OrangeFS

Versions and features

• Server-to-server communication infrastructure
• SSD option for storage of distributed metadata
• Full native Windows client support
• Replication for immutable files

• Direct interface for applications
• Client caching for the direct interface with multi-process single-system coherence
• Initial release of the webpack supporting WebDAV and S3 via Apache modules

• Updates, fixes and performance improvements

• Updates, fixes and performance improvements, native Hadoop support via JNI shim, support for newer Linux kernels

• Distributed Metadata for Directory Entries
• Capability-based security in 3 modes
• * Standard security
• * Key-based security
• * Certificate-based security with LDAP interface support
• Extended documentation

• Bug fixes and build changes to support recent distributions.
• The Linux upstream kernel client is the primary access method for Linux, the out-of-tree kernel module is deprecated.
• The OrangeFS Windows client has been refreshed

History

• Many were satisfied with the design goals of PVFS2 and needed it to remain relatively unchanged for future stability
• Others envisioned PVFS2 as a foundation on which to build an entirely new set of design objectives for more advanced applications of the future.

• * Proposed conversion of OrangeFS readahead to use folios, replacing traditional page-based handling, to improve read performance and align with modern kernel page cache trends.
• * Initial discussions and patches to simplify argument passing in OrangeFS's readdir functions, enhancing code clarity

• * Merged folio-based readahead for OrangeFS into the mm-unstable branch, improving read efficiency.
• * Fixed an incorrect error code in orangefs_sysfs_init, enhancing sysfs initialization reliability.
• * Introduced a splice-read stub for OrangeFS, enabling more efficient I/O operations via splice.
• * Converted OrangeFS to new ctime accessor functions, aligning with VFS timestamp handling improvements

• * Continued refinement of multigrain timestamp support, with OrangeFS adopting new atime/mtime accessors.
• * Moved orangefs_xattr_handlers to.rodata, improving security and efficiency by making it read-only.
• * Proposed replacing strncpy with strscpy in OrangeFS, addressing string truncation risks.

• * Finalized conversion to new timestamp accessors for atime, mtime, and ctime, ensuring VFS compatibility.
• * Simplified readdir argument passing, improving directory operation efficiency.
• * Removed unused ORANGEFS_CACHE_CREATE_FLAGS, streamlining code.

• * Fixed a NULL dereference in orangefs_mount, reported by Julia Lawall and kernel test robot, preventing mount crashes.
• * Addressed a superblock reference count leak, improving resource management.
• * Moved debug code to orangefs-debugfs.c, enhancing code organization.
• * Replaced strncpy with strscpy across OrangeFS, resolving truncation warnings and improving string safety.
• * Submitted updates for Linux 6.9, including string handling and debug code improvements.

• * Removed PG_error flag usage in OrangeFS, simplifying error handling as part of a kernel-wide cleanup.
• * Converted read/write operations to use iterators, replacing deprecated ->read/->write methods.
• * Finalized out-of-bounds fsid access fix, backported to multiple stable kernels.

• * Converted orangefs_write_begin and orangefs_write_end to use folios, enhancing write performance.
• * Rejected a duplicate fsid access issue, confirming prior fix sufficiency.
• * Submitted OrangeFS changes for Linux 6.12, likely including folio and stability fixes.

• * Reported and fixed a slab-out-of-bounds read in orangefs_debug_write, triggered by syzbot, preventing crashes or corruption.
• * Replaced strcpy with strscpy in orangefs_inode_getxattr, further improving string safety.
• * Reviewed folio memory management enhancements, supporting OrangeFS’s folio integration.

• * Fixed out-of-bounds debug_write issue, backported to 4.19–6.13, with multiple pull requests.
• * Improved orangefs_d_revalidate with stable VFS parent inode and name handling, and ensured debugfs safety.
• * Submitted a 9-patch series for Linux 6.15, including file size truncation fix, folio writeback enhancements, and code simplifications.
• * Fixed file size truncation bug, backported to 5.4–6.14, preventing data loss.
• * Addressed a 6.14 writeback regression, improving performance.
• * Fixed counting code bugs and resolved VFS integration issues.

• * Continued addressing the 6.14 writeback regression, ensuring data integrity.
• * Submitted OrangeFS pull requests for Linux 6.16, including ongoing fixes and enhancements.
• * Resolved duplicate patch issues in linux-next, improving integration.

• documentation updates.
• removed c-shell dependency.
• change default build linkage from static to shared
• change from bdb to lmdb for default database
• updated configure script for usrint.
• some code refactored to adhere to coding standards.

• truncate bstream before set size attr in case former fails.
• lock local lock for truncate as well as set size attr.

• adjust some functions in state machines to match their prototypes.
• add some gossip statements, and improve some others.
• restored ability to invoke "make dist".
• fixed db_display string processing.
• fixed bstream-resize truncation failure error handling.