teuthology: Add support for seek and sync in write_file #2010
Conversation
Signed-off-by: Christopher Hoffman <[email protected]>
Signed-off-by: Christopher Hoffman <[email protected]>
chrisphoffman
force-pushed
the
wip-dd-syncseek
branch
from
8b5e772 to
824260d
Compare
Thanks for the review. Comments addressed and tests added. PTAL |
|
|
||
| @patch("teuthology.orchestra.remote.RemoteShell.write_file") | ||
| def test_write_file_offset(self, m_write_file): | ||
| self.c.write_file("filename", "content", bs=1, offset=1024) | ||
| m_write_file.assert_called_with("filename", "content", bs=1, offset=1024) | ||
|
|
||
| @patch("teuthology.orchestra.remote.RemoteShell.write_file") | ||
| def test_write_file_sync(self, m_write_file): | ||
| self.c.write_file("filename", "content", sync=True) | ||
| m_write_file.assert_called_with("filename", "content", sync=True) |
There was a problem hiding this comment.
Those are test not for the remote.write_file but for cluster.remote_write, I guess there should be new tests added to test/test_remote.py module instead.
| def write_file(self, file_name, content, sudo=False, perms=None, owner=None, | ||
| **kwargs): |
There was a problem hiding this comment.
This change is unrelated directly to write_file. To keep PR simple if you really need seek and sync in cluster wide call, maybe create another PR.
| @@ -276,11 +277,20 @@ def write_file(self, path, data, sudo=False, mode=None, owner=None, | |||
| :param owner: set file owner if provided | |||
| :param mkdir: preliminary create the file directory, defaults False | |||
| :param append: append data to the file, defaults False | |||
| :param bs: write up to N bytes at a time, defaults 512 | |||
There was a problem hiding this comment.
I guess default is None, maybe just give a reference to dd, just add another line right after Write data to remote file above, like:
The data written in 512-byte blocks, provide `bs` to use bigger blocks.
Add support for seek and sync in write_file.
This patch will allow for partial writes and sync on completion of write.
This is to support strided writes for cephfs on a multi-mount system.