Notes from discussion about tasks for upcoming NDS5 workshop.
Prioritization of high-level tasks (will add tickets to JIRA with more detail, then provide estimates):
- Kubernetes cluster
- on Openstack (99%) (David ~4hrs) (multi node? need to confirm)
- Need to clean up existing tickets.
- New Task: Easy deployment of cluster on openstack. Allows anyone who wants to bring up a project to boot up a Kubernetes cluster on OpenStack (nebula only) (due 1/29)
- New Task: Testing (due 2/5)
- New Task: Update labs portal documentation (assigned to Mike, due 2/5)
- Development stack– NDS Dev (David + Mike ~?) (single node? need to confirm)
- NDS-82, 80%: Allows anyone to bring up Kubernetes
- Two parts:
- getting our devstack sorted out, how people will use ndsdev – 1/29
- Clowder under Kubernetes (95% done with the configuration) – will be done by 1/29. Outcome: kub config (yaml) that brings up an instance of vanilla Clowder that works end-to-end.
- New tasks: Testing (due 2/5- Craig), documentation (due 2/5 - Mike)
- on Openstack (99%) (David ~4hrs) (multi node? need to confirm)
- NDS Ops services (the thing that let's you deploy #3)
- NDS-92: Set of containers (logging, monitor, alert) = ELK? (David + Craig) - (1/29)
- New task: Testing, documentation
- Ability to scale it up and down (David)
- New task: Allows anyone (NDS Labs user) to scale up their cluster if they need to – adding more OpenStack resources. (Not for NDS5)
- NDS-92: Set of containers (logging, monitor, alert) = ELK? (David + Craig) - (1/29)
- Demo #1 (TERRA)
- Write additional pod specifications
- identify which ones
- Put those all together + glue
- More containerizing for Clowder
- Scale (elasticity, sharding mongo, sharding rabbit) see how it fits
- Write additional pod specifications
- Deployment GUI
- 1 day prototype/proof of concept (Mike)
- Actual GUI (est TBD)
- Do the TERRA demo validation (end-to-end walkthrough, Dev + OpenStack envs)
- Save it and make it a container, add it to the services (projects?)
- Demo #2 (MDF)
- MDF
- Workout with Globus
- Metasearch
- Flesh this out (Craig)
- Write the pod specifications
- Put those all together + glue
- MDF
- Workshop preparation (presentations)
There was a discussion of different services.
Question was asked whether the services list contains high-level component (e.g., clowder) or everything (e.g., clowder, rabbit, mongo, elastic). Answer: All.
Discussion of Kubernetes -- two steps: 1 deploy, 2 scale/replication,
Things to discuss later/
- Implementation that spans Nebula and Amazon? Once you get to #4 – all you need is the account. Idea: Terra is running on Nebula, migrates to Google Cloud?
- Is storage more important?