Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

dst: enhancements tracking issue #886

Open
2 of 7 tasks
asubiotto opened this issue May 30, 2024 · 0 comments
Open
2 of 7 tasks

dst: enhancements tracking issue #886

asubiotto opened this issue May 30, 2024 · 0 comments
Assignees
Labels
planned Planned work wont get closed by stalebot

Comments

@asubiotto
Copy link
Member

asubiotto commented May 30, 2024

When #883 merges, deterministic simulation testing will be part of our CI pipeline. Currently, deterministic simulation testing is very limited in that it only randomizes scheduling order of goroutines on local run queues. There is more work to do for more interesting failure simulation and infra work.

Simulation work

  • Simulate hard shutdowns: dst: add hardRestart command #895
  • Simulate filesystem. It feels like we should be able to do this at the wasip1 syscall layer. This might go hand-in-hand with exploring using the wazero runtime. vfs: add a VFS wrapper used in deterministic simulation tests #910
  • More extensive go runtime simulation
    • Is there a way to randomize scheduling even more?
    • Can we improve faketime? Currently time only advances when all goroutines are blocked, causing cases where sleeping goroutines can be starved of CPU. It would be interesting to spuriously wake them up by advancing time in non-conventional scenarios.

Infra work

  • Publish dst.wasm test builds from main. These can be run by anyone with any wasm runtime and GORANDSEED. cc @brancz
  • We currently use wasmtime for no good reason apart from it's the default runtime. Experiment with switching to wazero, which also seems to provide more knobs for runtime tuning (e.g. time and randomness). dst: run DST with wazero #896
  • Sometimes runs aren't exactly the same given the same random seed. We need to track this down.
@asubiotto asubiotto added the planned Planned work wont get closed by stalebot label May 30, 2024
@asubiotto asubiotto self-assigned this Jun 26, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
planned Planned work wont get closed by stalebot
Projects
None yet
Development

No branches or pull requests

1 participant