Blobfuse2 Limitations
Vikas Bhansali edited this page Oct 14, 2022
·
4 revisions
- mknod
- link (hard link api)
- setxattr
- getxattr
- listxattr
- removexattr
- access
- lock
- bmap
- ioctl
- poll
- write_buf
- read_buf
- flock
- fallocate
- copyfilerange
- lseek
- Creation of pipes, FIFO queues, Device files
- XAttrs for file or directory
- Hardlinks for file or directory
- Last access time and Last change time for any file or directory
- fsync() force deletes file from local cache and invalidates attribute cache. This forces blobfuse2 to refresh the file metadata and contents on next open call to that file.
- fsyncdir() invalidates metadata of that directory recursively. This forces blobfuse2 to refresh metadata of any child of that directory on next metadata query by kernel.
- Blobfuse2 does not support overlapping mount paths. While running multiple instances of Blobfuse2 make sure each instance has a unique and non-overlapping mount point.
- Blobfuse2 does not support co-existence with NFS on same mount path. Behavior in this case is undefined.
- For non-HNS accounts (flat name space), Blobfuse2 expects special directory marker files to identify any directory. If such files do not exist on storage and '--virtual-directory=true' cli flag is not provided, Blobfuse2 will fail to identify directories.
- On non-HNS accounts chmod operations are not permitted and Blobfuse2 will return back success in such cases.