# AiiDAlab Deployment Meeting 2024-02-05

###### tags: `Meeting`

:::info
- **Location:** https://psich.zoom.us/j/69116696576 *(Please request password on slack)*
- **Date:** xx xx, 2023 14h (CET)
- **Agenda**

- Current AiiDAlab deployments
- New organisation of the deployments
- Status of the computational resources
- Planned AiiDAlab deployments

- **Participants:**
Before the meeting, the host should check and update the list below.

- Jason [PSI] (jy)
- Sasha [Empa] (ay)
- Giovanni [PSI] (gp)
- Carlo [Empa] (cp)

- **Host:** jy

:::

[TOC]

## Current AiiDAlab deployments

Global:
- *Main* AiiDAlab server (**UP**): https://aiidalab.materialscloud.org. Mostly used by people from Empa and their collaborators. Open to all researchers affiliated with Materials Cloud partners.
- MarketPlace AiiDAlab server (**DOWN**): https://materials-marketplace.aiidalab.net. Open to all Materials Modeling MarketPlace users.
- NEP AiiDAlab server (**UP**): https://aiidalab-nep.materialscloud.org/. Open to all researchers affiliated with NFFA-Europe PILOT (NEP). mrcloud 2 orgnisation. Sasha will take care of making it alive.

Institutional:

- PSI's instance (**UP**): https://aiidalab.psi.ch/

## New organisation of the deployments

- *Global* instances are only for quick demos (quick meaning: start in 2-3 mins; OK to run for a few hours for longer QE demos). Main demo: QE app.
- We keep only one global instance demo.aiidalab.io
- Production instances are per-institute.

## The repository used for deployment

- Instructions on using Azure Terraform: https://github.com/aiidalab/aiidalab-on-azure
- Ansible role to deploy the server: https://github.com/aiidalab/ansible-role-aiidalab-server
- Kubernetes scripts:
- microk8s for deploy in a VM: https://github.com/aiidalab/aiidalab-microk8s-deploy
- Regular K8s instance: https://github.com/aiidalab/aiidalab-k8s

## Access policy

- Demo server (demo.aiidalab.io) only accessible for the GitHub users:
- Limit the number of pods to avoid resource overload (4-6 before production in March; increase to some number to decide after - 50? 100?)
- Monitoring needs to be enabled (both to see if it's up, but also of costs, to be monitored in the first few months)
- QE app installed by default, including all pseudos etc.
- First startup time needs to be almost 0
- Login page clearly explaining the conditions (no backups, auto-deletion of user's data)
- Two apps: demo app (it welcomes you and tells what to do) and the QE app.
- Pre-populated database with computed things (5-10 interesting materials?)
- QE app should inform that there are some examples available (talk to Edan about pop-up thing)

## Available computational platforms

- Castor machine: project is prolonged till September 2024, accessible till December 2024.
- Kubernetes@CSCS is rolling out, but not yet reliavble
- Microsoft Azure (demo will be deploye there)
- EGI (Sasha will contact Enol to see the status)

On-site resources:
- PSI (virtual machine on premises)
- Empa (still to be clarified).

## Plans for the AiiDAlab deployments

- Deploy Empa instance (https://aiidalab.empa.ch)
- Deploy Bristol instance (Daniel Hollas)
- Deploy AiiDAlab Demo instance:
- URL: https://demo.aiidalab.io
- Deployed on Azure
- Use GitHub auth (and maybe also GitLab?) - ensure no fake users mining cryptocurrency!

## Follow-ups

- What is the auto-culling policy on the demo server (when to kill the containers)
- How and when we delete the data?