Skip to content

ARROW-1980: [Python] Fix race condition in write_to_dataset #1468

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
jcrist wants to merge 1 commit into from
Closed

ARROW-1980: [Python] Fix race condition in write_to_dataset #1468

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
17 changes: 10 additions & 7 deletions python/pyarrow/parquet.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -966,6 +966,14 @@ def write_table(table, where, row_group_size=None, version='1.0',
""".format(_parquet_writer_arg_docs)


def _mkdir_if_not_exists(fs, path):
if fs._isfilestore() and not fs.exists(path):
try:
fs.mkdir(path)
except OSError:
assert fs.exists(path)
Copy link
Contributor Author

@jcrist jcrist Jan 9, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It might be worth adding exist_ok to the filesystem interface. Python has this kwarg for os.makedirs (but not os.mkdir) to say not to error if the directory already exists (defaults to False).

Copy link
Member

@xhochy xhochy Jan 10, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Before adding more things to the interface, I would prefer that we continue to settle the discussion on a uniform file interface between dask, Arrow and others. Then we actually know to what we all want to converge to.

Copy link
Member

@wesm wesm Jan 10, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Maybe worth opening a JIRA about this in either case



def write_to_dataset(table, root_path, partition_cols=None,
filesystem=None, preserve_index=True, **kwargs):
"""
Expand Down Expand Up @@ -1012,11 +1020,7 @@ def write_to_dataset(table, root_path, partition_cols=None,
else:
fs = _ensure_filesystem(filesystem)

if fs._isfilestore() and not fs.exists(root_path):
try:
fs.mkdir(root_path)
except OSError:
assert fs.exists(root_path)
_mkdir_if_not_exists(fs, root_path)

if partition_cols is not None and len(partition_cols) > 0:
df = table.to_pandas()
Expand All @@ -1034,8 +1038,7 @@ def write_to_dataset(table, root_path, partition_cols=None,
subtable = Table.from_pandas(subgroup,
preserve_index=preserve_index)
prefix = "/".join([root_path, subdir])
if fs._isfilestore() and not fs.exists(prefix):
fs.mkdir(prefix)
_mkdir_if_not_exists(fs, prefix)
outfile = compat.guid() + ".parquet"
full_path = "/".join([prefix, outfile])
with fs.open(full_path, 'wb') as f:
Expand Down