fsck_vxfs(1M)

NAME

fsck_vxfs: fsck — check and repair a VxFS file system

SYNOPSIS

fsck [-F vxfs] [-V] [-mnNpPsyY] [-pP] [-o p] [special ... ]

fsck [-F vxfs] [-V] [-mnNpPsyY] [-o full,mounted,nolog] [special ... ]

DESCRIPTION

fsck checks VxFS file systems for consistency. Because VxFS records pending file system updates in an intent log, fsck typically replays the intent log instead of doing a full structural file system check. You can use options (-o full or -y) to force a full structural file system check.

special specifies one or more special character devices, for example, /dev/rdsk/c1t0d0.

If multiple devices are specified, each device is checked in turn unless the -P option or the -o p suboption is also specified, in which case the devices are checked in parallel. If special is not specified, fsck prompts you with each file system listed in /etc/fstab to determine which file system to check, unless you specify -y or -Y, to automatically answer yes to the prompts.

Options

fsck recognizes the following options:

-F vxfs: Specify the VxFS file system type.
-m: Check whether or not the file system is marked clean. This option does not validate the file system. The file system could have been corrupted since it was marked clean (for example, by a system crash), and if so, a mount could fail. In that case, a full fsck would be required to clean it. Use fsck -n to test for file system corruption.
-n|-N: Assume a "no" response to all prompts by fsck; do not open the file system for writing, do not replay the intent log. A full file system check is performed.
-o specific_options: Specify VxFS file system specific options. See the subsection The -o Specific Options below.
-p: Produce messages that identify the device being checked.
-P: With VxFS, -P is used by fsck by default; it does not provide any functionality.
-s: Safe performance mode. To improve performance, a system wide sync() will not be issued (see sync(2)).
-V: Echo the completed command line, but do not execute the command. The command line is generated by incorporating the user specified options and other information derived from /etc/fstab. This option allows the user to verify the command line.
-y|-Y: Assume a "yes" response to all prompts by fsck. Additionally, if the file system requires a full file system check after the log replay, or if the nolog suboption causes the log replay to be skipped and the file system is not clean, then a full file system check is performed.

Because VxFS maintains an intent log, a complete check is generally not required; the default is to replay the intent log only. If fsck_vxfs detects file system damage or the log replay operation detects damage, an indication that a complete check is required is placed in the super-block. In this case, if the -y option was specified, the full check will be run after the log replay. If the -y option was not used, fsck must be run again, with the -o full option to perform the full structural check.

Operands

fsck recognizes the following operand:

special: Name of one or more special character devices which contain VxFS file systems.

The -o Specific Options

The -o option specifies VxFS file system specific options. These options can be a combination of the following in a comma-separated list:

full

Perform a full file system check.

mounted

Allows a full check of a mounted file system. -o mounted is only used internally as part of the primary cluster node recovery process after the primary fails. Never enter this option from the command line as it can destroy a file system if not used correctly.

nolog

Do not perform log replay. This option may be used if the log area was physically damaged.

Note: Use the -n option to verify whether there are file system inconsistencies. Use fsck -o full,nolog to fix a corrupted file system and avoid a log replay. If you run fsck -o full without nolog on a clean file system, it replays the intent log and performs a full file system check.

p

Allows parallel log replay for several VxFS file systems. Each message from fsck is prefixed with the device name to identify the device. This suboption does not perform a full file system check in parallel; that is still done sequentially on each device, even when multiple devices are specified. This option is compatible only with the -y|-Y option (that is, non-interactive full file system check), in which case a log replay is done in parallel on all specified devices. A sequential full file system check is performed on devices where needed. The number of devices that can be checked in parallel is determined by the amount of physical memory in the system. One instance of fsck on a single device can consume up to a maximum of 32 megabytes of memory.

Check a File System

A full check looks for the following inconsistencies:

Blocks claimed by more than one inode or the free list.
Blocks claimed by an inode outside the range of the file system.
Incorrect link counts.
Size checks:
-
Incorrect number of blocks.
-
Directory entry format.
Bad inode format.
Blocks not accounted for anywhere.
Directory checks:
-
File pointing to unallocated inode.
-
Inode number out of range.
-
Linkage to parent directory.
-
Hash chain linkage.
-
Free space count.
Super-block checks:
-
Checksum mismatch.
-
More blocks for inodes than there are in the file system.
Structural Files:
-
Fileset headers.
-
Object Location Table (OLT).
-
Inode list files.
-
Inode allocation summary files.
-
Attribute files (including Access Control Lists).
-
Attribute link counts.
Bad free block list format.
Total free block and/or free inode count incorrect.

Lost and Found Directory

Orphaned files and directories (allocated but unreferenced) are, with the user's agreement, reconnected by placing them in the lost+found directory. The name assigned is the inode number. The only restriction is that the directory lost+found must already exist in the file system's root directory.

Notes

Checking the raw device is almost always faster.

Unlike 2.x and earlier releases of VxFS, a full file system check does not always perform pending extended inode operations. Some extended operations can only be processed when the file system is mounted. A file system that has been marked CLEAN can still contain extended operations.

If a structural flaw is detected during the intent log replay, the full fsck flag is set on the file system without operator interaction.

If fsck encounters a large file on an older OS version, the command stops without completing the file system check.

RETURN VALUES

Structural errors discovered during a full check are displayed on standard output. Responses required during a full check are read from standard input.

The following return codes are used for the -m option for all devices other than the one used by the root file system:

0: The file system is unmounted and clean.
32: The file system is unmounted and needs checking.
33: The file system is mounted.
34: The stat of the device failed.
Other: The state could not be determined because of an error.

The following return codes are used for the -m option for the device used by the root file system:

0: The root file system is mounted read-only and is clean, or the root file system is mounted read/write and therefore is clean.
32: The root file system is mounted read-only and needs checking.
34: The stat of the device failed.
Other: The state could not be determined because of an error.

In most cases, fsck prints the following messages:

log replay in progress 
replay complete - marking super-block as CLEAN 

If the file system is already clean, fsck prints the following message instead:

file system is clean - log replay is not required 

If fsck prints any other messages, a full structural check is needed. If the -y option is specified, fsck performs (if necessary) a full check after running the intent log replay. If the -y option is not used, fsck must be invoked with the -o full option to perform a full structural check.

If -o p or -P is specified, fsck prints the following messages for a device, for example /dev/rdsk/c0t0d0:

/dev/rdsk/c0t0d0:log replay in progress 
/dev/rdsk/c0t0d0:replay complete - marking super-block as CLEAN 

DIAGNOSTICS

All error messages that relate to the contents of a file system produced during a log replay are displayed on standard output. All I/O failures and exit messages are displayed on standard error.

WARNINGS

-o mounted allows a full check of a mounted file system. -o mounted is only used internally as part of the primary cluster node recovery process after the primary fails. Never enter this option from the command line as it can destroy a file system if not used correctly.

The -s (safe performance mode) option will be obsoleted in future releases.