Release Notes

hp StorageWorks File System Extender Release Notes 21
• sockets
• pipes
Consequently, these files cannot be created on an HSM file system.
Directories with sticky bit not supported
The FSE implementation does not support handling of “sticky directories”, that is, directories with the sticky
bit set. Such directories cannot be copied to HSM file systems and the sticky bit cannot be applied to
existing directories on HSM file systems.
The sticky bit is represented with the attribute character t or T in the output of the ls command, as shown
in this example:
If you try to copy a sticky directory to an HSM file system or try to modify the attributes of an existing
directory on an HSM file system to include the sticky bit, the cp and chmod commands fail with the
following errors:
The reason for this limitation is Ext3 extended attributes, which are used by the FSE implementation, but
cannot be added to sticky directories.
Limited symbolic link support
Even though an HSM file system allows symbolic links to be created on it and permits other file operations
on these links, they are never migrated to FSE media. Since they occupy very little disk space and do not
obstruct normal HSM file system operation, this is not a drawback.
The problem arises if the HSM file system is completely destroyed and has to be recovered from FSE media.
In this case, the recovery of the HSM file system is not complete because of the missing symbolic links on
the FSE media.
File size limitation
On Linux hosts, the maximum size of a file created on an HSM file system is limited by the underlying Ext3
file system size limit. In the current version of the Linux kernel (2.4), this limit is two terabytes (2 TB).
Generally, the maximum size of a file that can be migrated from an HSM file system to FSE media is limited
by the size of disk buffer and by the amount of free space on all FSE media in the respective FSE media
pool. Files that are larger than the disk buffer cannot be migrated, and therefore, they permanently occupy
space on the HSM file system.
Append-only and immutable files cannot be released
Files on an HSM file system with the append-only or immutable status (set by the appropriate special file
attributes) cannot be released. They are migrated in the same way as all other files but are not released.
Consequently, these files are not recalled and permanently occupy space on the HSM file system.
Pathname length limitation
An FSE implementation cannot gain control of files whose full pathname exceeds 4096 characters. A full
pathname consists of the absolute path of a file (including path separators—slashes) and the name of the
file. An absolute path is the path from the root directory of the root file system. Files whose pathnames are
too long remain dirty forever; they are not migrated to FSE media and therefore cannot be released from
the HSM file system.
-rwxrwxrwx 1 root root 16932 Nov 26 08:02 file
drwxrwxrwx 2 root root 4096 Nov 26 09:38 non_sticky_dir
drwxrwxrwt 2 root root 4096 Nov 26 08:26 sticky_dir
drwxrwxrwT 2 root root 4096 Nov 26 09:36 sticky_dir_2
cp: cannot create directory `/mnt/hsmfs01/test': Operation not permitted
chmod: changing permissions of `/mnt/hsmfs01/test': Operation not
permitted