Skip to content

Refactor MG C++ tests (handle initialization)#2439

Merged
rapids-bot[bot] merged 19 commits intorapidsai:branch-22.08from
seunghwak:enh_mg_test
Jul 27, 2022
Merged

Refactor MG C++ tests (handle initialization)#2439
rapids-bot[bot] merged 19 commits intorapidsai:branch-22.08from
seunghwak:enh_mg_test

Conversation

@seunghwak
Copy link
Contributor

@seunghwak seunghwak commented Jul 22, 2022

Update MG C++ tests to initialize a RAFT handle_t object in SetUpTestCase().

Previously, a handle_t object is initialized inside individual tests. If an exception is thrown inside a single test, a stack unwinding process starts, and this calls handle_t's destructor, and the destructor invokes ncclCommDestroy(). If only a subset of workers fail, some processes will still advance and can be blocked on a NCCL call, This bars ncclCommDestroy() from finishing, and the stack unwinding process will hang. The outcome is no output in the console and testers may have no clue that exceptions are thrown.

With this restructuring, exceptions in individual tests will not destroy a handle_t object, so testers will see the exception messages.

@seunghwak seunghwak requested review from a team as code owners July 22, 2022 23:33
@seunghwak seunghwak changed the title Update MG C++ tests Refactor MG C++ tests (handle initialization) Jul 22, 2022
@seunghwak seunghwak self-assigned this Jul 22, 2022
@seunghwak seunghwak added 2 - In Progress improvement Improvement / enhancement to an existing function non-breaking Non-breaking change labels Jul 22, 2022
@seunghwak seunghwak added this to the 22.08 milestone Jul 22, 2022
@codecov-commenter
Copy link

codecov-commenter commented Jul 26, 2022

Codecov Report

❗ No coverage uploaded for pull request base (branch-22.08@d6b4cc5). Click here to learn what that means.
The diff coverage is n/a.

@@               Coverage Diff               @@
##             branch-22.08    #2439   +/-   ##
===============================================
  Coverage                ?   61.30%           
===============================================
  Files                   ?      106           
  Lines                   ?     5398           
  Branches                ?        0           
===============================================
  Hits                    ?     3309           
  Misses                  ?     2089           
  Partials                ?        0           

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update d6b4cc5...367fd88. Read the comment docs.

@BradReesWork
Copy link
Member

@gpucibot merge

@rapids-bot rapids-bot bot merged commit 830d113 into rapidsai:branch-22.08 Jul 27, 2022
@seunghwak seunghwak deleted the enh_mg_test branch August 11, 2022 23:23
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

improvement Improvement / enhancement to an existing function non-breaking Non-breaking change

Projects

None yet

Development

Successfully merging this pull request may close these issues.

6 participants