[goma] Enable auto-scaling.

- Migrate resource limits/requests to more reasonable numbers for the
individual pods.
- Specify `--memory-margin=` on the frontend deployments to prevent the
unnecessary /healthz 503'ing.
- Create HorizontalPodAutoscaler (HPA) configs for the deployments.
- Drop the `exec-server` readiness delay to 10 seconds (after toolchain
batch loading).
- Raise all other deployment readiness delays from 3 to 5 seconds.
- Use a rolling update strategy in all deployments.

Change-Id: I20f67ae2314897547acb9bc956578e695ee00b85
Reviewed-on: https://fuchsia-review.googlesource.com/c/infra/config/+/377034
Reviewed-by: Haowei Wu <haowei@google.com>
10 files changed
tree: eebaa5bca493a10cb493732a5f1b2209ca125ca1
  1. goma/
  2. manifest/
  3. repositories/
  4. .gitignore
  5. .style.yapf
  6. AUTHORS
  7. LICENSE
  8. OWNERS
  9. PATENTS
  10. README.md
README.md

Config

Currently the only configs here are for Tricium. All other configs live in the integration repo.