Skip to content

Conversation

@pierugo-dfinity
Copy link
Contributor

@pierugo-dfinity pierugo-dfinity commented Dec 3, 2025

NOTE: This branch currently has the changes introduced by pierugo/orchestrator/return-subnet-id. The actual changes of this PR will only contain the contents of rs/tests/consensus after the first is merged.

This PR introduces a nuance in how we corrupt a CUP: whether we corrupt it in a way that can still determine the subnet or not. If yes, then the recovery should be fairly identical to a normal one: we can use the same nodes and provision SSH keys: sr_app_recoverable_corrupt_cup_with_chain_keys_test. If not, then we can neither provision SSH keys nor use the same nodes: sr_app_failover_nodes_test.

@github-actions github-actions bot added the test label Dec 3, 2025
@pierugo-dfinity pierugo-dfinity changed the title test: distinguish tests for recoverable vs unrecoverable CUP corruptions test(recovery): distinguish tests for recoverable vs unrecoverable CUP corruptions Dec 3, 2025
@pierugo-dfinity pierugo-dfinity added the CI_ALL_BAZEL_TARGETS Runs all bazel targets and uploads them to S3 label Dec 3, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

CI_ALL_BAZEL_TARGETS Runs all bazel targets and uploads them to S3 test

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant