Skip to content

Three regression tests for real mainnet stalls were deleted, including #9,251,058 #379

Open

Description

@SDL-A1

From the external review of #374. Verified here.

Commit aa04535 removes three regression tests. main has five
test_reconcile_* tests in crates/sync/src/tests.rs; the PR head has two.

Removed:

  • test_reconcile_recovers_when_the_fork_has_no_canonical_pointer
  • test_reconcile_when_the_forks_parent_was_never_downloaded
  • test_reconcile_retreats_below_a_three_deep_fork

Kept: test_reconcile_leaves_a_genuine_tip_alone,
test_reconcile_rolls_back_orphaned_exec_head.

9_251_058 no longer appears anywhere in the PR's sources. That height is the
mainnet stall that wedged this node for three days — the PR's own source
comment records it as the reason the orphan-adoption path existed.

The specific worry

These guarded real, observed production failures. The PR replaces the mechanism
they covered, which is a legitimate reason for them to change — but the
replacements (beside-execution tests) are synthetic, and the mainnet chain
replay feeds blocks in canonical order, so it never exercises a live fork.

So the thing that broke mainnet twice has no regression test keyed to it any more.

What would settle it

Port the three removed cases to the new mechanism, or demonstrate that a
beside-execution test covers each one's scenario. If a case is genuinely
unreachable under the new design, say so in a comment where the test used to be,
with the reasoning.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions