults to 5.

• dns_search_domains - (Optional) A list of DNS search domains for the container to use. If you are using bridge networking mode with a network block in the task group, you must set all DNS options in the network.dns block instead.

• dns_options - (Optional) A list of DNS options for the container to use. If you are using bridge networking mode with a network block in the task group, you must set all DNS options in the network.dns block instead.

• dns_servers - (Optional) A list of DNS servers for the container to use (e.g. ["8.8.8.8", "8.8.4.4"]). Requires Docker v1.10 or greater. If you are using bridge networking mode with a network block in the task group, you must set all DNS options in the network.dns block instead.

• entrypoint - (Optional) A string list overriding the image's entrypoint.

• extra_hosts - (Optional) A list of hosts, given as host:IP, to be added to /etc/hosts. This option may not work as expected in bridge network mode when there is more than one task within the same group. Refer to the upgrade guide for more information.

• force_pull - (Optional) true or false (default). Always pull most recent image instead of using existing local image. Should be set to true if repository tags are mutable. If image's tag is latest or omitted, the image will always be pulled regardless of this setting.

• group_add - (Optional) A list of supplementary groups to be applied to the container user.

• healthchecks - (Optional) A configuration block for controlling how the docker driver manages HEALTHCHECK directives built into the container. Set healthchecks.disable to disable any built-in healthcheck.

  config {  healthchecks {    disable = true  }}

• hostname - (Optional) The hostname to assign to the container. When launching more than one of a task (using count) with this option set, every container the task starts will have the same hostname.

• init - (Optional) true or false (default). Enable init (tini) system when launching your container. When enabled, an init process will be used as the PID1 in the container. Specifying an init process ensures the usual responsibilities of an init system, such as reaping zombie processes, are performed inside the created container.

  The default init process used is the first docker-init executable found in the system path of the Docker daemon process. This docker-init binary, included in the default installation, is backed by tini.

• interactive - (Optional) true or false (default). Keep STDIN open on the container.

• isolation - (Optional) Specifies Windows isolation mode: "hyperv" or "process". Defaults to "hyperv".

• sysctl - (Optional) A key-value map of sysctl configurations to set to the containers on start.

  config {  sysctl = {    "net.core.somaxconn" = "16384"  }}

• ulimit - (Optional) A key-value map of ulimit configurations to set to the containers on start.

  config {  ulimit {    nproc = "4242"    nofile = "2048:4096"  }}

• privileged - (Optional) true or false (default). Privileged mode gives the container access to devices on the host. Note that this also requires the nomad agent and docker daemon to be configured to allow privileged containers.

• ipc_mode - (Optional) The IPC mode to be used for the container. The default is none for a private IPC namespace. Other values are host for sharing the host IPC namespace or the name or id of an existing container. Note that it is not possible to refer to Docker containers started by Nomad since their names are not known in advance. Note that setting this option also requires the Nomad agent to be configured to allow privileged containers.

• ipv4_address - (Optional) The IPv4 address to be used for the container when using user defined networks. Requires Docker 1.13 or greater.

• ipv6_address - (Optional) The IPv6 address to be used for the container when using user defined networks. Requires Docker 1.13 or greater.

• labels - (Optional) A key-value map of labels to set to the containers on start.

  config {  labels {    foo = "bar"    zip = "zap"  }}

  A more verbose syntax must be used to specify labels with keys that require quoting. For example DataDog's autodiscovery mechanism looks for labels with dots in the key which must be quoted:

  config {  labels = [    {      "com.datadoghq.ad.check_names"  =  "[\"openmetrics\"]"      "com.datadoghq.ad.init_configs" = "[{}]"    }  ]}

• load - (Optional) Load an image from a tar archive file instead of from a remote repository. Equivalent to the docker load -i <filename> command. If you're using an artifact block to fetch the archive file, you'll need to ensure that Nomad keeps the archive intact after download.

  artifact {  source = "http://path.to/redis.tar"  options {    archive = false  }}config {  load = "redis.tar"  image = "redis"}

• logging - (Optional) A key-value map of Docker logging options. Defaults to json-file with log rotation (max-file=2 and max-size=2m).

  config {  logging {    type = "fluentd"    config {      fluentd-address = "localhost:24224"      tag = "your_tag"    }  }}

• mac_address - (Optional) The MAC address for the container to use (e.g. "02:68:b3:29:da:98").

• memory_hard_limit - (Optional) The maximum allowable amount of memory used (megabytes) by the container. If set, the memory parameter of the task resource configuration becomes a soft limit passed to the docker driver as --memory_reservation, and memory_hard_limit is passed as the --memory hard limit. When the host is under memory pressure, the behavior of soft limit activation is governed by the Kernel.

• network_aliases - (Optional) A list of network-scoped aliases, provide a way for a container to be discovered by an alternate name by any other container within the scope of a particular network. Network-scoped alias is supported only for containers in user defined networks

  config {  network_mode = "user-network"  network_aliases = [    "${NOMAD_TASK_NAME}",    "${NOMAD_TASK_NAME}-${NOMAD_ALLOC_INDEX}"  ]}

• network_mode - (Optional) The network mode to be used for the container. In order to support userspace networking plugins in Docker 1.9 this accepts any value. The default is bridge for all operating systems but Windows, which defaults to nat. Other networking modes may not work without additional configuration on the host (which is outside the scope of Nomad). Valid values pre-docker 1.9 are default, bridge, host, none, or container:name.

  The default network_mode for tasks that use group networking in bridge mode will be container:<name>, where the name is the container name of the parent container used to share network namespaces between tasks. If you set the group network.mode to "bridge" you should not set this Docker network_mode config, otherwise the container will be unable to reach other containers in the task group. This will also prevent Connect-enabled tasks from reaching the Envoy sidecar proxy. You must also set any DNS options in the network.dns block and not in the task configuration.

  If you are in the process of migrating from the default Docker network to group-wide bridge networking, you may encounter issues preventing your containers from reaching networks outside of the bridge interface on systems with firewalld enabled. This behavior is often caused by the CNI plugin not registering the group network as trusted and can be resolved as described in the network block documentation.

• oom_score_adj - (Optional) A positive integer to indicate the likelihood of the task being OOM killed (valid only for Linux). Defaults to 0.

• pid_mode - (Optional) host or not set (default). Set to host to share the PID namespace with the host. Note that this also requires the Nomad agent to be configured to allow privileged containers. See below for more details.

• ports - (Optional) A list of port labels to map into the container (see below).

• port_map - (Optional) Deprecated A key-value map of port labels (see below).

• security_opt - (Optional) A list of string flags to pass directly to --security-opt. For example:

  config {  security_opt = [    "credentialspec=file://gmsaUser.json",  ]}

• shm_size - (Optional) The size (bytes) of /dev/shm for the container.

• storage_opt - (Optional) A key-value map of storage options set to the containers on start. This overrides the host dockerd configuration. For example:

  config {  storage_opt = {    size = "40G"  }}

• tty - (Optional) true or false (default). Allocate a pseudo-TTY for the container.

• uts_mode - (Optional) host or not set (default). Set to host to share the UTS namespace with the host. Note that this also requires the Nomad agent to be configured to allow privileged containers.

• userns_mode - (Optional) host or not set (default). Set to host to use the host's user namespace (effectively disabling user namespacing) when user namespace remapping is enabled on the docker daemon. This field has no effect if the docker daemon does not have user namespace remapping enabled.

• volumes - (Optional) A list of host_path:container_path strings to bind host paths to container paths. Mounting host paths outside of the allocation working directory is prevented by default and limits volumes to directories that exist inside the allocation working directory. You can allow mounting host paths outside of the allocation working directory on individual clients by setting the docker.volumes.enabled option to true in the client's configuration. We recommend using mount if you wish to have more control over volume definitions.

  config {  volumes = [    # Use absolute paths to mount arbitrary paths on the host    "/path/on/host:/path/in/container",     # Use relative paths to rebind paths already in the allocation dir    "relative/to/task:/also/in/container"  ]}

• volume_driver - (Optional) The name of the volume driver used to mount volumes. Must be used along with volumes. If volume_driver is omitted, then relative paths will be mounted from inside the allocation dir. If a "local" or other driver is used, then they may be named volumes instead. If docker.volumes.enabled is false then volume drivers and paths outside the allocation directory are disallowed.

  config {  volumes = [    # Use named volume created outside nomad.    "name-of-the-volume:/path/in/container"  ]  # Name of the Docker Volume Driver used by the container  volume_driver = "pxd"}

• work_dir - (Optional) The working directory inside the container.

• mount - Since 1.0.1 (Optional) Specify a mount to be mounted into the container. Volume, bind, and tmpfs type mounts are supported. May be specified multiple times.

  config {  # sample volume mount  mount {    type = "volume"    target = "/path/in/container"    source = "name-of-volume"    readonly = false    volume_options {      no_copy = false      labels {        foo = "bar"      }      driver_config {        name = "pxd"        options {          foo = "bar"        }      }    }  }   # sample bind mount  mount {    type = "bind"    target = "/path/in/container"    source = "/path/in/host"    readonly = false    bind_options {      propagation = "rshared"    }  }   # sample tmpfs mount  mount {    type = "tmpfs"    target = "/path/in/container"    readonly = false    tmpfs_options {      size = 100000 # size in bytes    }  }}

• mounts - (deprecated: Replaced by mount in 1.0.1) (Optional) A list of mounts to be mounted into the container. Volume, bind, and tmpfs type mounts are supported.

  config {  mounts = [    # sample volume mount    {      type = "volume"      target = "/path/in/container"      source = "name-of-volume"      readonly = false      volume_options = {        no_copy = false        labels = {          foo = "bar"        }        driver_config = {          name = "pxd"          options = {            foo = "bar"          }        }      }    },    # sample bind mount    {      type = "bind"      target = "/path/in/container"      source = "/path/in/host"      readonly = false      bind_options = {        propagation = "rshared"      }    },    # sample tmpfs mount    {      type = "tmpfs"      target = "/path/in/container"      readonly = false      tmpfs_options = {        size = 100000 # size in bytes      }    }  ]}

• devices - (Optional) A list of devices to be exposed the container. host_path is the only required field. By default, the container will be able to read, write and mknod these devices. Use the optional cgroup_permissions field to restrict permissions.

  config {  devices = [    {      host_path = "/dev/sda1"      container_path = "/dev/xvdc"      cgroup_permissions = "r"    },    {      host_path = "/dev/sda2"      container_path = "/dev/xvdd"    }  ]}

• cap_add - (Optional) A list of Linux capabilities as strings to pass directly to --cap-add. Effective capabilities (computed from cap_add and cap_drop) must be a subset of the allowed capabilities configured with the allow_caps plugin option key in the client node's configuration. Note that "all" is not permitted here if the allow_caps field in the driver configuration doesn't also allow all capabilities. For example:

config {  cap_add = ["net_raw", "sys_time"]}

• cap_drop - (Optional) A list of Linux capabilities as strings to pass directly to --cap-drop. Effective capabilities (computed from cap_add and cap_drop) must be a subset of the allowed capabilities configured with the allow_caps plugin option key in the client node's configuration. For example:

config {  cap_drop = ["mknod"]}

• cpu_hard_limit - (Optional) true or false (default). Use hard CPU limiting instead of soft limiting. By default this is false which means soft limiting is used and containers are able to burst above their CPU limit when there is idle capacity.

• cpu_cfs_period - (Optional) An integer value that specifies the duration in microseconds of the period during which the CPU usage quota is measured. The default is 100000 (0.1 second) and the maximum allowed value is 1000000 (1 second). See here for more details.

• advertise_ipv6_address - (Optional) true or false (default). Use the container's IPv6 address (GlobalIPv6Address in Docker) when registering services and checks. See IPv6 Docker containers for details.

• readonly_rootfs - (Optional) true or false (default). Mount the container's filesystem as read only.

• runtime - (Optional) A string representing a configured runtime to pass to docker. This is equivalent to the --runtime argument in the docker CLI For example, to use gVisor:

  config {  # gVisor runtime is runsc  runtime = "runsc"}

• pids_limit - (Optional) An integer value that specifies the pid limit for the container. Defaults to unlimited.

Additionally, the docker driver supports customization of the container's user through the task's user option.

Container Name

Nomad creates a container after pulling an image. Containers are named {taskName}-{allocId}. This is necessary in order to place more than one container from the same task on a host (e.g. with count > 1). This also means that each container's name is unique across the cluster.

This is not configurable.

Authentication

If you want to pull from a private repo (for example on dockerhub or quay.io), you will need to specify credentials in your job via:

• the auth option in the task config.

• by storing explicit repository credentials or by specifying Docker credHelpers in a file and setting the auth config value on the client in the plugin options.

• by specifying an auth helper on the client in the plugin options.

The auth object supports the following keys:

• username - (Optional) The account username.

• password - (Optional) The account password.

• email - (Optional) The account email.

• server_address - (Optional) The server domain/IP without the protocol. Docker Hub is used by default.

Example task-config:

task "example" {  driver = "docker"   config {    image = "secret/service"     auth {      username = "dockerhub_user"      password = "dockerhub_password"    }  }}

Example docker-config, using two helper scripts in $PATH, "docker-credential-ecr-login" and "docker-credential-vault":

{  "auths": {    "internal.repo": {      "auth": "`echo -n '<username>:<password>' | base64 -w0`"    }  },  "credHelpers": {    "<XYZ>.dkr.ecr.<region>.amazonaws.com": "ecr-login"  },  "credsStore": "secretservice"}

Example agent configuration, using a helper script "docker-credential-ecr-login" in $PATH

client {  enabled = true} plugin "docker" {  config {    auth {      # Nomad will prepend "docker-credential-" to the helper value and call      # that script name.      helper = "ecr-login"    }  }}

Insecure Registries

In order to pull images from a registry without TLS, you must configure the Docker daemon's insecure-registries flag. No additional Nomad client configuration is required. You should only allow insecure registries for registries running locally on the client or when the communication to the registry is otherwise encrypted. List the insecure-registries flag in the dockerd configuration file.

{  "insecure-registries": ["example.local:5000"]}

Networking

Docker supports a variety of networking configurations, including using host interfaces, SDNs, etc. Nomad uses bridged networking by default, like Docker.

You can specify other networking options, including custom networking plugins in Docker 1.9. You may need to perform additional configuration on the host in order to make these work. This additional configuration is outside the scope of Nomad.

Allocating Ports

You can allocate ports to your task using the port syntax described on the networking page. Here is a recap:

group {  network {    port "http" {}    port "https" {}  }  task "example" {    driver = "docker"    config {      ports = ["http", "https"]    }  }}

Forwarding and Exposing Ports

A Docker container typically specifies which port a service will listen on by specifying the EXPOSE directive in the Dockerfile.

Because dynamic ports will not match the ports exposed in your Dockerfile, Nomad will automatically expose any ports specified in the ports field.

These ports will be identified via environment variables. For example:

group {  network {    port "http" {}  }  task "api" {    driver = "docker"    config {      ports = ["http"]    }  }}

If Nomad allocates port 23332 to your api task for http, 23332 will be automatically exposed and forwarded to your container, and the driver will set an environment variable NOMAD_PORT_http with the value 23332 that you can read inside your container.

This provides an easy way to use the host networking option for better performance.

Using the Port Map

If you prefer to use the traditional port-mapping method, you can specify the the to field in the port configuration. It looks like this:

group "example" {  network {    port "redis" { to = 6379 }  }  task "example" {    driver = "docker"     config {      image = "redis"      ports = ["redis"]    }  }}

If Nomad allocates port 23332 to your allocation, the Docker driver will automatically setup the port mapping from 23332 on the host to 6379 in your container, so it will just work.

Note that by default this only works with bridged networking mode. It may also work with custom networking plugins which implement the same API for expose and port forwarding.

Deprecated port_map Syntax

Up until Nomad 0.12, ports could be specified in a task's resource block and set using the docker port_map field. As more features have been added to the group network resource allocation, task based network resources are deprecated. With it the port_map field is also deprecated and can only be used with task network resources.

Users should migrate their jobs to define ports in the group network block and specified which ports a task maps with the ports field.

Advertising Container IPs

When using network plugins like weave that assign containers a routable IP address, that address will automatically be used in any service advertisements for the task. You may override what address is advertised by using the address_mode parameter on a service. See service for details.

Networking Protocols

The Docker driver configures ports on both the tcp and udp protocols.

This is not configurable.

Other Networking Modes

Some networking modes like container or none will require coordination outside of Nomad. First-class support for these options may be improved later through Nomad plugins or dynamic job configuration.

Capabilities

The docker driver implements the following capabilities.

Client Requirements

Nomad requires Docker to be installed and running on the host alongside the Nomad agent.

By default Nomad communicates with the Docker daemon using the daemon's Unix socket. Nomad will need to be able to read/write to this socket. If you do not run Nomad as root, make sure you add the Nomad user to the Docker group so Nomad can communicate with the Docker daemon.

For example, on Ubuntu you can use the usermod command to add the nomad user to the docker group so you can run Nomad without root:

$ sudo usermod -G docker -a nomad

Nomad clients manage a cpuset cgroup for each task to reserve or share CPU cores. In order for Nomad to be compatible with Docker's own cgroups management, it must write to cgroups owned by Docker, which requires running as root. If Nomad is not running as root, CPU isolation and NUMA-aware scheduling will not function correctly for workloads with resources.cores, including workloads using task drivers other than docker on the same host.

For the best performance and security features you should use recent versions of the Linux Kernel and Docker daemon.

If you would like to change any of the options related to the docker driver on a Nomad client, you can modify them with the plugin block syntax. Below is an example of a configuration (many of the values are the default). See the next section for more information on the options.

plugin "docker" {  config {    endpoint = "unix:///var/run/docker.sock"     auth {      config = "/etc/docker-auth.json"      helper = "ecr-login"    }     tls {      cert = "/etc/nomad/nomad.pub"      key  = "/etc/nomad/nomad.pem"      ca   = "/etc/nomad/nomad.cert"    }     extra_labels = ["job_name", "job_id", "task_group_name", "task_name", "namespace", "node_name", "node_id"]     gc {      image       = true      image_delay = "3m"      container   = true       dangling_containers {        enabled        = true        dry_run        = false        period         = "5m"        creation_grace = "5m"      }    }     volumes {      enabled      = true      selinuxlabel = "z"    }     allow_privileged = false    allow_caps       = ["chown", "net_raw"]  }}

Plugin Options

• endpoint - If using a non-standard socket, HTTP or another location, or if TLS is being used, docker.endpoint must be set. If unset, Nomad will attempt to instantiate a Docker client using the DOCKER_HOST environment variable and then fall back to the default listen address for the given operating system. Defaults to unix:///var/run/docker.sock on Unix platforms and npipe:////./pipe/docker_engine for Windows.

• allow_privileged - Defaults to false. Changing this to true will allow containers to use privileged mode, which gives the containers full access to the host's devices. Note that you must set a similar setting on the Docker daemon for this to work.

• pull_activity_timeout - Defaults to 2m. If Nomad receives no communication from the Docker engine during an image pull within this timeframe, Nomad will time out the request that initiated the pull command. (Minimum of 1m)

• pids_limit - Defaults to unlimited (0). An integer value that specifies the pid limit for all the Docker containers running on that Nomad client. You can override this limit by setting pids_limit in your task config. If this value is greater than 0, your task pids_limit must be less than or equal to the value defined here.

• allow_caps - A list of allowed Linux capabilities. Defaults to

["audit_write", "chown", "dac_override", "fowner", "fsetid", "kill", "mknod", "net_bind_service", "setfcap", "setgid", "setpcap", "setuid", "sys_chroot"]

which is the same list of capabilities allowed by docker by default (without NET_RAW). Allows the operator to control which capabilities can be obtained by tasks using cap_add and cap_drop options. Supports the value "all" as a shortcut for allow-listing all capabilities supported by the operating system. Note that due to a limitation in Docker, tasks running as non-root users cannot expand the capabilities set beyond the default. They can only have their capabilities reduced.

• allow_runtimes - defaults to ["runc", "nvidia"] - A list of the allowed docker runtimes a task may use.

• auth block:

  • config - Allows an operator to specify a JSON file which is in the dockercfg format containing authentication information for a private registry, from either (in order) auths, credsStore or credHelpers.

  • helper - Allows an operator to specify a credsStore like script on $PATH to lookup authentication information from external sources. The script's name must begin with docker-credential- and this option should include only the basename of the script, not the path.

    If you set an auth helper, it will be tried for all images, including public images. If you mix private and public images, you will need to include auth_soft_fail=true in every job using a public image.

• tls block:

  • cert - Path to the server's certificate file (.pem). Specify this along with key and ca to use a TLS client to connect to the docker daemon. endpoint must also be specified or this setting will be ignored.

  • key - Path to the client's private key (.pem). Specify this along with cert and ca to use a TLS client to connect to the docker daemon. endpoint must also be specified or this setting will be ignored.

  • ca - Path to the server's CA file (.pem). Specify this along with cert and key to use a TLS client to connect to the docker daemon. endpoint must also be specified or this setting will be ignored.

• disable_log_collection - Defaults to false. Setting this to true will disable Nomad logs collection of Docker tasks. If you don't rely on nomad log capabilities and exclusively use host based log aggregation, you may consider this option to disable nomad log collection overhead.

• extra_labels - Extra labels to add to Docker containers. Available options are job_name, job_id, task_group_name, task_name, namespace, node_name, node_id. Globs are supported (e.g. task*)

• logging block:

  • type - Defaults to "json-file". Specifies the logging driver docker should use for all containers Nomad starts. Note that for older versions of Docker, only json-file file or journald will allow Nomad to read the driver's logs via the Docker API, and this will prevent commands such as nomad alloc logs from functioning.

  • config - Defaults to { max-file = "2", max-size = "2m" }. This option can also be used to pass further configuration to the logging driver.

• gc block:

  • image - Defaults to true. Changing this to false will prevent Nomad from removing images from stopped tasks.

  • image_delay - A time duration, as defined here, that defaults to 3m. The delay controls how long Nomad will wait between an image being unused and deleting it. If a task is received that uses the same image within the delay, the image will be reused. If an image is referenced by more than one tag, image_delay may not work correctly.

  • container - Defaults to true. This option can be used to disable Nomad from removing a container when the task exits. Under a name conflict, Nomad may still remove the dead container.

  • dangling_containers block for controlling dangling container detection and cleanup:

    • enabled - Defaults to true. Enables dangling container handling.

    • dry_run - Defaults to false. Only log dangling containers without cleaning them up.

    • period - Defaults to "5m". A time duration that controls interval between Nomad scans for dangling containers.

    • creation_grace - Defaults to "5m". Grace period after a container is created during which the GC ignores it. Only used to prevent the GC from removing newly created containers before they are registered with the GC. Should not need adjusting higher but may be adjusted lower to GC more aggressively.

• volumes block:

  • enabled - Defaults to false. Allows tasks to bind host paths (volumes) inside their container and use volume drivers (volume_driver). Binding relative paths is always allowed and will be resolved relative to the allocation's directory.

  • selinuxlabel - Allows the operator to set a SELinux label to the allocation and task local bind-mounts to containers. If used with docker.volumes.enabled set to false, the labels will still be applied to the standard binds in the container.

• infra_image - This is the Docker image to use when creating the parent container necessary when sharing network namespaces between tasks. Defaults to registry.k8s.io/pause-<goarch>:3.3. The image will only be pulled from the container registry if its tag is latest or the image doesn't yet exist locally.

• infra_image_pull_timeout - A time duration that controls how long Nomad will wait before cancelling an in-progress pull of the Docker image as specified in infra_image. Defaults to "5m".

• windows_allow_insecure_container_admin - Indicates that on windows, docker checks the task.user field or, if unset, the container image manifest after pulling the container, to see if it's running as ContainerAdmin. If so, exits with an error unless the task config has privileged=true. Defaults to false.

Client Configuration

The docker driver has the following client configuration options:

• docker.endpoint - If using a non-standard socket, HTTP or another location, or if TLS is being used, docker.endpoint must be set. If unset, Nomad will attempt to instantiate a Docker client using the DOCKER_HOST environment variable and then fall back to the default listen address for the given operating system. Defaults to unix:///var/run/docker.sock on Unix platforms and npipe:////./pipe/docker_engine for Windows.

• docker.auth.config - Allows an operator to specify a JSON file which is in the dockercfg format containing authentication information for a private registry, from either (in order) auths, credsStore or credHelpers.

• docker.auth.helper - Allows an operator to specify a credsStore -like script on \$PATH to lookup authentication information from external sources. The script's name must begin with docker-credential- and this option should include only the basename of the script, not the path.

• docker.tls.cert - Path to the server's certificate file (.pem). Specify this along with docker.tls.key and docker.tls.ca to use a TLS client to connect to the docker daemon. docker.endpoint must also be specified or this setting will be ignored.

• docker.tls.key - Path to the client's private key (.pem). Specify this along with docker.tls.cert and docker.tls.ca to use a TLS client to connect to the docker daemon. docker.endpoint must also be specified or this setting will be ignored.

• docker.tls.ca - Path to the server's CA file (.pem). Specify this along with docker.tls.cert and docker.tls.key to use a TLS client to connect to the docker daemon. docker.endpoint must also be specified or this setting will be ignored.

• docker.cleanup.image Defaults to true. Changing this to false will prevent Nomad from removing images from stopped tasks.

• docker.cleanup.image.delay A time duration, as defined here, that defaults to 3m. The delay controls how long Nomad will wait between an image being unused and deleting it. If a tasks is received that uses the same image within the delay, the image will be reused.

• docker.volumes.enabled: Defaults to false. Allows tasks to bind host paths (volumes) inside their container and use volume drivers (volume_driver). Binding relative paths is always allowed and will be resolved relative to the allocation's directory.

• docker.volumes.selinuxlabel: Allows the operator to set a SELinux label to the allocation and task local bind-mounts to containers. If used with docker.volumes.enabled set to false, the labels will still be applied to the standard binds in the container.

• docker.privileged.enabled Defaults to false. Changing this to true will allow containers to use privileged mode, which gives the containers full access to the host's devices. Note that you must set a similar setting on the Docker daemon for this to work.

• docker.caps.allowlist: A list of allowed Linux capabilities. Defaults to "CHOWN,DAC_OVERRIDE,FSETID,FOWNER,MKNOD,NET_RAW,SETGID,SETUID,SETFCAP, SETPCAP,NET_BIND_SERVICE,SYS_CHROOT,KILL,AUDIT_WRITE", which is the list of capabilities allowed by docker by default, as defined here. Allows the operator to control which capabilities can be obtained by tasks using cap_add and cap_drop options. Supports the value "ALL" as a shortcut for allowlisting all capabilities.

• docker.cleanup.container: Defaults to true. This option can be used to disable Nomad from removing a container when the task exits. Under a name conflict, Nomad may still remove the dead container.

• docker.nvidia_runtime: Defaults to nvidia. This option allows operators to select the runtime that should be used in order to expose Nvidia GPUs to the container.

Note: When testing or using the -dev flag you can use DOCKER_HOST, DOCKER_TLS_VERIFY, and DOCKER_CERT_PATH to customize Nomad's behavior. If docker.endpoint is set Nomad will only read client configuration from the config file.

An example is given below:

client {  options {    "docker.cleanup.image" = "false"  }}

Client Attributes

The docker driver will set the following client attributes:

• driver.docker - This will be set to "1", indicating the driver is available.

• driver.docker.bridge_ip - The IP of the Docker bridge network if one exists.

• driver.docker.version - This will be set to version of the docker server.

Here is an example of using these properties in a job file:

job "docs" {  # Require docker version higher than 1.2.  constraint {    attribute = "${attr.driver.docker.version}"    operator  = ">"    version   = "1.2"  }}

Resource Isolation

CPU

Nomad limits containers' CPU based on CPU shares. CPU shares allow containers to burst past their CPU limits. CPU limits will only be imposed when there is contention for resources. When the host is under load your process may be throttled to stabilize QoS depending on how many shares it has. You can see how many CPU shares are available to your process by reading NOMAD_CPU_LIMIT. 1000 shares are approximately equal to 1 GHz.

Please keep the implications of CPU shares in mind when you load test workloads on Nomad.

If resources cores is set, the task is given an isolated reserved set of CPU cores to make use of. The total set of cores the task may run on is the private set combined with the variable set of unreserved cores. The private set of CPU cores is available to your process by reading NOMAD_CPU_CORES.

Memory

Nomad limits containers' memory usage based on total virtual memory. This means that containers scheduled by Nomad cannot use swap. This is to ensure that a swappy process does not degrade performance for other workloads on the same host.

Since memory is not an elastic resource, you will need to make sure your container does not exceed the amount of memory allocated to it, or it will be terminated or crash when it tries to malloc. A process can inspect its memory limit by reading NOMAD_MEMORY_LIMIT, but will need to track its own memory usage. Memory limit is expressed in megabytes so 1024 = 1 GB.

IO

Nomad's Docker integration does not currently provide QoS around network or filesystem IO. These will be added in a later release.

Security

Docker provides resource isolation by way of cgroups and namespaces. Containers essentially have a virtual file system all to themselves. If you need a higher degree of isolation between processes for security or other reasons, it is recommended to use full virtualization like QEMU.

Caveats

Dangling Containers

Nomad has a detector and a reaper for dangling Docker containers, containers that Nomad starts yet does not manage or track. Though rare, they lead to unexpectedly running services, potentially with stale versions.

When Docker daemon becomes unavailable as Nomad starts a task, it is possible for Docker to successfully start the container but return a 500 error code from the API call. In such cases, Nomad retries and eventually aims to kill such containers. However, if the Docker Engine remains unhealthy, subsequent retries and stop attempts may still fail, and the started container becomes a dangling container that Nomad no longer manages.

The newly added reaper periodically scans for such containers. It only targets containers with a com.hashicorp.nomad.allocation_id label, or match Nomad's conventions for naming and bind-mounts (i.e. /alloc, /secrets, local). Containers that don't match Nomad container patterns are left untouched.

Operators can run the reaper in a dry-run mode, where it only logs dangling container ids without killing them, or disable it by setting the gc.dangling_containers config block.

Docker for Windows

Docker for Windows only supports running Windows containers. Because Docker for Windows is relatively new and rapidly evolving you may want to consult the list of relevant issues on GitHub.