Applications on Demand Service - information for providers

Main

EGI.eu operations services

Support

Documentation

Tools

Activities

Performance

Technology

Catch-all Services

Resource Allocation

Security

Applications on Demand Service menu:

Home •

Documentation for providers •

Documentation for developers •

Architecture

This page provides information about the 'EGI Access Platform', hereafter referred to as the Platform, for supporting the long tail. The long-tail of science refers to the individual researchers and small laboratories who - opposed to large, expensive collaborations - do not have access to computational resources and online services to manage and analyse large amount of data.

This Platform allows individual researchers and small research teams to perform compute and data-intensive simulations on large, distributed networks of computers in a user friendly way.

If you are interested in the activity that developed and now maintains the Platform, please jump to the Long-tail of science activity page.

Overview

• Science Gateways/Portals and resource providers must accept and follow the platform security policy: https://documents.egi.eu/public/ShowDocument?docid=2734
  
• Science Gateways/Portals providers must sign OLA with EGI.eu.
• Science Gateway/Portals providers must integrate with the User Registration Portal to enable the single sign-on capability for users.
• Science Gateway/Portals providers must integrate with the per-user subproxy solution to offer traceable user authentication towards the e-infrastructure VO.
• Science Gateways/Portals must implement user resource usage quota (to prohibit a user consuming all the resources from the platform).
• Resource providers must support the per-user subproxy solution and join the e-infrastructure VO.

The below subsections provide guidance to complete these steps.

How to join

How to connect a Science Gateway to the Platform

Connecting the Science Gateway with the User Registration Portal (URP)

Client service Registration

1. Open the GGUS ticket to operations that include return URIs

2. UNITY team send Client clientID and secretKey

Authorization procedure Unity with Client:

1] The Client sends a request to the OpenID Provider

address: https://unity.egi.eu/oauth2-as/oauth2-authz

parameters:

response_type:code
redirect_uri: [[Redirect url]]
client_id:unity-oauth-egrantstate: [[You should generate your own state eg. md5(uniqid(rand(), TRUE));]]

scope:profile openid 

example:

[https://unity.egi.eu/oauth2-as/oauth2-authz https://unity.egi.eu/oauth2-as/oauth2-authz]
response_type=code
&client_id=123123123
&redirect_uri=https%3A%2F%2Fclient.pl%2Fauth
&scope=openid%20profile



&state=a123a123a123

2] Authorization Server authenticates the End-User.

3] Authorization Server obtains End-User Consent/Authorization.

4] Authorization Server sends the End-User back to the redirect uri from the first request (Redirect url) with code.

example of the response

Location: [https://client.pl/auth https://client.pl/auth]




code=uniquecode123
&state=a123a123a123

5] Client sends the code to the Token Endpoint to receive an Access Token and ID Token in the response.

POST /token HTTP/1.1
Host: [http://client.pl/ client.pl]


Authorization: Basic czZCaGRSa3F0MzpnWDFmQmF0M2JW


Content-Type: application/x-www-form-urlencoded


grant_type=authorization_code&code=uniquecode123



&redirect_uri=https%3A%2F%2Fclient.pl%2Fauth

6] Client validates the tokens and retrieves the End-User's Subject Identifier.

example:

HTTP/1.1 200 OK
Content-Type: application/json


Cache-Control: no-store


Pragma: no-cache


{


"access_token":"accessToken123",


"token_type":"Bearer",


"expires_in":3600,


"refresh_token":"refreshToken123",


"id_token":"idToken123123"



}

You should decode id_token and make some validation (more information: http://openid.net/specs/openid-connect-basic-1_0.html)

7] Client Gets some information from userpoint endpoint (https://unity.egi.eu/oauth2/userinfo)

example

https://unity.egi.eu/oauth2/userinfo?schema=openid&access_token=accessToken123"

8] User gets information about user such as email or name in json format

important data:

unity.server.clientId=  [YOUR CLIENT ID]
unity.server.clientSecret= [YOUR SECRET KEY]

unity.server.authorize=[https://unity.egi.eu/oauth2-as/oauth2-authz https://unity.egi.eu/oauth2-as/oauth2-authz]

unity.server.token=[https://unity.egi.eu/oauth2/token https://unity.egi.eu/oauth2/token]

unity.server.base=[https://unity.egi.eu/ https://unity.egi.eu]

full configuration:

https://unity.egi.eu/oauth2/.well-known/openid-configuration

List of attributes released by Unity

for scope 'profile':

name (string)

email (string)

confirmedRegistration (true/false) - tells if user has confirmed affiliation request

hasActiveSla - (true/false) - tells if user has confirmed resource request

both confirmedRegistration and hasActiveSla have to be set to 'true' to accept user as full ltos user

for scope 'additional':

persistent (string) - persistent ID of user given by UNITY. Attribute should be used by SG providers as a source of unique ltos userID.

OpenID Connect for Liferay

OpenId Connect for Liferay is a very rough but effective implementation of the OpenId connect protocol for Liferay. Use this module to authenticate with any OpenId Connect provider.

OpenID Connect for Django

To configure Django project to support OpenID Authentication it is possible to use the django-oidc-auth module.

How to use

$ pip install django-oidc-auth

Then configure the module:

INSTALLED_APPS += ['oidc_auth']

AUTHENTICATION_BACKENDS = ('oidc_auth.auth.OpenIDConnectBackend',) + AUTHENTICATION_BACKENDS

OIDC_AUTH = {
    'DEFAULT_PROVIDER': {
        'issuer': 'https://unity.egi.eu/oauth2',
        'client_id': AUTH_OIDC_CLIENT_ID,
        'client_secret': AUTH_OIDC_SECRET,
    },
    'SCOPES': ['openid', 'profile'],
}

Finally, add this to your urls.py:

urlpatterns = patterns('your.views',
    # ...
    url(r'oidc/', include('oidc_auth.urls')),
)

Run

python manage.py migrate

Connecting the Science Gateways/Portals with Per-User Sub-Proxies

The Platform serving the long tail uses per-user sub-proxies (PSUPs) for user's authentication. Any connected Science Gateway must generate per-user sub proxies for their users and must use these for any interaction with VO resources on behalf of the users.

A Science Gateway can generate PSUPs in two ways:

1. From a robot certificate that is physically hosted on the gateway server itself. OR
2. From a remote robot certificate that is hosted in the e-Token Server by INFN Catania.

We recommend you to choose the first option. If there is an IGTF CA in your country which issues robot certificates then obtain a robot from this CA. If such robots are not available in your country or region, then EGI can issue a robot for you from the SEEGRID catch-all CA. The next subsections provide details information to complete these steps.

Generic requirements

The Per-User Sub-Proxy (PUSP) and End-Entity Certificate (EEC) must satisfy the following requirements:

1. The EEC is a valid robot certificate:
   • it either contains OID 1.2.840.113612.5.2.3.3.1, see https://www.eugridpma.org/objectid/?oid=1.2.840.113612.5.2.3.3.1
   • or its DN matches the regular expression "/CN=[rR]obot[^/[:alnum:]]" i.e. containing a CN field which starts with robot or Robot and is followed by a non-alphanumerical non-slash character. see https://www.eugridpma.org/guidelines/robot/ section 3.
2. The PUSP is RFC 3820 compliant, i.e. no legacy GT2 or GT3 proxies
3. The PUSP is the first proxy delegation
4. If the same user enters via the same portal, he must get the same PUSP DN
5. No two distinct identified users will have the same PUSP DN.

A robot EEC that generates PUSP credentials SHOULD NOT be used for any other purpose; for example, it should not be used to generate non-PUSP proxy credentials and should not be use for direct authenticating.

The machine/service that will take care of PUSP generation and management should respect the following rules:

1. Documented response procedures in case of incidents (that are periodically tested).
2. A listed/accredited CSIRT team.
3. Internal risk assessment and an actuarial team to calculate the effective risk

Using a robot certificate from your national IGTF CA

1. Obtain a robot certificate from your national IGTF Certification Authority following the instructions here.
2. Register the robot in the vo.access.egi.eu VO: https://perun.metacentrum.cz/cert/registrar/?vo=vo.access.egi.eu
3. Generate proxies from the robot using this script: https://ndpfsvn.nikhef.nl/viewvc/mwsec/trunk/lcmaps-plugins-robot/tools/

Obtaining a robot certificate from EGI catch-all CA

1. Contact long-tail-support@mailman.egi.eu and send a short description of your gateway service and the way it would be integrated with platform resources. The team will arrange a robot certificate for your gateway from the SEEGRID CA (which operates as a 'catch-all' CA in EGI) and will register this in the VO and in the e-Token Server in Italy.
2. Register the robot in the vo.access.egi.eu VO: https://perun.metacentrum.cz/cert/registrar/?vo=vo.access.egi.eu
3. Generate proxies from the robot using this script: https://ndpfsvn.nikhef.nl/viewvc/mwsec/trunk/lcmaps-plugins-robot/tools/

Instructions to use the e-Token Server

1. Contact long-tail-support@mailman.egi.eu and send a short justification why you would like to use the eToken server (instead of hosting the robot certificate locally). Describe your gateway service and the way it would be integrated with platform resources. The team will arrange a robot certificate for your gateway from the SEEGRID CA and will register it in vo.access.egi.eu.
2. Provide long-tail-support@mailman.egi.eu with the static IP address of your gateway server, so proxy requests can be authorized from this address on the e-Token Server.
3. Generate proxies from the e-Token server following this guideline:

There are two available e-Token Server instances for availability and reliability reasons:

• etokenserver.ct.infn.it
• etokenserver2.ct.infn.it

The following rest API is available to get a PUSP given a unique identifier:

https://[eToken Server instance]:8443/eTokenServer/eToken/[Robot Certificate ID]?voms=[VO]:/[VO]&proxy-renewal=[true|false]&disable-voms-proxy=[true|false]&rfc-proxy=[true|false]&cn-label=user:[user unique identifier]

• Robot cetificate ID: it is the ID of your robot certificate in the e-Token server. It will be generated after the setup of your robot into the e-Token Server.
• VO: the VO you want to use to perform any action on the EGI infrastructure. The robot certificate must be a member of this VO.
• proxy-renewal: this option is used to enable (true) or disable (false) the automatic registration of a long-term proxy into a MyProxy Server.
• disable-voms-proxy: this option is used to generate plain (true) or VOMS proxy certificate (false).
• rfc-proxy: this option is used to generate standard RFC proxies (true) or legacy proxies (false).
• cn-label: this option is used to generate a PUSP for the given unique identifier.

below an example:

https://[eToken Server instance]:8443/eTokenServer/eToken/27br90771bba31acb942efe4c8209e69?voms=training.egi.eu:/training.egi.eu&proxy-renewal=false&disable-voms-proxy=false&rfc-proxy=true&cn-label=user:test1

Connecting the Science Gateways/portals with the EGI monitoring system

To monitor a new Science Gateway (SG)/portal with the EGI monitoring system based on ARGO, the SG system administrator HAS TO:

• Register the Science Gateway/portal in the GOCDB under EGI.eu NGI
  • For that, please contact the EGI Operations Team (<operations@egi.eu>)

• Register the Science Gateway/portal in the GOCDB as service type: egi.Portal
  • For that, please contact the EGI Operations Team (<operations@egi.eu>)

• Change in the GOCDB the Certification Status of the SG in Certified
  • For that, please contact the EGI Operations Team (<operations@egi.eu>);

• • By default ARGO will automatically fetches the Science Gateway endpoint from the GOCDB and implements simple http checks using standard NAGIOS probe.

ARGO uses a bundle of connectors/sync components to monitor different service components, but if necessary new ones can be easily developed and added.

How to join as a grid/cloud resource provider

Any EGI resource provider can join the Platform to offer capacity for members of the long-tail of science. The site needs to run one of the supported grid or cloud middleware software, enable per-user sub-proxies (for user authentication and authorisation), and join the vo.access.egi.eu Virtual Organisation. Please email long-tail-support@egi.eu if you wish to join as a resource provider.

In order to provide authorization to the users of the LToS VO, a couple of DNs (Distinghished Names) are required to be configured on the services to be enabled. For instance, for the CREAM CE the usual grid-mapfile is the place where to add them, for OpenStack it's /etc/keystone/voms.json. You can find below the instructions for each service.

The following Robot Certificate DNs must be configured:

/DC=EU/DC=EGI/C=HU/O=Robots/O=MTA SZTAKI/CN=Robot:zfarkas@sztaki.hu
/C=IT/O=INFN/OU=Robot/L=Catania/CN=Robot: Catania Science Gateway  - Roberto Barbera

Instructions for OpenStack providers

Keystone-VOMS has support for per-user subproxies in the special branch called subproxy_support available in the github repository https://github.com/enolfc/keystone-voms (code is in progress of being integrated into the main branch of Keystone-VOMS). You can install the code from the repository following these instructions:

 git clone -b subproxy_support https://github.com/enolfc/keystone-voms.git
 cd keystone-voms
 pip install .

Configuration and deployment of the plugin does not change from the normal Keystone-VOMS plugin, follow the Keystone-VOMS documentation to deploy it.

There are new parameters to configure in your keystone config file, under the [voms] section:

• allow_subproxy, should be set to True for enabling PUSP support.
• subproxy_robots, should be set to * (recommended) or to a list of the DNs that are allowed to create PUSP in the system.
• subproxy_user_prefix, determines the expected prefix for the PUSP user specification. It is safe to leave it undefined so the default value (CN=eToken is used.

Instructions for OpenNebula providers

OpenNebula sites are not yet supported in the platform, there is some unofficial instructions (the development is not 100% complete, hence no instructions and partial support is available). Here some hints, please use at your own risks!

For the PUSP support on an up-to-date rOCCI-server, CESGA can enable partial PUSP support by adding

SetEnv ROCCI_SERVER_AUTHN_VOMS_ROBOT_SUBPROXY_IDENTITY_ENABLE yes

to Apache2 VHost configuration for rOCCI-server. This enables PUSP support in accounting but does NOT provide yet isolation between different PUSP users of the same robot.

Instructions for EMI 3/gLite providers

There is an EGI manual that shows how to set up a per-user sub-proxy to allow identification of the individual users under a common robot certificate. You can find the guide here: https://wiki.egi.eu/wiki/MAN12

How to join the user support team

If you wish to support platform users from your country, region or scientific disciplinary area, then please email long-tail-support@egi.eu. We can train you and then register you as a supporter in our team.

How to operate your service

• To operate your service, please refer to the official procedure PROC09 Resource Centre Registration and Certification

• To describe the Intervention in case of unscheduled failure of central operational tool, please refer to the documentation Tool Intervention Management

• To manage Service Intervention, please refer to the following documentation Service Intervention Management

Notifying users about downtimes

Scheduled downtime

• Scheduled interventions MUST be declared at least 24h in advance, specifying reason and duration.
• Existing scheduled interventions CAN be extended, provided that it is done 24h in advance.

1. For Grid/Cloud Resource Providers:

• Resource Provider (RP) has to register the scheduled downtime in the GOCDB by clicking on the 'Add a Downtime' link in the sidebar, specify the downtime information and the full list of impacted services.
  • In case there is more than one or select a site to select all the sites associated services.

2. For Science Gateway/portal providers:

• Science Gateway provider has to register the scheduled downtime in the GOCDB by clicking on the 'Add a Downtime' link in the sidebar.
• Display downtime information on the front-page of your Science Gateway.
• The Science Gateway picks up the downtime information from GOCDB via API, and sends out a broadcast email message to active users of the platform. (under development)
• The Science Gateway displays downtime information in a dedicated section of the front page (under development).

• GOCDB Programming Interface documentation can be found here

Unscheduled (unexpected) downtime

• Any intervention declared less than 24h in advance will be considered unscheduled.
• Both grid/cloud resource providers AND science gateway/portal providers MUST declare unscheduled interventions as soon as they are detected to inform the users. Register the scheduled downtime in the GOCDB by clicking on the 'Add a Downtime' link in the sidebar, specify the downtime information and the full list of impacted services in case there is more than one or select a site to select all the sites associated services.
• Unscheduled interventions CAN be declared up to 48h in the past (retroactive information to the user community).
• Unscheduled downtimes can be fetched from the GOCDB via API as well and notified to the end-users.
• GOCDB Programming Interface documentation can be found here

Display downtimes

• There is a JSR-286 portlet on github that picks up the downtime information from GOCDB via API, and sends out a broadcast email message to active users of the platform.
• The User Registration Portal (URP) reports, in a table-view, the list of downtimes affecting the services of the platform on the dashboard.