e2e-testing / starship configured to use more RAM than necessary

[0xpatrickdev]

What is the Problem Being Solved?

e2e-testing/config.yaml, modeled from agoric-sdk/multichain-testing/config.yaml, currently requires 10 GB of ram allocated in Docker to start.

This in turn:

• requires developers to override Docker RAM allocations, adding an extra setup step and friction point
• excludes developers working from lower-powered machines

The error usually manifests itself as pods in a Pending for a long time, and an "Insufficient memory" error in the kubectl describe pod agoriclocal-genesis-0 output.

However:

• it seems much less resources are used, so lowering allocations shouldn't affect the ability to run the service
• starship advertises much lower resource reqs in their docs
• agoric-sdk/multichain-testing/config.yaml is aimed to run performantly in CI, and is not necessarily suitable for local development

Description of the Design

config.yaml without, or with less, resources overrides is tested and checked in for dapp-orchestration-basics.

A minimal goal should be less than < 8GB, so we don't need to adjust Docker settings. It would best to go as low as we can, which might be cpu: "0.2" and memory: "400M" according the starship docs linked above.

CPU should also be considered in this exercise.

We might also consider checking in multiple starship configs, one could just run a single remote chain and relayer instead of two of each.

Test Plan

Manual testing is likely sufficient, but should be verified by multiple parties.

[0xpatrickdev]

The docs should also give guidance on minimum system reqs:

• RAM, Disk Storage. Currently it seems 10GB of RAM and 10-12GB of Disk space are needed

[dckc]

I see this in some hackathon project feedback:

Challenges Faced (and WAGMI!)

• Agoric's Multi Chain Testing Kit was difficult to run initially in our local as it required high end specifications but with the help of their team, we got a low specification version as well and were able to complete it.