Skip to content

Latest commit

 

History

History
457 lines (337 loc) · 16.8 KB

README.md

File metadata and controls

457 lines (337 loc) · 16.8 KB

OrderPortal

A portal for orders (requests, project proposals, etc) to a facility from its users.

The GitHub wiki contains some additional information.

Background

The OrderPortal system allows basically any type of form-based submission of information. It was designed to work for any type of service facility that handles discrete orders or project proposals, reports, etc.

The OrderPortal system was originally created to satisfy the needs of the National Genomics Infrastructure (NGI) Sweden, which is a service facility for DNA sequencing and genotyping on samples provided by external researchers. The facility processes the samples according to the order from the researcher.

The OrderPortal system is not hardcoded for any specific area such as DNA sequencing. On the contrary, considerable effort has gone into making sure the design is as general as possible. This means that installation of the system requires specific settings suite your particular needs.

Features

  • Allow users to register an account, which is approved by the facility staff.
  • Allow the user to specify an order according to a predefined form.
  • Allow the user to submit the order to the facility.
  • Let the facility staff keep track of review and agreements.
  • The facility staff can review, accept or decline an order.
  • Allow input from the user of required project data, such as sample sheets.
  • Allow attaching various documents to an order.
  • Display project status reports.
  • Allow keeping track of Key Performance Indicators (KPIs), facilitating resource usage reports for the facility.

Design outline

The system is a portal for orders (requests, project proposals, etc) to a single facility from its users. A user is a researcher external to the service facility, and may or may not be the Principal Investigator (PI) for one or more projects.

An order is created from a template which is called a form in this system. The facility administrators must set up and publish the order forms for a researcher to be able to create an order. Each form contains fields for data input by the researcher.

The OrderPortal system is designed for only one facility, displaying up to about 8 different order forms. There is no hard limit to the number of simultaneously published forms, but the current design does not work well with more than 8 forms. If an organization has different facilities requiring different sets of forms, then it is best to set up different instances of the OrderPortal system, with different back-end database instances.

A user account is defined within each OrderPortal instance. We decided against a design based on a single central user account database for all facilities. The email address of the user is used as the user account identifier.

The design of an order form is fairly general; there is nothing that is hardcoded for specific domains of science. The content of the order forms consists of a different fields for different types of input data. The order forms are defined by the facility administrators. An order form can be used only when it has been enabled. An outdated order form can be disabled; its orders will still be accessible.

The design of the system is kept as flexible and general as possible. Customisation of site logo and title is possible, and the information pages are under control of the facility administrators.

The order form

The order form fields are fully configurable by the facility administrators via the web interface. The field definitions are generic, and allow order forms to be designed for a wide variety of input data.

The order form allows hierarchical grouping of fields, with dynamic display according to rules. This allows for cases where a top-level selection of e.g. a specific technology determines which further input fields are required to be filled in.

When the order form for a facility is changed, previously submitted orders are not affected. It is not possible to change the form for previously created orders.

Simple info pages

There is a very simple information page subsystem. This is not by far a full-fledged wiki, so it can be used only for basic needs. All administrators in the system can edit these pages via the web interface. This feature can be disabled by modifying the settings.

Facility

The term facility is used for the organisation providing the service specified by the order form. One instance (database) of the system handles one facility. All entities in the database belong to one and only one facility.

There are three reasons for this design choice:

  1. Security between facilities. The existence and contents of a particular project in one facility must not be visible to the administrators or staff of another facility. This is a strict requirement for some facilities, and it is easier to implement if the databases for each facility is separate from one another.

  2. The styling of an order portal is much easier to implement if each facility has its own portal instance.

  3. The introduction, or elimination, of a facility in the overall organisation becomes much easier if every instance of the system is independent of the other.

One drawback with this design choice is that it complicates the communication between and linking of different but related projects in different facilities.

Users

A user is an account in the system. Almost all operation require that the user is logged in. The email address is the user account identifier.

There are three kinds of users:

  1. User: An external scientist, who uses the portal to place one or more orders, and to follow the progress of their own orders. The "customer" of the facility.

  2. Staff: Facility staff, who may view all orders, but not change anything.

  3. Admin: Facility administrators, a.k.a. project coordinators, who are allowed to view and edit all aspects of the system that can be modified via the web interface. This includes processing orders, modifying the order fields, and handling user accounts.

User accounts can be set as disabled, for example if the person leaves her position, or as a means of blocking invalid use. Deletion of a user account is not allowed, to allow full traceability of old orders. An account can always be enabled again.

An external scientist applies for a user account by providing the relevant information. Such an account is created with a status of "pending". The administrator reviews the pending user account and enables it if it appears legitimate. The user gets an email about the account having been enabled and with instructions on how to set the password for it.

Some configuration operations (settings variables) can only be done on the command line on the machine hosting the system.

Access privileges

The user can place orders as soon has she has logged in.

A user is allowed to specify which other users will be able to access to her orders. Access can also be granted to specific users for each individual order.

Order: form and fields

The administrator designs the forms and their set of fields which determine what the user must fill in for an order. The administrator can clone a form in order to make a new variant. Old forms can be disabled, and new forms enabled, as needed.

Adding a field in a new form requires deciding on the followin parameters:

  • Field identifier
  • Field data type
  • Is a field value required?
  • Field description
  • Value options, if relevant
  • Hierarchy and order, including conditional visibility
  • Visibility to the user; some fields may be visible only to the staff

When an order is created, its fields definitions are copied from the form. Thus, an order is always self-contained. Once an order has been created, its fields and selected options are effectively frozen, and remain fixed. Only the values of the fields may be changed, not the definition of them.

A field may be conditional, meaning that it is displayed only of some other field has been assigned a specific value. This is necessary for orders where the relevant fields depend on some high-level choice, such a type of technology to use for a project.

Order

An order contains a copy of all fields from its form. It belongs to a user. It is editable by the user until it has been submitted for approval by the facility staff.

An order may contain fields which require a value. An order lacking a required value can be saved, but it cannot be submitted. This allows the user to create and fill in orders only partially, and to return to the order at a later date to complete it.

An order can have one and only one status. The statuses available for an order is configurable. Here is a list of standard statuses:

State Semantics
PREPARATION Created, and possibly edited.
SUBMITTED Submitted by user.
REVIEW Under review by the facility.
ACCEPTED Checked and accepted by the facility.
REJECTED Rejected by facility.
PROCESSING The facility is working on the project.
ABORTED The project has been stopped.
CLOSED All work and steps for the order have been done.

The statuses and the allowed transitions between them are defined in a YAML configuration file. This allows a facility to define other statuses and transitions than the standard ones. It also allows new statuses and transitions to be added to an existing setup. Removing existing statuses may break the system, and should not be attempted; instead, the transitions should be modified to avoid the redundant status.

Interface

There are two main interfaces to the system, the web and the API. The web interface behaves differently depending on the type of the user account logged in.

User web interface

This is for ordinary users. It should provide the user with sufficient help and visual cues to allow filling in the form in a productive manner. Missing values and values outside of allowed ranges must be highlighted to help the user prepare a valid order.

Staff web interface

This is for normal facility staff usage. Display of orders according to status and other parameters.

Administrator web interface

For administrators only, this interface enables editing orders, users and fields. This includes the ability to move orders along into different statuses.

The Application Programming Interface (API)

The API allows other systems to interact with the order portal. It is based on RESTful principles using JSON and linked data to allow other systems to access and/or modify various data entities in the portal.

Attached files

Files such as agreements, specifications, images, etc, can be attached to an order or to a sample.

Links

Links to other web pages can be set for orders and samples, allowing users to navigate to other relevant information systems. This feature can be disabled by modifying the settings.

Log

Each change of an order is logged, and the information "who, when, what" is stored. The log trace is written by the system automatically and cannot be edited by any user.

Installation

The current version requires Python 3.6 or higher.

This instruction is based on the procedure used for the instances running on the SciLifeLab server. It will have to be adapted for your site.

The Linux account nginx is used to host the instance files. Change according to the policy at your site.

The name xyz is used below as a placeholder for the name of your instance.

Instructions for upgrading the OrderPortal software is given below under Updates.

Source code setup

Clone the GitHub repo. The commands below use the base GitHub repo; substitute by whichever fork you are using.

$ cd /var/www/apps
$ sudo mkdir xyz
$ sudo chown nginx.nginx xyz
$ cd xyz
$ sudo -u nginx git clone https://github.com/pekrau/OrderPortal.git

Create the site subdirectory for your instance using the site_template directory (since version 4.1):

$ cd OrderPortal
$ sudo -u nginx cp -r site_template site

Some files in your site directory most likely need to be modified. In particular, the YAML files may need to be modified.

Download and install the required third-party Python modules using the requirements.txt file as approprate for your environment.

$ sudo pip install -r requirements.txt

Settings file

Edit the site/settings.yaml file according to your setup. Some of the settings depend on actions described below.

$ cd /var/www/apps/xyz/OrderPortal/site
$ sudo -u nginx chmod go-r settings.yaml
$ sudo -u nginx emacs settings.yaml

See the comments in the settings file.

CouchDB setup

It is assumed that you already have a CouchDB instance running. Follow the standard CouchDB procedures.

  • Go to the CouchDB web interface.
  • Create the CouchDB user orderportal_xyz.
  • Log in as CouchDB admin. Set the password for the user orderportal_xyz:
    • Go to the database _users and open the document for the user orderportal_xyz.
    • Create a new field with the key "password", and set its value to the secret password. This password must also be edited into the settings file.
    • When you save the document, CouchDB will hash the password and remove the password field.
  • Create the database orderportal_xyz in CouchDB.
  • Click on "Security..." and ensure that only the proper user can access it:
    • In the Names field of Admins, add the user name like so: ["orderportal_xyz"] (a string in a list).
    • In the Names field of Members, add a dummy user name like so: ["dummy"] (a string in a list).
    • The Roles fields should not be changed.

Initialize the database in CouchDB using the CLI. This requires a valid settings file.

$ cd /var/www/apps/xyz/OrderPortal/orderportal
$ sudo -u nginx PYTHONPATH=/var/www/apps/xyz/OrderPortal python3 cli.py initialize

Create the first admin account in the database using the CLI:

$ sudo -u nginx PYTHONPATH=/var/www/apps/xyz/OrderPortal python3 cli.py create-admin

Logging

The settings file may define the file path of the log file (variable LOGGING_FILEPATH), if any. The log file must be located in a directory which the tornado server can write to. For example:

$ cd /var/log
$ sudo mkdir orderportal_xyz
$ sudo chown nginx.nginx orderportal_xyz

Tornado server

The tornado server should be executed as a system service. The mechanism for this depends on the operating system. For SELinux, a template systemd file is available at orderportal/site/orderportal_xyz.service. Copy, rename and edit it.

$ cd /etc/systemd/system
$ sudo cp /var/www/apps/ddd/OrderPortal/orderportal/site/orderportal_xyz.service orderportal_ddd.service
$ sudo emacs orderportal_ddd.service

HTTP nginx configuration

In our case, the tornado server is accessed through a reverse-proxy via nginx. The template nginx file is available at orderportal/site_template/orderportal_xyz.conf. Copy, rename and edit it. In particular, ensure that the URL and port is specified correctly.

$ cd /etc/nginx/conf.d
$ sudo cp /var/www/apps/xyz/OrderPortal/orderportal/site_template/orderportal_xyz.conf orderportal_xyzconf
$ sudo emacs orderportal_xyz.conf

Backup

Backup relies on running the CLI to dump all data in the CouchDB database to a tar file. Create a backup directory:

$ sudo mkdir /home/backup/backup_files/orderportal_xyz

Copy and rename the template bash backup script orderportal/site_template/dump_orderportal_xyz.bash. It will also need editing, since it uses a SciLifeLab-specific tape-backup setup:

$ cd /etc/scripts
$ sudo cp /var/www/apps/xyz/OrderPortal/orderportal/site_template/dump_orderportal_xyz.bash dump_orderportal_xyz.bash

Edit the crontab file:

$ export EDITOR=emacs
$ crontab -e

The line in the crontab file should look something like this:

45 22 * * * /etc/scripts/dump_orderportal_xyz.bash

Updates

To update the source code from the GitHub repo:

$ cd /var/www/apps/xyz/OrderPortal
$ sudo -u nginx git pull
$ sudo systemctl restart orderportal_xyz

Since version 3.6.19, the design documents are automatically updated when the tornado server is restarted.

Please note that if your installation uses a repo forked from the base, ensure that you have updated that repo first by making a pull request and executing it.

Set up git commit signing.