Skip to content

Latest commit

 

History

History
431 lines (281 loc) · 22.2 KB

README.md

File metadata and controls

431 lines (281 loc) · 22.2 KB
Raw

Bridgehead

The Bridgehead is a secure, low-effort solution to connect your research institution to a federated research network. It bundles interoperable, open-source software components into a turnkey package for installation on one of your secure servers. The Bridgehead is pre-configured with sane defaults, centrally monitored and with an absolute minimum of "moving parts" on your side, making it an extremely low-maintenance gateway to data sharing.

This repository is the starting point for any information and tools you will need to deploy a Bridgehead. If you have questions, please contact us.

  1. Requirements
  2. Deployment
  3. Site-specific configuration
  4. Things you should know
  5. Troubleshooting
  6. License

Requirements

The data protection officer at your site will probably want to know exactly what our software does with patient data, and you may need to get their approval before you are allowed to install a Bridgehead. To help you with this, we have provided some data protection concepts:

Hardware

Hardware requirements strongly depend on the specific use-cases of your network as well as on the data it is going to serve. Most use-cases are well-served with the following configuration:

  • 4 CPU cores
  • 32 GB RAM
  • 160GB Hard Drive, SSD recommended

We recommend using a dedicated VM for the Bridgehead, with no other applications running on it. While the Bridgehead can, in principle, run on a shared VM, you might run into surprising problems such as resource conflicts (e.g., two apps using tcp port 443).

Software

You are strongly recommended to install the Bridgehead under a Linux operating system (but see the section Non-Linux OS). You will need root (administrator) priveleges on this machine in order to perform the deployment. We recommend the newest Ubuntu LTS server release.

Ensure the following software (or newer) is installed:

  • git >= 2.0
  • docker >= 20.10.1
  • docker-compose >= 2.xx (docker-compose and docker compose are both supported).
  • systemd
  • curl

We recommend to install Docker(-compose) from its official sources as described on the Docker website.

📝 Note for Ubuntu: Snap versions of Docker are not supported.

Network

A Bridgehead communicates to all central components via outgoing HTTPS connections.

Your site might require an outgoing proxy (i.e. HTTPS forward proxy) to connect to external servers; you should discuss this with your local systems administration. In that case, you will need to note down the URL of the proxy. If the proxy requires authentication, you will also need to make a note of its username and password. This information will be used later on during the installation process. TLS terminating proxies are also supported, see here. Apart from the Bridgehead itself, you may also need to configure the proxy server in git and docker.

The following URLs need to be accessible (prefix with https://):

  • To fetch code and configuration from git repositories
    • github.com
    • git.verbis.dkfz.de
  • To fetch docker images
    • docker.verbis.dkfz.de
    • Official Docker, Inc. URLs (subject to change, see official list)
      • hub.docker.com
      • registry-1.docker.io
      • production.cloudflare.docker.com
  • To report bridgeheads operational status
    • healthchecks.verbis.dkfz.de
  • only for DKTK/CCP
    • broker.ccp-it.dktk.dkfz.de
  • only for BBMRI-ERIC
    • broker.bbmri.samply.de
    • gitlab.bbmri-eric.eu
  • only for German Biobank Node
    • broker.bbmri.de

📝 This URL list is subject to change. Instead of the individual names, we highly recommend whitelisting wildcard domains: *.dkfz.de, github.com, *.docker.com, *.docker.io, *.samply.de, *.bbmri.de.

📝 Ubuntu's pre-installed uncomplicated firewall (ufw) is known to conflict with Docker, more info here.

Deployment

Site name

You will need to choose a short name for your site. This is not a URL, just a simple identifying string. For the examples below, we will use "your-site-name", but you should obviously choose something that is meaningful to you and which is unique.

Site names should adhere to the following conventions:

  • They should be lower-case.
  • They should generally be named after the city where your site is based, e.g. karlsruhe.
  • If you have a multi-part name, please use a hypen ("-") as separator, e.g. le-havre.
  • If your site is for testing purposes, rather than production, please append "-test", e.g. zaragoza-test.
  • If you are a developer and you are making changes to the Bridgehead, please use your name and prepend "dev-", e.g. dev-joe-doe.

GitLab repository

In order to be able to install, you will need to have your own repository in GitLab for your site's configuration settings. This allows automated updates of the Bridgehead software.

To request a new repository, please contact your research network administration or send an email to one of the project specific addresses:

Mention:

  • which project you belong to, i.e. "bbmri" or "ccp"
  • site name (According to conventions listed above)
  • operator name and email

We will set the repository up for you. We will then send you:

  • A Repository Short Name (RSN). Beware: this is distinct from your site name.
  • Repository URL containing the acces token eg. https://BH_Dummy:[email protected]/-bridgehead-configs/dummy.git

During the installation, your Bridgehead will download your site's configuration from GitLab and you can review the details provided to us by email.

Base Installation

First, download your site specific configuration repository:

sudo mkdir -p /etc/bridgehead/
sudo git clone <REPO_URL_FROM_EMAIL> /etc/bridgehead/

Review the site configuration:

sudo cat /etc/bridgehead/bbmri.conf

Pay special attention to:

  • SITE_NAME
  • SITE_ID
  • OPERATOR_FIRST_NAME
  • OPERATOR_LAST_NAME
  • OPERATOR_EMAIL

Clone the bridgehead repository:

sudo mkdir -p /srv/docker/
sudo git clone https://github.com/samply/bridgehead.git /srv/docker/bridgehead

Then, run the installation script:

cd /srv/docker/bridgehead
sudo ./bridgehead install <PROJECT>

Register with Samply.Beam

Many Bridgehead services rely on the secure, performant and flexible messaging middleware called Samply.Beam. You will need to register ("enroll") with Samply.Beam by creating a cryptographic key pair for your bridgehead:

cd /srv/docker/bridgehead
sudo ./bridgehead enroll <PROJECT>

... and follow the instructions on the screen. Please send your default Collection ID and the display name of your site together with the certificate request when you enroll. You should then be prompted to do the next step:

Starting and stopping your Bridgehead

If you followed the above steps, your Bridgehead should already be configured to autostart (via systemd). If you would like to start/stop manually:

To start, run

sudo systemctl start bridgehead@<PROJECT>.service

To stop, run

sudo systemctl stop bridgehead@<PROJECT>.service

To enable/disable autostart, run

sudo systemctl [enable|disable] bridgehead@<PROJECT>.service

Testing your new Bridgehead

After starting the Bridgehead, you can watch the initialization process with the following command:

/srv/docker/bridgehead/bridgehead logs <project> -f

if this exits with something similar to the following:

[email protected]: Main process exited, code=exited, status=1/FAILURE

Then you know that there was a problem with starting the Bridgehead. Scroll up the printout to find the cause of the error.

Once the Bridgehead is running, you can also view the individual Docker processes with:

docker ps

There should be 6 - 10 Docker proceses. If there are fewer, then you know that something has gone wrong. To see what is going on, run:

/srv/docker/bridgehead/bridgehead logs <Project> -f

This translates to a journalctl command so all the regular journalctl flags can be used.

Once the Bridgehead has passed these checks, take a look at the landing page:

https://localhost

You can either do this in a browser or with curl. If you visit the URL in the browser, you will neet to click through several warnings, because you will initially be using a self-signed certificate. With curl, you can bypass these checks:

curl -k https://localhost

Should the landing page not show anything, you can inspect the logs of the containers to determine what is going wrong. To do this you can use ./bridgehead docker-logs <Project> -f to follow the logs of the container. This transaltes to a docker compose logs command meaning all the ususal docker logs flags work.

If you have chosen to take part in our monitoring program (by setting the MONITOR_APIKEY variable in the configuration), you will be informed by email when problems are detected in your Bridgehead.

De-installing a Bridgehead

You may decide that you want to remove a Bridgehead installation from your machine, e.g. if you want to migrate it to a new location or if you want to start a fresh installation because the initial attempts did not work.

To do this, run:

sh bridgehead uninstall

Site-specific configuration

HTTPS Access

Even within your internal network, the Bridgehead enforces HTTPS for all services. During the installation, a self-signed, long-lived certificate was created for you. To increase security, you can simply replace the files under /etc/bridgehead/traefik-tls with ones from established certification authorities such as Let's Encrypt or DFN-AAI.

TLS terminating proxies

All of the Bridgehead's outgoing connections are secured by transport encryption (TLS) and a Bridgehead will refuse to connect if certificate verification fails. If your local forward proxy server performs TLS termination, please place its CA certificate in /etc/bridgehead/trusted-ca-certs as a .pem file, e.g. /etc/bridgehead/trusted-ca-certs/mylocalca.pem. Then, all Bridgehead components will pick up this certificate and trust it for outgoing connections.

To find the certificate file, first run the following:

curl -v https://broker.bbmri.samply.de/v1/health

In the output, look out for the line:

successfully set certificate verify locations:

Here a file will be mentioned, perhaps in the directory /etc/ssl/certs. The exact location will depend on your operating system. This is the file that you need to copy.

File structure

  • /srv/docker/bridgehead contains this git repository with the shell scripts and project-specific configuration. In here, all files are identical for all sites. You should not make any changes here.
  • /etc/bridgehead contains your site-specific configuration synchronized from your site-specific git repository as part of the base installation. To change anything here, please consult your git repository (find out its URL via git -C /etc/bridgehead remote -v).
    • /etc/bridgehead/<PROJECT>.conf is your main site-specific configuration, all bundled into one concise config file. Do not change it here but via the central git repository.
    • /etc/bridgehead/<PROJECT>.local.conf contains site-specific parameters to be known to your Bridgehead only, e.g. local access credentials. The file is ignored via git, and you may edit it here via a text editor.
    • /etc/bridgehead/traefik-tls contains your Bridgehead's reverse proxies TLS certificates for HTTPS access.
    • /etc/bridgehead/pki contains your Bridgehead's private key (e.g., but not limited to Samply.Beam), generated as part of the Samply.Beam enrollment.
    • /etc/bridgehead/trusted-ca-certs contains third-party certificates to be trusted by the Bridgehead. For example, you want to place the certificates of your TLS-terminating proxy here.

Your Bridgehead's actual data is not stored in the above directories, but in named docker volumes, see docker volume ls and docker volume inspect <volume_name>.

BBMRI-ERIC Directory entry needed

If you run a biobank, you should be listed together with your collections with in the Directory, a BBMRI-ERIC project that catalogs biobanks.

To do this, contact the BBMRI-ERIC national node for the country where your biobank is based, see the list of nodes.

Once you have added your biobank to the Directory you got persistent identifier (PID) for your biobank and unique identifiers (IDs) for your collections. The collection IDs are necessary for the biospecimens assigning to the collections and later in the data flows between BBMRI-ERIC tools. In case you cannot distribute all your biospecimens within collections via assigning the collection IDs, you should choose one of your sample collections as a default collection for your biobank. This collection will be automatically used to label any samples that have not been assigned a collection ID in your ETL process. Make a note of this default collection ID, you will need it later on in the installation process.

Directory sync tool

The Bridgehead's Directory Sync is an optional feature that keeps the Directory up to date with your local data, e.g. number of samples. Conversely, it also updates the local FHIR store with the latest contact details etc. from the Directory. You must explicitly set your country specific directory URL, username and password to enable this feature.

You should talk with your local data protection group regarding the information that is published by Directory sync.

Full details can be found in directory_sync_service.

To enable it, you will need to set these variables to the bbmri.conf file of your GitLab repository. Here is an example config:

DS_DIRECTORY_USER_NAME=your_directory_username
DS_DIRECTORY_USER_PASS=your_directory_password

Please contact your National Node to obtain this information.

Optionally, you may change when you want Directory sync to run by specifying a cron expression, e.g. DS_TIMER_CRON="0 22 * * *" for 10 pm every evening.

Once you edited the gitlab config, the bridgehead will autoupdate the config with the values and will sync the data.

There will be a delay before the effects of Directory sync become visible. First, you will need to wait until the time you have specified in TIMER_CRON. Second, the information will then be synchronized from your national node with the central European Directory. This can take up to 24 hours.

Loading data

The data accessed by the federated search is held in the Bridgehead in a FHIR store (we use Blaze).

You can load data into this store by using its FHIR API:

https://<Name of your server>/bbmri-localdatamanagement/fhir

The name of your server will generally be the full name of the VM that the Bridgehead runs on. You can alternatively supply an IP address.

The FHIR API uses basic auth. You can find the credentials in /etc/bridgehead/<project>.local.conf.

Note that if you don't have a DNS certificate for the Bridgehead, you will need to allow an insecure connection. E.g. with curl, use the -k flag.

The storage space on your hard drive will depend on the number of FHIR resources that you intend to generate. This will be the sum of the number of patients/subjects, the number of samples, the number of conditions/diseases and the number of observations. As a general rule of thumb, you can assume that each resource will consume about 2 kilobytes of disk space.

For more information on Blaze performance, please refer to import performance.

ETL for BBMRI and GBA

Normally, you will need to build your own ETL to feed the Bridgehead. However, there is one case where a short cut might be available:

  • If you are using CentraXX as a BIMS and you have a FHIR-Export License, then you can employ standard mapping scripts that access the CentraXX-internal data structures and map the data onto the BBMRI FHIR profile. It may be necessary to adjust a few parameters, but this is nonetheless significantly easier than writing your own ETL.

You can find the profiles for generating FHIR in Simplifier.

Things you should know

Auto-Updates

Your Bridgehead will automatically and regularly check for updates. Whenever something has been updates (e.g., one of the git repositories or one of the docker images), your Bridgehead is automatically restarted. This should happen automatically and does not need any configuration.

If you would like to understand what happens exactly and when, please check the systemd units deployed during the installation via systemctl cat bridgehead-update@<PROJECT>.service and systemctl cat bridgehead-update@<PROJECT>.timer.

Auto-Backups

Some of the components in the bridgehead will store persistent data. For those components, we integrated an automated backup solution in the bridgehead updates. It will automatically save the backup in multiple files

  1. Last-XX, were XX represents a weekday to allow re-import of at least one version of the database for each of the past seven days.
  2. Year-KW-XX, were XX represents the calendar week to allow re-import of at least one version per calendar week
  3. Year-Month, to allow re-import of at least one version per month

To enable the Auto-Backup feature, please set the Variable BACKUP_DIRECTORY in your sites configuration.

Development Installation

By using ./bridgehead dev-install <projectname> instead of install, you can install a developer bridgehead. The difference is, that you can provide an arbitrary configuration repository during the installation, meaning that it does not have to adhere to the usual naming scheme. This allows for better decoupling between development and production configurations.

Non-Linux OS

The installation procedures described above have only been tested under Linux.

Below are some suggestions for getting the installation to work on other operating systems. Note that we are not able to provide support for these routes!

We believe that it is likely that installation would also work with FreeBSD and MacOS.

Under Windows, you have 2 options:

  • Virtual machine
  • WSL

We have tested the installation procedure with an Ubuntu 22.04 guest system running on a VMware virtual machine. That worked flawlessly.

Installation under WSL ought to work, but we have not tested this.

Troubleshooting

Docker Daemon Proxy Configuration

Docker has a background daemon, responsible for downloading images and starting them. Sometimes, proxy configuration from your system won't carry over and it will fail to download images. In that case, you'll need to configure the proxy inside the system unit of docker by creating the file /etc/systemd/system/docker.service.d/proxy.conf with the following content:

[Service]
Environment="HTTP_PROXY=http://proxy.example.com:3128"
Environment="HTTPS_PROXY=https://proxy.example.com:3128"
Environment="NO_PROXY=localhost,127.0.0.1,some-local-docker-registry.example.com,.corp"

After saving the configuration file, you'll need to reload the system daemon for the changes to take effect:

sudo systemctl daemon-reload

and restart the docker daemon:

sudo systemctl restart docker

For more information, please consult the official documentation.

Monitoring

To keep all Bridgeheads up and working and detect any errors before a user does, a central monitoring

  • Your Bridgehead itself will report relevant system events, such as successful/failed updates, restarts, performance metrics or version numbers.
  • Your Bridgehead is also monitored from the outside by your network's central components. For example, the federated search will regularly perform a black-box test by sending an empty query to your Bridgehead and checking if the results make sense.

In all monitoring cases, obviously no sensitive information is transmitted, in particular not any patient-related data. Aggregated data, e.g. total amount of datasets, may be transmitted for diagnostic purposes.

License

Copyright 2019 - 2022 The Samply Community

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at

http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.