Launches "temporary" Jupyter notebook servers.
tmpnb launches a docker container for each user that requests one. In practice, this gets used to provide temporary notebooks, demo the IPython notebook as part of a Nature article, or even provide Jupyter kernels for publications.
People have used it at user groups, meetups, and workshops to provide temporary access to a full system without any installation whatsoever.
Get Docker, then:
docker pull jupyter/minimal-notebook
export TOKEN=$( head -c 30 /dev/urandom | xxd -p )
docker run --net=host -d -e CONFIGPROXY_AUTH_TOKEN=$TOKEN --name=proxy jupyter/configurable-http-proxy --default-target http://127.0.0.1:9999
docker run --net=host -d -e CONFIGPROXY_AUTH_TOKEN=$TOKEN --name=tmpnb -v /var/run/docker.sock:/docker.sock jupyter/tmpnb python orchestrate.py --container-user=jovyan --command='jupyter notebook --no-browser --port {port} --ip=0.0.0.0 --NotebookApp.base_url={base_path} --NotebookApp.port_retries=0 --NotebookApp.token="" --NotebookApp.disable_check_xsrf=True'
NOTE! This will disable Jupyter Notebook's token security. The option for container-user assures that the notebook will not run as a privileged user. You can set --NotebookApp.token to a string if you want to add a minimal layer of security.
BAM! Visit your Docker host on port 8000 and you have a working tmpnb setup. The -v /var/run/docker.sock:/docker.sock bit causes the orchestrator container to mount the docker client, which allows the orchestrator container to spawn docker containers on the host (see this article for more information).
If you are running docker using docker-machine, as is now the standard, get the IP address of your Docker host by running docker-machine ls. If you are using boot2docker, then you can find your docker host's ip address by running the following command in your console: boot2docker ip
If it didn't come up, try running docker ps -a and docker logs tmpnb to help diagnose issues.
Alternatively, you can choose to setup a docker-compose.yml file:
httpproxy:
image: jupyter/configurable-http-proxy
environment:
CONFIGPROXY_AUTH_TOKEN: 716238957362948752139417234
container_name: tmpnb-proxy
net: "host"
command: --default-target http://127.0.0.1:9999
ports:
- 8000:8000
tmpnb_orchestrate:
image: jupyter/tmpnb
net: "host"
container_name: tmpnb_orchestrate
environment:
CONFIGPROXY_AUTH_TOKEN: 716238957362948752139417234
volumes:
- /var/run/docker.sock:/docker.sock
command: python orchestrate.py --command='jupyter notebook --no-browser --port {port} --ip=0.0.0.0 --NotebookApp.base_url={base_path} --NotebookApp.port_retries=0 --NotebookApp.token="" --NotebookApp.disable_check_xsrf=True'
Then, you can launch the container with docker-compose up, no building is required since they pull directly from images.
If you need to set the docker-version or other options, they can be passed to jupyter/tmpnb directly:
docker run --net=host -d -e CONFIGPROXY_AUTH_TOKEN=$TOKEN -v /var/run/docker.sock:/docker.sock jupyter/tmpnb python orchestrate.py --cull-timeout=60 --docker-version="1.13" --command="jupyter notebook --NotebookApp.base_url={base_path} --ip=0.0.0.0 --port {port}"
Note that if you do not pass a value to docker-version, tmpnb will automatically use the Docker API version provided by the server.
The tmpnb server has two APIs: a public one that receives HTTP requests under the / proxy route and an administrative one available only on the private, localhost interface by default. You can configure the interfaces (--ip, --admin_ip) and ports (--port, --admin_port) of both APIs using command line arguments.
If you decide to expose the admin API on a public interface, you can protect it by specifying a secret token in the environment variable ADMIN_AUTH_TOKEN when starting the tmpnb container. Thereafter, all requests made to the admin API must include it in an HTTP header like so:
Authorization: token <admin secret token here>
Likewise, if you only want to allow programmatic access to your tmpnb cluster by select clients, you can specify a separate secret token in the environment variable API_AUTH_TOKEN when starting the tmpnb container. All requests made to the public API must include it in an HTTP header in the same manner as depicted for the admin token above. Note that when this token is set, only the /api/* resources of the tmpnb server are available. All human-facing paths are disabled.
If you want to see the resources available in both the admin and user APIs, look at the handler paths registered in the orchestrate.py file. You should consider both APIs to be unstable.
tmpnb can run any Docker container provided by the --image option, so long as the --command option tells where the {base_path} and {port}. Those are literal strings, complete with curly braces that tmpnb will replace with an assigned base_path and port.
docker run --net=host -d -e CONFIGPROXY_AUTH_TOKEN=$TOKEN \
-v /var/run/docker.sock:/docker.sock \
jupyter/tmpnb python orchestrate.py --image='jupyter/demo' --command="jupyter notebook --NotebookApp.base_url={base_path} --ip=0.0.0.0 --port {port}"
Using jupyter/docker-stacks images
When using the latest jupyter/docker-stacks images with tmpnb, you can use the start-notebook.sh script or invoke the jupyter notebook command directly to run your notebook servers as user jovyan. Substitute your desired docker-stacks image name in the command below.
docker run -d \
--net=host \
-e CONFIGPROXY_AUTH_TOKEN=$TOKEN \
-v /var/run/docker.sock:/docker.sock \
jupyter/tmpnb \
python orchestrate.py --image='jupyter/minimal-notebook' \
--command='start-notebook.sh \
"--NotebookApp.base_url={base_path} \
--ip=0.0.0.0 \
--port={port} \
--NotebookApp.trust_xheaders=True"'Usage: orchestrate.py [OPTIONS]
Options:
--help show this help information
tornado/log.py options:
--log-file-max-size max size of log files before rollover
(default 100000000)
--log-file-num-backups number of log files to keep (default 10)
--log-file-prefix=PATH Path prefix for log files. Note that if you
are running multiple tornado processes,
log_file_prefix must be different for each
of them (e.g. include the port number)
--log-rotate-interval The interval value of timed rotating
(default 1)
--log-rotate-mode The mode of rotating files(time or size)
(default size)
--log-rotate-when specify the type of TimedRotatingFileHandler
interval other options:('S', 'M', 'H', 'D',
'W0'-'W6') (default midnight)
--log-to-stderr Send log output to stderr (colorized if
possible). By default use stderr if
--log_file_prefix is not set and no other
logging is configured.
--logging=debug|info|warning|error|none
Set the Python log level. If 'none', tornado
won't touch the logging configuration.
(default info)
orchestrate.py options:
--admin-ip ip for the admin server to listen on
[default: 127.0.0.1] (default 127.0.0.1)
--admin-port port for the admin server to listen on
(default 10000)
--allow-credentials Sets the Access-Control-Allow-Credentials
header.
--allow-headers Sets the Access-Control-Allow-Headers
header.
--allow-methods Sets the Access-Control-Allow-Methods
header.
--allow-origin Set the Access-Control-Allow-Origin header.
Use '*' to allow any origin to access.
--assert-hostname Verify hostname of Docker daemon. (default
False)
--command Command to run when booting the image. A
placeholder for {base_path} should be
provided. A placeholder for {port} and {ip}
can be provided. (default jupyter notebook
--no-browser --port {port} --ip=0.0.0.0
--NotebookApp.base_url={base_path}
--NotebookApp.port_retries=0)
--container-ip Host IP address for containers to bind to.
If host_network=True, the host IP address
for notebook servers to bind to. (default
127.0.0.1)
--container-port Within container port for notebook servers
to bind to. If host_network=True, the
starting port assigned to notebook servers
on the host. (default 8888)
--container-user User to run container command as
--cpu-quota Limit CPU quota, per container.
Units are CPU-µs per 100ms, so 1
CPU/container would be:
--cpu-quota=100000
--cpu-shares Limit CPU shares, per container
--cull-max Maximum age of a container (s), regardless
of activity. Default: 14400
(4 hours) A container that
has been running for this long will be
culled, even if it is not idle.
(default 14400)
--cull-period Interval (s) for culling idle containers.
(default 600)
--cull-timeout Timeout (s) for culling idle containers.
(default 3600)
--docker-version Version of the Docker API to use (default
auto)
--expose-headers Sets the Access-Control-Expose-Headers
header.
--extra-hosts Extra hosts for the containers, multiple
hosts can be specified by using a
comma-delimited string, specified in the
form hostname:IP (default [])
--host-directories Mount the specified directory as a data
volume, multiple directories can be
specified by using a comma-delimited string,
directory path must provided in full
(eg: /home/steve/data/:r), permissions
default to rw
--host-network Attaches the containers to the host
networking instead of the default docker
bridge. Affects the semantics of
container_port and container_ip. (default
False)
--image Docker container to spawn for new users.
Must be on the system already (default
jupyter/minimal-notebook)
--ip ip for the main server to listen on
[default: all interfaces]
--max-age Sets the Access-Control-Max-Age header.
--max-dock-workers Maximum number of docker workers (default 2)
--mem-limit Limit on Memory, per container (default
512m)
--pool-name Container name fragment used to identity
containers that belong to this instance.
--pool-size Capacity for containers on this system. Will
be prelaunched at startup. (default 10)
--port port for the main server to listen on
(default 9999)
--redirect-uri URI to redirect users to upon initial
notebook launch (default /tree)
--static-files Static files to extract from the initial
container launch
--user-length Length of the unique /user/:id path
generated per container (default 12)
WARNING The Makefile used in the commands below assume your
containers can be deleted. Please work on an isolated machine and read
the cleanup target in the Makefile prior to executing.
git clone https://github.com/jupyter/tmpnb.git
cd tmpnb
# Kick off the proxy and run the server.
# Runs on all interfaces on port 8000 by default.
# NOTE: stops and deletes all containers
make dev
If you are receiving 500 errors after changing the proxy port, make sure that you are using the correct internal port (proxy port+1 unless you specify it otherwise).