sync-barrier-sync in MPI I/O tests
Nobody has claimed this yet.
- Dominant language
- C
- Stars
- 122
- Forks
- 34
- PR merge metrics
- No merged PRs in 30d
Description
We should review our own use of MPI I/O to be sure we have the necessary sync-barrier-sync for any write/read test where a process may be trying to read data written by another process, e.g., we have a single sync here.
https://github.com/LLNL/UnifyFS/blob/fed21e57b262132939413afbd96430a9c4b157db/examples/src/testutil_rdwr.h#L249
This includes any test that queries for the file size.
Currently, I think a single ``MPI_File_sync`` call will be sufficient for UnifyFS, since that will call ``fsync()`` to flush any writes to the server. However, for proper MPI I/O usage, we should consider using the sync-barrier-sync type of construct that the MPI standard says is required.
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start with the single MPI_File_sync call in examples/src/testutil_rdwr.h around line 249, then search the MPI I/O tests for write/read and file-size queries. Compare each case with the required synchronization pattern and verify that all tests involving data written by another process use the appropriate sync-barrier-sync sequence.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- c
- Domain
- distributed-systems, testing-qa
- Issue type
- Bug
- Difficulty
- 3/5
- Estimated time
- 1-2 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 45/100