Training Infrastructure as a Service (TIaaS)

Overview

Questions:
  • How to deploy EU’s TIaaS

Objectives:
  • Setup TIaaS

  • Request and manage trainings

  • Join a training

Requirements:
Time estimation: 30 minutes
Supporting Materials:
Last modification: Aug 6, 2021
License: Tutorial Content is licensed under Creative Commons Attribution 4.0 International License The GTN Framework is licensed under MIT

Overview

Agenda

  1. Introduction
  2. Setting up TIaaS
  3. Job Configuration

Introduction

Galaxy is widely used for teaching. In order to facilitate instructors, Galaxy Europe has developed Training Infrastructure as a Service (TIaaS). Workshop instructors can apply for TIaaS, and on the day of their workshop, their participants will be placed in a special group and use dedicated resources, thus reducing queue times on the day of the training.

TIaaS concept.
Figure 1: With TIaaS, all of your users visit the same server. In the background, the scheduler recognises which users are training users, and directs their jobs to special resources. In the EU deployment of TIaaS jobs preferentially use private resources, but can spill over to the main queue if there is not enough space available.

This tutorial will go cover how to set up such a service on your own Galaxy server.

Setting up TIaaS

hands_on Hands-on: Setup TIaaS

  1. In your requirements.yml add the TIaaS ansible role:

    --- a/requirements.yml
    +++ b/requirements.yml
    @@ -34,3 +34,5 @@
       version: 0.14.2
     - src: dj-wasabi.telegraf
       version: 0.12.0
    +- src: usegalaxy_eu.tiaas2
    +  version: 0.0.8
       
    

    And run the install step:

    code-in Input: Bash

    ansible-galaxy install -p roles -r requirements.yml
    
  2. In your galaxyservers group variables file, add the following:

    --- a/group_vars/galaxyservers.yml
    +++ b/group_vars/galaxyservers.yml
    @@ -214,3 +214,12 @@ telegraf_plugins_extra:
           - timeout = "10s"
           - data_format = "influx"
           - interval = "15s"
    +
    +# TIaaS setup
    +tiaas_dir: /opt/tiaas
    +tiaas_user: tiaas
    +tiaas_group: tiaas
    +tiaas_version: master
    +tiaas_admin_user: admin
    +tiaas_admin_pass: changeme
    +tiaas_listen_url: "127.0.0.1:6000"
       
    
  3. In the galaxyservers group variables file, we also need to set the database permissions correctly for TIaaS. It needs to be able to access some Galaxy tables, and we will carefully define only the ones we really need:

    --- a/group_vars/galaxyservers.yml
    +++ b/group_vars/galaxyservers.yml
    @@ -8,6 +8,7 @@ pip_package: python3-pip                               # geerlingguy.pip
     postgresql_objects_users:
       - name: galaxy
       - name: telegraf
    +  - name: tiaas
     postgresql_objects_databases:
       - name: galaxy
         owner: galaxy
    @@ -16,6 +17,27 @@ postgresql_objects_privileges:
         roles: telegraf
         privs: SELECT
         objs: ALL_IN_SCHEMA
    +  - database: galaxy
    +    roles: tiaas
    +    objs: galaxy_user,galaxy_session,job,history,workflow,workflow_invocation
    +    type: table
    +    privs: SELECT
    +  - database: galaxy
    +    roles: tiaas
    +    objs: user_group_association,galaxy_group,role,group_role_association
    +    type: table
    +    privs: SELECT,INSERT
    +  - database: galaxy
    +    roles: tiaas
    +    objs: group_role_association
    +    type: table
    +    privs: DELETE
    +  - database: galaxy
    +    roles: tiaas
    +    objs: role_id_seq,galaxy_group_id_seq,group_role_association_id_seq,user_group_association_id_seq
    +    type: sequence
    +    privs: USAGE,SELECT
    +
     # PostgreSQL Backups
     postgresql_backup_dir: /data/backups
     postgresql_backup_local_dir: "{{ '~postgres' | expanduser }}/backups"
       
    

    tip Why does TIaaS get DELETE privileges on Galaxy’s Database?

    The DELETE privilege is limited in scope to one table: group_role_association. This allows TIaaS to disassociate training groups from roles in the Galaxy database after the training event date has passed, so that users who participated in a training return to using normal (non-training) resources after the training ends.

    The usegalaxy_eu.tiaas2 role will create a cron job to perform this process every night at midnight. You can control when this runs (or disable it) using the tiaas_disassociate_training_roles variable.

  4. We need to add the usegalaxy_eu.tiaas2 role to the end of the playbook (galaxy.yml)

    --- a/galaxy.yml
    +++ b/galaxy.yml
    @@ -34,3 +34,4 @@
         - galaxyproject.cvmfs
         - galaxyproject.gxadmin
         - dj-wasabi.telegraf
    +    - usegalaxy_eu.tiaas2
       
    
  5. Lastly we should add the routes for TIaaS to the NGINX template for Galaxy:

    --- a/templates/nginx/galaxy.j2
    +++ b/templates/nginx/galaxy.j2
    @@ -61,4 +61,19 @@ server {
             proxy_pass http://127.0.0.1:3000/;
         }
        
    +    location /tiaas {
    +        uwsgi_pass {{ tiaas_listen_url }};
    +        uwsgi_param UWSGI_SCHEME $scheme;
    +        include uwsgi_params;
    +    }
    +
    +    location /tiaas/static {
    +        alias /opt/tiaas/static;
    +    }
    +
    +    location /join-training {
    +        uwsgi_pass {{ tiaas_listen_url }};
    +        uwsgi_param UWSGI_SCHEME $scheme;
    +        include uwsgi_params;
    +    }
     }
       
    
  6. Run the playbook

    code-in Input: Bash

    ansible-playbook galaxy.yml
    

TIaaS should be available now! The following routes on your server are now configured (we will run through these in the next section)

URL Use Who
https://<server>/tiaas/new/ Request a new TIaaS training Instructors
https://<server>/tiaas/admin/ Approve and Manage requests Admin
https://<server>/tiaas/stats/ Overall TIaaS statistics (EU Stats) Admins, Funding Agencies
https://<server>/tiaas/calendar/ Calendar of trainings (EU Calendar) Admins, Funding Agencies
https://<server>/join-training/<training-id> Join an TIaaS training Participants
https://<server>/join-training/<training-id>/status Dashboard with job states of trainees. Instructors

Let’s see it in action!

hands_on Hands-on: Using TIaaS

  1. Create a new TIaaS request
    • Go to https://<server>/tiaas/new/
    • Here you will find the request form users will fill in to request TIaaS: TIaaS request form.
    • For “Training Identifier”, fill in gryffindor (or remember this value if you enter something different)
      • This is the <training-id> used in the URLs listed above used for:
        1. Workshop participants to join the tiaas group
        2. Workshop instructors to monitor the progress of their participants.
    • Fill in the rest of the form as you like
    • Submit the form and you should see a confirmation dialog: TIaaS requested successfully.
  2. Approve TIaaS request
    • Next, the request will have to be approved by an admin
    • Go to https://<server>/tiaas/admin
    • Log in using the values you configured tiaas_admin_user and tiaas_admin_pass in your group variables file
      • Default values were admin:changeme
    • You should now see the admin panel: TIaaS admin console.
    • Click on Trainings, you should see the TIaaS request listed here: TIaaS request list.
    • Approve the request
      • Click on the training
      • Scroll down to the bottom
      • Change “Processed” to Approved and Save Approve TIaaS.
    • At this point, you would likely email the person who made the request to inform them of approval
  3. Join TIaaS Training
    • Make sure you are logged in to Galaxy
    • On the day of the workshop, participants will visit a following URL to join the TIaaS group
      • https://<server>/join-training/gryffindor
      • A confirmation dialog should appear if all went well: Join TIaaS.
  4. Monitor TIaaS status
    • This is very useful for instructors to monitor the job state of their participants
    • Go to https://<server>/join-training/gryffindor/status
    • In the Dasboard you should see that one user (you) has joined the training \
    • Run some jobs to see the dashboard in action TIaaS dashboard.
    • Scroll down to get some more information on a per-user level (anonymized)
      • Every user designated by their own identifier and colour, but no personal information TIaaS dashboard.

comment Note: GDPR assistance

Since this setup tracks additional personal information (submitter name & email, users in the queue view), TIaaS includes some always-on features to assist with your GDPR compliance.

  • Users in public status dashboard are only visible by an anonymized identifier and colour
  • Email addressses in the TIaaS admin panel will be automatically expunged 60 days after a training event

Of course you need to review any GDPR compliance concerns with your group’s legal representative(s), this only attempts to ensure some protections exist for the users of the system.

Job Configuration

While observability for teachers or trainers is already a huge benefit, one of the primary benefits of TIaaS from UseGalaxy.eu is that your jobs get sent to dedicated compute resources, which won’t be used by anyone else, during the period of the training. We will send all of the training jobs to pulsar if you have completed that tutorial, or one of the slurm destinations from the job configuration training.

In order to achieve this, we first need some way to sort the jobs of the training users into these private queues, while letting the other jobs continue on. So let’s create a sorting hat to figure out where jobs belong.

hands_on Hands-on: Writing a dynamic job destination

  1. Create and open templates/galaxy/dynamic_job_rules/hogwarts.py

    --- /dev/null
    +++ b/templates/galaxy/dynamic_job_rules/hogwarts.py
    @@ -0,0 +1,19 @@
    +from galaxy.jobs import JobDestination
    +from galaxy.jobs.mapper import JobMappingException
    +import os
    +
    +def sorting_hat(app, user):
    +    # Check that the user is not anonymous
    +    if not user:
    +        return app.job_config.get_destination('slurm')
    +
    +    # Collect the user's roles
    +    user_roles = [role.name for role in user.all_roles() if not role.deleted]
    +
    +    # If any of these are prefixed with 'training-'
    +    if any([role.startswith('training-') for role in user_roles]):
    +        # Then they are a training user, we will send their jobs to pulsar,
    +        # Or give them extra resources
    +        return app.job_config.get_destination('slurm-2c') # or pulsar, if available
    +
    +    return app.job_config.get_destination('slurm')
       
    

    This destination will check that the user_email is in a training group (role starting with training-).

  2. As usual, we need to instruct Galaxy of where to find this file. Edit your group variables file and add the following:

    --- a/group_vars/galaxyservers.yml
    +++ b/group_vars/galaxyservers.yml
    @@ -140,6 +140,7 @@ galaxy_local_tools:
     galaxy_dynamic_job_rules:
     - my_rules.py
     - map_resources.py
    +- hogwarts.py
        
     # systemd
     galaxy_manage_systemd: yes
       
    
  3. We next need to configure this plugin in our job configuration (templates/galaxy/config/job_conf.xml.j2):

    --- a/templates/galaxy/config/job_conf.xml.j2
    +++ b/templates/galaxy/config/job_conf.xml.j2
    @@ -13,7 +13,7 @@
                 <param id="manager">_default_</param>
             </plugin>
         </plugins>
    -    <destinations default="slurm">
    +    <destinations default="sorting_hat">
             <destination id="local_destination" runner="local_plugin"/>
             <destination id="pulsar" runner="pulsar_runner" >
                 <param id="default_file_action">remote_transfer</param>
    @@ -25,6 +25,10 @@
                 <param id="transport">curl</param>
                 <param id="outputs_to_working_directory">False</param>
             </destination>
    +        <destination id="sorting_hat" runner="dynamic">
    +            <param id="type">python</param>
    +            <param id="function">sorting_hat</param>
    +        </destination>
             <destination id="slurm" runner="slurm">
                 <param id="singularity_enabled">true</param>
                 <env id="LC_ALL">C</env>
    @@ -63,6 +67,7 @@
             <group id="testing">cores,time</group>
         </resources>
         <tools>
    +        <tool id="upload1" destination="slurm"/>
             <tool id="testing" destination="dynamic_cores_time" resources="testing" />
             <tool id="bwa" destination="pulsar"/>
             <tool id="bwa_mem" destination="pulsar"/>
       
    

    This is a Python function dynamic destination. Galaxy will load all python files in the {{ galaxy_dynamic_rule_dir }}, and all functions defined in those will be available to be used in the job_conf.xml.j2. Additionally it will send all jobs through the sorting hat, but we want upload jobs to stay local. They should always run locally.

  4. Run the playbook

    code-in Input: Bash

    ansible-playbook galaxy.yml
    
  5. Ensure your user is joined to a training

  6. Run a job and observe the logs to see where it goes (journalctl -u galaxy -f)

Congratulations! you have now set up TIaaS on your Galaxy server.

Key points

  • TIaaS is an additional service you can deploy which can help you provide a better service to your users

Frequently Asked Questions

Have questions about this tutorial? Check out the FAQ page for the Galaxy Server administration topic to see if your question is listed there. If not, please ask your question on the GTN Gitter Channel or the Galaxy Help Forum

Feedback

Did you use this material as an instructor? Feel free to give us feedback on how it went.
Did you use this material as a learner or student? Click the form below to leave feedback.

Click here to load Google feedback frame

Citing this Tutorial

  1. Helena Rasche, Saskia Hiltemann, 2021 Training Infrastructure as a Service (TIaaS) (Galaxy Training Materials). https://training.galaxyproject.org/training-material/topics/admin/tutorials/tiaas/tutorial.html Online; accessed TODAY
  2. Batut et al., 2018 Community-Driven Data Analysis Training for Biology Cell Systems 10.1016/j.cels.2018.05.012

details BibTeX

@misc{admin-tiaas,
author = "Helena Rasche and Saskia Hiltemann",
title = "Training Infrastructure as a Service (TIaaS) (Galaxy Training Materials)",
year = "2021",
month = "08",
day = "06"
url = "\url{https://training.galaxyproject.org/training-material/topics/admin/tutorials/tiaas/tutorial.html}",
note = "[Online; accessed TODAY]"
}
@article{Batut_2018,
    doi = {10.1016/j.cels.2018.05.012},
    url = {https://doi.org/10.1016%2Fj.cels.2018.05.012},
    year = 2018,
    month = {jun},
    publisher = {Elsevier {BV}},
    volume = {6},
    number = {6},
    pages = {752--758.e1},
    author = {B{\'{e}}r{\'{e}}nice Batut and Saskia Hiltemann and Andrea Bagnacani and Dannon Baker and Vivek Bhardwaj and Clemens Blank and Anthony Bretaudeau and Loraine Brillet-Gu{\'{e}}guen and Martin {\v{C}}ech and John Chilton and Dave Clements and Olivia Doppelt-Azeroual and Anika Erxleben and Mallory Ann Freeberg and Simon Gladman and Youri Hoogstrate and Hans-Rudolf Hotz and Torsten Houwaart and Pratik Jagtap and Delphine Larivi{\`{e}}re and Gildas Le Corguill{\'{e}} and Thomas Manke and Fabien Mareuil and Fidel Ram{\'{\i}}rez and Devon Ryan and Florian Christoph Sigloch and Nicola Soranzo and Joachim Wolff and Pavankumar Videm and Markus Wolfien and Aisanjiang Wubuli and Dilmurat Yusuf and James Taylor and Rolf Backofen and Anton Nekrutenko and Björn Grüning},
    title = {Community-Driven Data Analysis Training for Biology},
    journal = {Cell Systems}
}
                

Congratulations on successfully completing this tutorial!