Skip to content

Parallel-exec hardening: residual functional test failures under EXEC3_PARALLEL=true unrelated to commitment-accumulator races #21108

@mh0lt

Description

@mh0lt

Symptom

Under EXEC3_PARALLEL=true (the #21017 matrix), several test groups still have functional failures after PR #21088 closed the commitment-accumulator race family. These are NOT race-detector hits — go test -race on the four parallel race-test groups reports 0 races post-#21088. They're pre-existing functional divergences between serial and parallel exec, e.g.:

  • TestEngineApiBAL* (block access list engine-API tests)
  • various TestEngineApi* built-block / fork-choice / reorg-recovery tests
  • some rpc/jsonrpc tests (e.g. TestGetBlockByTimestamp, TestGetBlockReceiptsByBlockHash, ...)
  • txnprovider/shutter tests

~20 unique test names across core-rpc, execution-other, etc. (exact list reproducible by running the parallel race-test matrix on main after #21088 lands).

What to do

This is the long tail of "make #21017 fully green in parallel mode".

Related: #21017, #21088

Metadata

Metadata

Assignees

Type

No type
No fields configured for issues without a type.

Projects

No projects

Relationships

None yet

Development

No branches or pull requests

Issue actions