ANS4023E file input/output error

Resolving the problem

1) Most commonly, due to filesystem corruption.
2) Anti-virus software is attempting to scan the filesystem during the TSM backup.
3) The operating system is not allowing the filesystem or directory to be backed up.
4) There are other problems with filesystem besides filesystem corruption, such as stale NFS mounts.

Filesystem corruption typically requires a filesystem check to fix. If the corruption is with a UNIX '/' (root) filesystem, halt the OS on the TSM Client, then reboot in single user or maintenance mode to unmount and run 'fsck' on the '/' filesystem. If the corruption is on a Windows filesystem, run the MS-DOS utility 'chkdsk drive_letter: /f' to fix filesystem errors, which typically requires a system reboot. Filesystem corruption can exist even on RAID arrays (despite error correction), and can occur even if the hardware does not list errors at the OS-level.

If the filesystem corruption persists after running 'fsck' or 'chkdsk /f', it is recommended that you consult with the OS support group to determine how to recover the data or work around the corruption.

If the failing filesystem is NFS-mounted (network attached), investigate whether the issue could be caused by a stale NFS mount. The TSM Client always gathers information about all mounted filesystems at the moment the backup begins. The DOMAIN option indicates which filesystems will be backed up (if 'incremental' command is used without arguments), but the information about all mounted filesystems is gathered anyway. To avoid problems with stale mount points, the TSM Client has a special option called NFSTIMEOUT which can allow the TSM Client to bypass a stale mountpoint. This options defines how long (in seconds) the TSM Client should wait for the answer from the NFS server; the default is zero (wait forever). Using a short NFSTIMEOUT value (between 1 and 10 seconds) will allow the TSM Client to fail only the NFS-mounted filesystem and continue the backup.

The RC=164 error is generated after the TSM Client attempts OS-level "readdir_r" call for a particular directory structure. The TSM Client is trying to open a directory to get more information whether it be for the next directory entry, attributes, availability, and so on. From the return code alone, the OS does not tell TSM if the object is no longer available (deleted) or if there is just a temporary problem. The TSM Client is designed to treat this error as a hard failure because otherwise it would have to make assumptions about deleted files; possibly adding the directory to an expire list, causing valid data on the TSM Server to be deleted.
Previous
Next Post »