The ioerror mount option for VxFS filesystems sets the policy how the filesystem should handle I/O errors. There are five policies for handling these errors:
- disable
- nodisable
- wdisable
- mwdisable
- mdisable
Disable means that the system disables the erroneous filesystem by refusing any further I/O operations to it. In these cases one needs to umount and repair the filesystem, a full fsck is needed in most of the cases.
Nodisable does the opposite: although the VX_DATAIOERR flag will be set on the filesystem superblock, the filesystem countinues to run in degraded mode. For metadata failures, VX_FULLFSCK flag will be set, in case of a metadata write failure also the VX_METAIOERR flag will be marked.
The wdisable and mwdisable policies are for disabling the filesystem in case of write or metadata-write failures, otherwise the filesystem will be declared as degraded. The latter is the default mount option for local filesystems.
With the mdisable policy the filesystem will be disabled only if a metadata read or write fails. If the I/O error is related to the data segment, the system continues to operate in degraded mode. This is the default policy in case of shared cluster filesystems.
Source: symantec.com
