Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Manage write access to HDF5 files using POSIX lockf file locking (equivalent to fcntl on Linux), which works over GPFS. I hope this will let us write incrementally to HDF5 files as variables are computed.
The writer thread waits to be sent some data, then acquires the lock and opens the HDF5 file. While it has the file open, it tries to do multiple writes if possible, but if nothing comes within 0.2 seconds (arbitrary choice), it closes the file and releases the lock, so another writer can take a turn.
I believe we need to reopen the file whenever another process may have modified it, because HDF5 can cache some data from open files, and that cache may become invalid if we keep the file open.
I made a demo of processes on two different hosts writing to the same file with the
WriterThread
class here, slowed down withsleep()
calls to illustrate what's going on.