Skip to content

Latest commit

 

History

History
73 lines (50 loc) · 3.56 KB

SERVICES_README.md

File metadata and controls

73 lines (50 loc) · 3.56 KB
Raw

Print & Probability Workbench Architecture

Workbench Services

The Print & Probability 'workbench' site is comprised of a number of services as seen in the diagram above. The site utilizes Django, a Python app that interfaces with an underlying PostgreSQL database that maintains the project's data tables as well as admin tables for the site. Gunicorn acts as the HTTP web server that communicates with Django and Nginx acts as a proxy server to the outside web. A Vue front end (shown below as 'Front-end Artifacts') communicates the interactions of the site's visitors to the Django app. The workbench also uses an IIIF server to serve the site images. The Django App, Gunicorn, and the PostgreSQL database run on Bridges virtual machine vm012.bridges2.psc.edu while the IIIF server runs on Bridges virtual machine vm013.bridges2.psc.edu. The IIIF server itself runs within a singularity instance iiif1. In order to log onto any of these virtual machines type ssh <your_bridges_username>@vm<vm_number>.bridges2.psc.edu.

Vue/Django/Gunicorn

Restarting

  1. Log into vm012.bridges2.psc.edu
  2. cd /usr/local/bin - This directory contains commands to start/stop the backend
  3. kill_gunicorn - Stops the gunicorn server that works with the django app
  4. start_printprob - Starts the django app and gunicorn

App location

  • /data/printprob-db via any Bridges login
  • NOTE: Any changes to the Django app or Vue frontend must be placed here.

PostgreSQL database

Preparation before use

  • Make a .pgpass file in your Bridges home directory containing:
    • localhost:5432:pp:app:CwyYmvkA897FUy3ANCRiXDkMJjrZAohcrpcUL

Access to make queries

  • psql -h localhost -p 5432 -U app -d pp

Restarting

  • sudo /usr/bin/systemctl restart postgresql-12

Backup

  • As of May 2024, Sriram has been backing up the database weekly in /ocean/projects/hum160002p/shared/db_backups, removing the previous week's backup when done. Below is information on that backup process.

Backup steps

  1. vacuumdb --dbname=pp --host=localhost --username=app
  2. reindexdb --dbname=pp --host=localhost --username=app --verbose
  3. bash /data/backup_pg.sh
  4. Delete the old backup file from the backups directory in /data

Weekly Vacuuming

# Vacuum database at midnight every Sunday
0 0 * * SUN vacuumdb --dbname=pp --host=localhost --username=app --quiet
0 0 * * SUN touch /data/last_pp_db_vacuumed

# Re-index 'pp' database weekly once every Sunday at 01:00 - after vacuum finishes
0 1 * * SUN reindexdb --dbname=pp --host=localhost --username=app --quiet
0 1 * * SUN touch /data/last_pp_db_reindexed

Nginx

Status

  • service nginx status

Restarting

  • service nginx restart
  • NOTE: Nginx needs to be restarted every 6 months to a year. server.crt and trustchain.crt are both changed by Bridges staff. Need to communicate with them to do so.

IIIF Server

Restarting

  1. Log into vm013.bridges2.psc.edu
  2. cd /ocean/projects/hum160002p/shared/iiif-server
  3. singularity instance stop iiif1 - Stop any currently-running instance
  4. ./startup.sh - Create the singularity instance
  5. nohup singularity run --env VIPS_DISC_THRESHOLD=250m instance://iiif1 > iiif.log & - Activate the IIIF server inside the contianer

NOTE: The new argument --env VIPS_DISC_THRESHOLD=250m is necessary for an image memory cap we were exceeding with the IIIF server. This sets the environment variable VIPS_DISC_THRESHOLD inside the singularity instance in which the IIIF server program is running.