Chapter 20 Resolving UFS File System Inconsistencies (Tasks)

This chapter describes the fsck error messages and the possible responses you can make to resolve the error messages.

This is a list of the information in this chapter:

• General fsck Error Messages

• Initialization Phase fsck Messages

• Phase 1: Check Blocks and Sizes Messages

• Phase 1B: Rescan for More DUPS Messages

• Solaris 10: Solaris 10: Phase 1B: Rescan for More DUPS Messages

• Phase 2: Check Path Names Messages

• Phase 3: Check Connectivity Messages

• Phase 4: Check Reference Counts Messages

• Phase 5: Check Cylinder Groups Messages

• Solaris 10: Phase 5: Check Cylinder Groups Messages

• fsck Summary Messages

• Solaris 10: Cleanup Phase Messages

For information about the fsck command and how to use it to check file system integrity, see Chapter 21, Checking UFS File System Consistency (Tasks), in System Administration Guide: Devices and File Systems.

New fsck Error Messages

Solaris 10 6/06: In this Solaris release, error messages that are displayed when you run the fsck command have changed. This section includes the revised fsck error messages. If you are not running at least the Solaris 10 6/06 release, refer to the error messages in this chapter that are labeled “Solaris 10”. For a detailed description of all the fsck improvements in the current Solaris release, see System Administration Guide: Devices and File Systems.

fsck Error Messages

Normally, the fsck command is run non-interactively to preen the file systems after an abrupt system halt in which the latest file system changes were not written to disk. Preening automatically fixes any basic file system inconsistencies and does not try to repair more serious errors. While preening a file system, the fsck command fixes the inconsistencies it expects from such an abrupt halt. For more serious conditions, the command reports the error and terminates.

When you run the fsck command interactively, it reports each inconsistency found and fixes innocuous errors. However, for more serious errors, the command reports the inconsistency and prompts you to choose a response. When you run the fsck command with the -y or -n options, your response is predefined as yes or no to the default response suggested by the fsck command for each error condition.

Some corrective actions will result in some loss of data. The amount and severity of data loss might be determined from the fsck diagnostic output.

The fsck command is a multipass file system check program. Each pass invokes a different phase of the fsck command with different sets of messages. After initialization, the fsck command performs successive passes over each file system, checking blocks and sizes, path names, connectivity, reference counts, and the map of free blocks (possibly rebuilding it). It also performs some cleanup.

The phases (passes) performed by the UFS version of the fsck command are:

• Initialization

• Phase 1 – Check Blocks and Sizes

• Phase 2a – Check Duplicated Names

• Phase 2b – Check Pathnames

• Phase 3 – Check Connectivity

• Phase 3b – Verify Shadows/ACLs

• Phase 4 – Check Reference Counts

• Phase 5 – Check Cylinder Groups

The next sections describe the error conditions that might be detected in each phase, the messages and prompts that result, and possible responses you can make.

Messages that might appear in more than one phase are described in General fsck Error Messages. Otherwise, messages are organized alphabetically by the phases in which they occur.

The following table lists many of the abbreviations included in the fsck error messages.

Many of the messages also include variable fields, such as inode numbers, which are represented in this book by an italicized term, such as inode-number. For example, this screen message:

INCORRECT BLOCK COUNT I=2529

is shown as follows:

INCORRECT BLOCK COUNT I=inode-number

General fsck Error Messages

The error messages in this section might be displayed in any phase after initialization. Although they offer the option to continue, it is generally best to regard them as fatal. They reflect a serious system failure and should be handled immediately. When confronted with such a message, terminate the program by entering n(o). If you cannot determine what caused the problem, contact your local service provider or another qualified person.

CANNOT SEEK: BLK disk-block-number (CONTINUE)

Solaris 10:

CANNOT SEEK: BLK block-number (CONTINUE)

Cause

A request to move to the specified block number, disk-block-number, in the file system failed. This message indicates a serious problem, probably a hardware failure.

Solaris 10: A request to move to the specified block number, block-number, in the file system failed. This message indicates a serious problem, probably a hardware failure.

If you want to continue the file system check, fsck will retry the move and display a list of sector numbers that could not be moved. If the block was part of the virtual memory buffer cache, fsck will terminate with a fatal I/O error message.

Action

If the disk is experiencing hardware problems, the problem will persist. Run fsck again to recheck the file system.

If the recheck fails, contact your local service provider or another qualified person.

CANNOT READ: DISK BLOCK disk-block-number: I/O ERROR 
 CONTINUE?

Solaris 10:

CANNOT READ: DISK BLOCK block-number: I/O ERROR 
 CONTINUE?

Cause

A request to read the specified block number, disk-block-number, in the file system failed. The message indicates a serious problem, probably a hardware failure.

Solaris 10: A request to read a specified block number, block-number, in the file system failed. The message indicates a serious problem, probably a hardware failure.

If you want to continue the file system check, fsck will retry the read and display a list of sector numbers that could not be read. If the block was part of the virtual memory buffer cache, fsck will terminate with a fatal I/O error message. If fsck tries to write back one of the blocks on which the read failed, it will display the following message:

WRITING ZERO'ED BLOCK sector-numbersTO DISK

Action

If the disk is experiencing hardware problems, the problem will persist. Run fsck again to recheck the file system. If the recheck fails, contact your local service provider or another qualified person.

CANNOT WRITE: BLK disk-block-number (CONTINUE)

Solaris 10:

CANNOT WRITE: BLK block-number (CONTINUE)

Cause

A request to write the specified block number, disk-block-number, in the file system failed.

If you continue the file system check, fsck will retry the write and display a list of sector numbers that could not be written. If the block was part of the virtual memory buffer cache, fsck will terminate with a fatal I/O error message.

Solaris 10: A request to write a specified block number, block-number, in the file system failed.

If you continue the file system check, fsck will retry the write and display a list of sector numbers that could not be written. If the block was part of the virtual memory buffer cache, fsck will terminate with a fatal I/O error message.

Action

The disk might be write-protected. Check the write-protect lock on the drive. If the disk has hardware problems, the problem will persist. Run fsck again to recheck the file system. If the write-protect is not the problem or the recheck fails, contact your local service provider or another qualified person.

Initialization Phase fsck Messages

In the initialization phase, command-line syntax is checked. Before the file system check can be performed, fsck sets up tables and opens files.

The messages in this section relate to error conditions resulting from command-line options, memory requests, the opening of files, the status of files, file system size checks, and the creation of the scratch file. All such initialization errors terminate fsck when it is preening the file system.

Can't roll the log for device-name.

DISCARDING THE LOG MAY DISCARD PENDING TRANSACTIONS.
DISCARD THE LOG AND CONTINUE?

Cause

fsck was unable to flush the transaction log of a logging UFS file system prior to checking the file system for errors.

Action

Answering yes means the file system operations that were in the log, but had not been applied to the file system, are lost. In this case, fsck runs the same checks it always runs and asks the following question in phase 5:

FREE BLK COUNT(S) WRONG IN SUPERBLK (SALVAGE)

Answering yes at this point reclaims the blocks that were used for the log. The next time the file system is mounted with logging enabled, the log will be recreated.

Answering no preserves the log and exits, but the file system isn't mountable.

bad inode number inode-number to ginode

Cause

An internal error occurred because of a nonexistent inode inode-number. fsck exits.

Action

Contact your local service provider or another qualified person.

cannot alloc size-of-block map bytes for blockmap
cannot alloc size-of-free map bytes for freemap
cannot alloc size-of-state map bytes for statemap
cannot alloc size-of-lncntp bytes for lncntp

Cause

Request for memory for its internal tables failed. fsck terminates. This message indicates a serious system failure that should be handled immediately. This condition might occur if other processes are using a very large amount of system resources.

Action

Killing other processes might solve the problem. If not, contact your local service provider or another qualified person.

Can't open checklist file: filename

Cause

The file system checklist file filename (usually /etc/vfstab) cannot be opened for reading. fsck terminates.

Action

Check if the file exists and if its access modes permit read access.

Can't open filename

Cause

fsck cannot open file system filename. When running interactively, fsck ignores this file system and continues checking the next file system given.

Action

Check to see if read and write access to the raw device file for the file system is permitted.

Can't stat root

Cause

fsck request for statistics about the root directory failed. fsck terminates.

Action

This message indicates a serious system failure. Contact your local service provider or another qualified person.

Can't stat filename
Can't make sense out of name filename

Cause

fsck request for statistics about the file system filename failed. When running interactively, fsck ignores this file system and continues checking the next file system given.

Action

Check if the file system exists and check its access modes.

filename: (NO WRITE)

Cause

Either the -n option was specified or fsck could not open the file system filename for writing. When fsck is running in no-write mode, all diagnostic messages are displayed, but fsck does not attempt to fix anything.

Action

If -n was not specified, check the type of the file specified. It might be the name of a regular file.

IMPOSSIBLE MINFREE=percent IN SUPERBLOCK (SET TO DEFAULT)

Cause

The superblock minimum space percentage is greater than 99 percent or less than 0 percent.

Action

To set the minfree parameter to the default 10 percent, type y at the default prompt. To ignore the error condition, type n at the default prompt.

filename: BAD SUPER BLOCK: message
USE AN ALTERNATE SUPER-BLOCK TO SUPPLY NEEDED INFORMATION;
e.g., fsck[-f ufs] -o b=# [special ...]
where # is the alternate superblock.  See fsck_ufs(1M)

Cause

The superblock has been corrupted.

Action

One of the following messages might be displayed:

CPG OUT OF RANGE FRAGS PER BLOCK OR FRAGSIZE WRONG INODES PER GROUP OUT OF RANGE INOPB NONSENSICAL RELATIVE TO BSIZE MAGIC NUMBER WRONG NCG OUT OF RANGE NCYL IS INCONSISTENT WITH NCG*CPG NUMBER OF DATA BLOCKS OUT OF RANGE NUMBER OF DIRECTORIES OUT OF RANGE ROTATIONAL POSITION TABLE SIZE OUT OF RANGE SIZE OF CYLINDER GROUP SUMMARY AREA WRONG SIZE TOO LARGE BAD VALUES IN SUPERBLOCK

Try to rerun fsck with an alternative superblock. Specifying block 32 is a good first choice. You can locate an alternative copy of the superblock by running the newfs -N command on the slice. Be sure to specify the -N option; otherwise, newfs overwrites the existing file system.

UNDEFINED OPTIMIZATION IN SUPERBLOCK (SET TO DEFAULT)

Cause

The superblock optimization parameter is neither OPT_TIME nor OPT_SPACE.

Action

To minimize the time to perform operations on the file system, type y at the SET TO DEFAULT prompt. To ignore this error condition, type n.

Phase 1: Check Blocks and Sizes Messages

This phase checks the inode list. It reports error conditions encountered while:

• Checking inode types

• Setting up the zero-link-count table

• Examining inode block numbers for bad or duplicate blocks

• Checking inode size

• Checking inode format

All errors in this phase except INCORRECT BLOCK COUNT, PARTIALLY TRUNCATED INODE, PARTIALLY ALLOCATED INODE, and UNKNOWN FILE TYPE terminate fsck when it is preening a file system.

These messages (in alphabetical order) might occur in phase 1:

block-number BAD I=inode-number

Cause

Inode inode-number contains a block number block-number with a number lower than the number of the first data block in the file system or greater than the number of the last block in the file system. This error condition might generate the EXCESSIVE BAD BLKS error message in phase 1 if inode inode-number has too many block numbers outside the file system range. This error condition generates the BAD/DUP error message in phases 2 and 4.

Action

N/A

BAD MODE: MAKE IT A FILE?

Cause

The status of a given inode is set to all 1s, indicating file system damage. This message does not indicate physical disk damage, unless it is displayed repeatedly after fsck -y has been run.

Action

Type y to reinitialize the inode to a reasonable value.

BAD STATE state-number TO BLKERR

Cause

An internal error has scrambled the fsck state map so that it shows the impossible value state-number. fsck exits immediately.

Action

Contact your local service provider or another qualified person.

fragment-number DUP I=inode-number

Solaris 10:

block-number DUP I=inode-number

Cause

Inode inode-number contains a block number fragment-number, which is already claimed by the same or another inode. This error condition might generate the EXCESSIVE DUP BLKS error message in phase 1 if inode inode-number has too many block numbers claimed by the same or another inode. This error condition invokes phase 1B and generates the BAD/DUP error messages in phases 2 and 4.

Solaris 10: Inode inode-number contains a block number block-number, which is already claimed by the same or another inode. This error condition might generate the EXCESSIVE DUP BLKS error message in phase 1 if inode inode-number has too many block numbers claimed by the same or another inode. This error condition invokes phase 1B and generates the BAD/DUP error messages in phases 2 and 4.

Action

N/A

DUP TABLE OVERFLOW (CONTINUE)

Cause

fsck could not allocate memory to track duplicate fragments. If the -o p option is specified, the program terminates.

Solaris 10: There is no more room in an internal table in fsck containing duplicate block numbers. If the -o p option is specified, the program terminates.

Action

To continue the program, type y at the CONTINUE prompt. When this error occurs, a complete check of the file system is not possible. If another duplicate fragment is found, this error condition repeats. Increase the amount of virtual memory available (by killing some processes, increasing swap space) and run fsck again to recheck the file system. To terminate the program, type n.

Solaris 10: To continue the program, type y at the CONTINUE prompt. When this error occurs, a complete check of the file system is not possible. If another duplicate block is found, this error condition repeats. Increase the amount of virtual memory available (by killing some processes, increasing swap space) and run fsck again to recheck the file system. To terminate the program, type n.

EXCESSIVE BAD FRAGMENTS I=inode-number (CONTINUE)

Solaris 10:

EXCESSIVE BAD BLOCKS I=inode-number (CONTINUE)

Cause

Too many (usually more than 10) fragments indicate an invalid disk address. If the -o p (preen) option is specified, the program terminates.

Solaris 10: Too many (usually more than 10) blocks have a number lower than the number of the first data block in the file system or greater than the number of the last block in the file system associated with inode inode-number. If the -o p (preen) option is specified, the program terminates.

Action

To continue the program, type y at the CONTINUE prompt. When this error occurs, a complete check of the file system is not possible. You should run fsck again to recheck the file system. To terminate the program, type n.

EXCESSIVE DUP BLKSDUPLICATE FRAGMENTS I=inode-number (CONTINUE)

Solaris 10:

EXCESSIVE DUP BLKS I=inode-number (CONTINUE)

Cause

Too many (usually more than 10) fragments are claimed by the same or another inode or by a free-list. If the -o p option is specified, the program terminates.

Solaris 10: Too many (usually more than 10) blocks are claimed by the same or another inode or by a free-list. If the -o p option is specified, the program terminates.

Action

To continue the program, type y at the CONTINUE prompt. When this error occurs, a complete check of the file system is not possible. You should run fsck again to recheck the file system. To terminate the program, type n.

INCORRECT DISK BLOCK COUNT I=inode-number (number-of-BAD-DUP-or-missing-blocks
should be number-of-blocks-in-filesystem) (CORRECT)

Solaris 10:

INCORRECT BLOCK COUNT I=inode-number (number-of-BAD-DUP-or-missing-blocks
should be number-of-blocks-in-filesystem) (CORRECT)

Cause

The disk block count for inode inode-number is incorrect.. When preening, fsck corrects the count.

Solaris 10: The block count for inode inode-number is number-of-BAD-DUP-or-missing-blocks, but should be number-of-blocks-in-filesystem. When preening, fsck corrects the count.

Action

To correct the disk block count of inode inode-number by number-of-blocks-in-file, type y at the CORRECT prompt. .

Solaris 10: To replace the block count of inode inode-number by number-of-blocks-in-filesystem, type y at the CORRECT prompt. To terminate the program, type n.

LINK COUNT TABLE OVERFLOW (CONTINUE)

Cause

There is no more room in an internal table for fsck containing allocated inodes with a link count of zero. If the -o p (preen) option is specified, the program exits and fsck has to be completed manually.

Action

To continue the program, type y at the CONTINUE prompt. If another allocated inode with a zero-link count is found, this error condition repeats. When this error occurs, a complete check of the file system is not possible. You should run fsck again to recheck the file system. Increase the virtual memory available by killing some processes or increasing swap space, then run fsck again. To terminate the program, type n.

PARTIALLY ALLOCATED INODE I=inode-number (CLEAR)

Cause

Inode inode-number is neither allocated nor unallocated. If the -o p (preen) option is specified, the inode is cleared.

Action

To deallocate the inode inode-number by zeroing out its contents, type y. This might generate the UNALLOCATED error condition in phase 2 for each directory entry pointing to this inode. To ignore the error condition, type n. A no response is appropriate only if you intend to take other measures to fix the problem.

PARTIALLY TRUNCATED INODE I=inode-number (SALVAGE)

Cause

fsck has found inode inode-number whose size is shorter than the number of fragments allocated to it. This condition occurs only if the system crashes while truncating a file. When preening the file system, fsck completes the truncation to the specified size.

Solaris 10: fsck has found inode inode-number whose size is shorter than the number of blocks allocated to it. This condition occurs only if the system crashes while truncating a file. When preening the file system, fsck completes the truncation to the specified size.

Action

To complete the truncation to the size specified in the inode, type y at the SALVAGE prompt. To ignore this error condition, type n.

UNKNOWN FILE TYPE I=inode-number (CLEAR)

Cause

The mode word of the inode inode-number shows that the inode is not a pipe, character device, block device, regular file, symbolic link, FIFO file, or directory inode. If the -o p option is specified, the inode is cleared.

Solaris 10: The mode word of the inode inode-number shows that the inode is not a pipe, special character inode, special block inode, regular inode, symbolic link, FIFO file, or directory inode. If the -o p option is specified, the inode is cleared.

Action

To deallocate the inode inode-number by zeroing its contents, which results in the UNALLOCATED error condition in phase 2 for each directory entry pointing to this inode, type y at the CLEAR prompt. To ignore this error condition, type n.

Solaris 10: Phase 1B: Rescan for More DUPS Messages

This sections contains phase 1B fsck messages in the current Solaris release.

When a duplicate fragment is found in the file system, this message is displayed:

fragment DUP I=inode-number

Cause

Inode inode-number contains a fragment number fragment-number that is already claimed by the same or another inode. This error condition generates the BAD/DUP error message in phase 2. Inodes that have overlapping fragments might be determined by examining this error condition and the DUP error condition in phase 1. This is simplified by the duplicate fragment report produced at the fsck run.

Action

When a duplicate block is found, the file system is rescanned to find the inode that previously claimed that block.

Phase 1B: Rescan for More DUPS Messages

This section contains fsck messages that are displayed in the Solaris 10 release.

When a duplicate block is found in the file system, this message is displayed:

block-number DUP I=inode-number

Cause

Inode inode-number contains a block number block-number that is already claimed by the same or another inode. This error condition generates the BAD/DUP error message in phase 2. Inodes that have overlapping blocks might be determined by examining this error condition and the DUP error condition in phase 1.

Action

When a duplicate block is found, the file system is rescanned to find the inode that previously claimed that block.

Phase 2: Check Path Names Messages

This phase removes directory entries pointing to bad inodes found in phases 1 and 1B. It reports error conditions resulting from:

• Incorrect root inode mode and status

• Directory inode pointers out of range

• Directory entries pointing to bad inodes

• Directory integrity checks

When the file system is being preened (-o -poption), all errors in this phase terminate fsck, except those related to directories not being a multiple of the block size, duplicate and bad blocks, inodes out of range, and extraneous hard links.

These messages (in alphabetical order) might occur in phase 2:

BAD INODE state-number TO DESCEND

Cause

An fsck internal error has passed an invalid state state-number to the routine that descends the file system directory structure. fsck exits.

Action

If this error message is displayed, contact your local service provider or another qualified person.

BAD INODE NUMBER FOR '.' I=inode-number OWNER=UID MODE=file-mode 
SIZE=file-size MTIME=modification-time DIR=filename (FIX)

Cause

A directory inode-number has been found whose inode number for “.” does not equal inode-number.

Action

To change the inode number for “.” to be equal to inode-number, type y at the FIX prompt To leave the inode numbers for “.” unchanged, type n.

BAD INODE NUMBER FOR '..' I=inode-number OWNER=UID MODE=file-mode 
SIZE=file-size MTIME=modification-time DIR=filename (FIX)

Cause

A directory inode-number has been found whose inode number for “..” does not equal the parent of inode-number.

Action

To change the inode number for “..” to be equal to the parent of inode-number, type y at the FIX prompt. (Note that “..'' in the root inode points to itself.) To leave the inode number for “..” unchanged, type n.

BAD RETURN STATE state-number FROM DESCEND

Cause

An fsck internal error has returned an impossible state state-number from the routine that descends the file system directory structure. fsck exits.

Action

If this message is displayed, contact your local service provider or another qualified person.

BAD STATE state-number FOR ROOT INODE

Cause

An internal error has assigned an impossible state state-number to the root inode. fsck exits.

Action

If this error message is displayed, contact your local service provider or another qualified person.

BAD STATE state-number FOR INODE=inode-number

Cause

An internal error has assigned an impossible state state-number to inode inode-number. fsck exits.

Action

If this error message is displayed, contact your local service provider or another qualified person.

DIRECTORY TOO SHORT I=inode-number OWNER=UID MODE=file-mode 
SIZE=file-size MTIME=modification-time DIR=filename (FIX)

Cause

A directory filename has been found whose size file-size is less than the minimum directory size. The owner UID, mode file-mode, size file-size, modify time modification-time, and directory name filename are displayed.

Action

To increase the size of the directory to the minimum directory size, type y at the FIX prompt. To ignore this directory, type n.

DIRECTORY filename: LENGTH file-size NOT MULTIPLE OF disk-block-size (ADJUST)

Solaris 10:

DIRECTORY filename: LENGTH file-size NOT MULTIPLE OF block-number (ADJUST)

Cause

A directory filename has been found with size file-size that is not a multiple of the directory block size disk-block-size.

Solaris 10:

A directory filename has been found with size file-size that is not a multiple of the directory block size block-number.

Action

To round up the length to the appropriate disk block size, type y. When preening the file system (-o p option), fsck only displays a warning and adjusts the directory. To ignore this condition, type n.

Solaris 10:

To round up the length to the appropriate block size, type y. When preening the file system (-o p option), fsck only displays a warning and adjusts the directory. To ignore this condition, type n.

DIRECTORY CORRUPTED I=inode-number OWNER=UID MODE=file-mode 
SIZE=file-size MTIME=modification-time DIR=filename (SALVAGE)

Cause

A directory with an inconsistent internal state has been found.

Action

To throw away all entries up to the next directory boundary (usually a 512-byte boundary), type y at the SALVAGE prompt. This drastic action can throw away up to 42 entries. Take this action only after other recovery efforts have failed. To skip to the next directory boundary and resume reading, but not modify the directory, type n.

DUP/BAD I=inode-number OWNER=O MODE=M SIZE=file-size 
MTIME=modification-time TYPE=filename (REMOVE)

Cause

Phase 1 or phase 1B found duplicate fragments or bad fragments associated with directory or file entry filename, inode inode-number. The owner UID, mode file-mode, size file-size, modification time modification-time, and directory or file name filename are displayed. If the -op (preen) option is specified, the duplicate/bad fragments are removed.

Solaris 10:

Phase 1 or phase 1B found duplicate blocks or bad blocks associated with directory or file entry filename, inode inode-number. The owner UID, mode file-mode, size file-size, modification time modification-time, and directory or file name filename are displayed. If the -op (preen) option is specified, the duplicate/bad blocks are removed.

Action

To remove the directory or file entry filename, type y at the REMOVE prompt. To ignore this error condition, type n.

DUPS/BAD IN ROOT INODE (REALLOCATE)

Cause

Phase 1 or phase 1B has found duplicate fragments or bad fragments in the root inode, ( inode number 20, of the file system.

Solaris 10:

Phase 1 or phase 1B has found duplicate blocks or bad blocks in the root inode (usually inode number 2 of the file system.

Action

To clear the existing contents of the root inode and reallocate it, type y at the REALLOCATE prompt. The files and directories usually found in the root inode will be recovered in phase 3 and put into the lost+found directory. If the attempt to allocate the root fails, fsck will exit with: CANNOT ALLOCATE ROOT INODE. Type n to get the CONTINUE prompt. Type: y to respond to the CONTINUE prompt, and ignore the DUPS/BAD error condition in the root inode and continue running the file system check. If the root inode is not correct, this might generate many other error messages. Type n to terminate the program.

EXTRA '.' ENTRY I=inode-number OWNER=UID MODE=file-mode 
SIZE=file-size MTIME=modification-time DIR=filename (FIX)

Cause

A directory inode-number has been found that has more than one entry for “.”.

Action

To remove the extra entry for “.” type y at the FIX prompt. To leave the directory unchanged, type n.

EXTRA '..' ENTRY I=inode-number OWNER=UID MODE=file-mode 
SIZE=file-size MTIME=modification-time DIR=filename (FIX)

Cause

A directory inode-number has been found that has more than one entry for “..” (the parent directory).

Action

To remove the extra entry for `..' (the parent directory), type y at the FIX prompt. To leave the directory unchanged, type n.

hard-link-number IS AN EXTRANEOUS HARD LINK TO A DIRECTORY filename (REMOVE)

Cause

fsck has found an extraneous hard link hard-link-number to a directory filename. When preening (-o p option), fsck ignores the extraneous hard links.

Action

To delete the extraneous entry hard-link-number type y at the REMOVE prompt. To ignore the error condition, type n.

inode-number OUT OF RANGE I=inode-number NAME=filename (REMOVE)

Cause

A directory entry filename has an inode number inode-number that is greater than the end of the inode list. If the -p (preen) option is specified, the inode will be removed automatically.

Action

To delete the directory entry filename type y at the REMOVE prompt. To ignore the error condition, type n.

MISSING '.' I=inode-number OWNER=UID MODE=file-mode SIZE=file-size 
MTIME=modification-time DIR=filename (FIX)

Cause

A directory inode-number has been found whose first entry (the entry for “.”) is unallocated.

Action

To build an entry for “.” with inode number equal to inode-number, type y at the FIX prompt. To leave the directory unchanged, type n.

MISSING '.' I=inode-number OWNER=UID MODE=file-mode SIZE=file-size 
MTIME=modification-time DIR=filename CANNOT FIX, FIRST ENTRY IN 
DIRECTORY CONTAINS filename

Cause

A directory inode-number has been found whose first entry is filename. fsck cannot resolve this problem.

Action

If this error message is displayed, contact your local service provider or another qualified person.

MISSING '.' I=inode-number OWNER=UID MODE=file-mode SIZE=file-size 
MTIME=modification-time DIR=filename CANNOT FIX, INSUFFICIENT 
SPACE TO ADD '.'

Cause

A directory inode-number has been found whose first entry is not “.”. fsck cannot resolve the problem.

Action

If this error message is displayed, contact your local service provider or another qualified person.

MISSING '..' I=inode-number OWNER=UID MODE=file-mode SIZE=file-size 
MTIME=modification-time DIR=filename (FIX)

Cause

A directory inode-number has been found whose second entry is unallocated.

Action

To build an entry for “..” with inode number equal to the parent of inode-number, type y at the FIX prompt. (Note that “..'' in the root inode points to itself.) To leave the directory unchanged, type n.

MISSING '..' I=inode-number OWNER=UID MODE=file-mode SIZE=file-size 
MTIME=modification-time DIR=filename CANNOT FIX, SECOND ENTRY IN 
DIRECTORY CONTAINS filename

Cause

A directory inode-number has been found whose second entry is filename. fsck cannot resolve this problem.

Action

If this error message is displayed, contact your local service provider or another qualified person.

MISSING '..' I=inode-number OWNER=UID MODE=file-mode SIZE=file-size 
MTIME=modification-time DIR=filename CANNOT FIX, INSUFFICIENT SPACE 
TO ADD '..'

Cause

A directory inode-number has been found whose second entry is not “..” (the parent directory). fsck cannot resolve this problem.

Action

If this error message is displayed, contact your local service provider or another qualified person.

NAME TOO LONG filename

Cause

An excessively long path name has been found, which usually indicates loops in the file system name space. This error can occur if a privileged user has made circular links to directories.

Action

Remove the circular links.

ROOT INODE UNALLOCATED (ALLOCATE)

Cause

The root inode (usually inode number 2) has no allocate-mode bits.

Action

To allocate inode 2 as the root inode, type y at the ALLOCATE prompt. The files and directories usually found in the root inode will be recovered in phase 3 and put into the lost+found directory. If the attempt to allocate the root inode fails, fsck displays this message and exits: CANNOT ALLOCATE ROOT INODE. To terminate the program, type n.

ROOT INODE NOT DIRECTORY (REALLOCATE)

Cause

The root inode (usually inode number 2) of the file system is not a directory inode.

Action

To clear the existing contents of the root inode and reallocate it, type y at the REALLOCATE prompt. The files and directories usually found in the root inode will be recovered in phase 3 and put into the lost+found directory. If the attempt to allocate the root inode fails, fsck displays this message and exits: CANNOT ALLOCATE ROOT INODE. To have fsck prompt with FIX, type n.

UNALLOCATED I=inode-number OWNER=UID MODE=file-mode SIZE=file-size 
MTIME=modification-time type=filename(REMOVE)

Cause

A directory or file entry filename points to an unallocated inode inode-number. The owner UID, mode file-mode, size file-size, modify time modification-time, and file name filename are displayed.

Action

To delete the directory entry filename, type y at the REMOVE prompt. To ignore the error condition, type n.

ZERO LENGTH DIRECTORY I=inode-number OWNER=UID MODE=file-mode 
SIZE=file-size MTIME=modification-time DIR=filename (REMOVE)

Cause

A directory entry filename has a size file-size that is zero. The owner UID, mode file-mode, size file-size, modify time modification-time, and directory name filename are displayed.

Action

To remove the directory entry filename, type y at the REMOVE prompt. This results in the BAD/DUP error message in phase 4. To ignore the error condition, type n.

Phase 3: Check Connectivity Messages

This phase checks the directories examined in phase 2 and reports error conditions resulting from:

• Unreferenced directories

• Missing or full lost+found directories

These messages (in alphabetical order) might occur in phase 3:

BAD INODE state-number TO DESCEND

Cause

An internal error has caused an impossible state state-number to be passed to the routine that descends the file system directory structure. fsck exits.

Action

If this occurs, contact your local service provider or another qualified person.

DIR I=inode-number1 CONNECTED. PARENT WAS I=inode-number2

Cause

This is an advisory message indicating a directory inode inode-number1 was successfully connected to the lost+found directory. The parent inode inode-number2 of the directory inode inode-number1 is replaced by the inode number of the lost+found directory.

Action

N/A

DIRECTORY filename LENGTH file-size NOT MULTIPLE OF disk-block-size (ADJUST)

Solaris 10:

DIRECTORY filename LENGTH file-size NOT MULTIPLE OF block-number (ADJUST)

Cause

A directory filename has been found with size file-size that is not a multiple of the directory block size B. (This condition can recur in phase 3 if it is not adjusted in phase 2.)

Action

To round up the length to the appropriate disk block size, type y at the ADJUST prompt. When preening, fsck displays a warning and adjusts the directory. To ignore this error condition, type n.

Solaris 10:

To round up the length to the appropriate block size, type y at the ADJUST prompt. When preening, fsck displays a warning and adjusts the directory. To ignore this error condition, type n.

lost+found IS NOT A DIRECTORY (REALLOCATE)

Cause

The entry for lost+found is not a directory.

Action

To allocate a directory inode and change the lost+found directory to reference it, type y at the REALLOCATE prompt. The previous inode reference by the lost+found directory is not cleared and it will either be reclaimed as an unreferenced inode or have its link count adjusted later in this phase. Inability to create a lost+found directory displays the message: SORRY. CANNOT CREATE lost+found DIRECTORY and aborts the attempt to link up the lost inode, which generates the UNREF error message in phase 4. To abort the attempt to link up the lost inode, which generates the UNREF error message in phase 4, type n.

NO lost+found DIRECTORY (CREATE)

Cause

There is no lost+found directory in the root directory of the file system. When preening, fsck tries to create a lost+found directory.

Action

To create a lost+found directory in the root of the file system, type y at the CREATE prompt. This might lead to the message NO SPACE LEFT IN / (EXPAND). If the lost+found directory cannot be created, fsck displays the message: SORRY. CANNOT CREATE lost+found DIRECTORY and aborts the attempt to link up the lost inode. This in turn generates the UNREF error message later in phase 4. To abort the attempt to link up the lost inode, type n.

NO SPACE LEFT IN /lost+found (EXPAND)

Cause

Another entry cannot be added to the lost+found directory in the root directory of the file system because no space is available. When preening, fsck expands the lost+found directory.

Action

To expand the lost+found directory to make room for the new entry, type y at the EXPAND prompt. If the attempted expansion fails, fsck displays: SORRY. NO SPACE IN lost+found DIRECTORY and aborts the request to link a file to the lost+found directory. This error generates the UNREF error message later in phase 4. Delete any unnecessary entries in the lost+found directory. This error terminates fsck when preening is in effect. To abort the attempt to link up the lost inode, type n.

UNREF DIR I=inode-number OWNER=UID MODE=file-mode SIZE=file-size 
MTIME=modification-time (RECONNECT)

Cause

The directory inode inode-number was not connected to a directory entry when the file system was traversed. The owner UID, mode file-mode, size file-size, and modification time modification-time of directory inode inode-number are displayed. When preening, fsck reconnects the non-empty directory inode if the directory size is non-zero. Otherwise, fsck clears the directory inode.

Action

To reconnect the directory inode inode-number into the lost+found directory, type y at the RECONNECT prompt. If the directory is successfully reconnected, a CONNECTED message is displayed. Otherwise, one of the lost+found error messages is displayed. To ignore this error condition, type n. This error causes the UNREF error condition in phase 4.

Phase 4: Check Reference Counts Messages

This phase checks the link count information obtained in phases 2 and 3. It reports error conditions resulting from:

• Unreferenced files

• A missing or full lost+found directory

• Incorrect link counts for files, directories, symbolic links, or special files

• Unreferenced files, symbolic links, and directories

• Bad or duplicate fragments in files and directories

  Solaris 10:

  Bad or duplicate blocks in files and directories

• Incorrect total free-inode counts

All errors in this phase (except running out of space in the lost+found directory) are correctable when the file system is being preened.

These messages (in alphabetical order) might occur in phase 4:

BAD/DUP type I=inode-number OWNER=UID MODE=file-mode SIZE=file-size 
MTIME=modification-time (CLEAR)

Cause

Phase 1 or phase 1B found duplicate fragments or bad fragments associated with file or directory inode inode-number. The owner UID, mode file-mode, size file-size, and modification time modification-time of inode inode-number are displayed.

Solaris 10:

Phase 1 or phase 1B found duplicate blocks or bad blocks associated with file or directory inode inode-number. The owner UID, mode file-mode, size file-size, and modification time modification-time of inode inode-number are displayed.

Action

To deallocate inode inode-number by zeroing its contents, type y at the CLEAR prompt. To ignore this error condition, type n.

(CLEAR)

Cause

The inode mentioned in the UNREF error message immediately preceding cannot be reconnected. This message does not display if the file system is being preened because lack of space to reconnect files terminates fsck.

Action

To deallocate the inode by zeroing out its contents, type y at the CLEAR prompt. To ignore the preceding error condition, type n.

LINK COUNT type I=inode-number OWNER=UID MODE=file-mode 
SIZE=file-size
MTIME=modification-time COUNT link-count SHOULD BE 
corrected-link-count (ADJUST)

Cause

The link count for directory or file inode inode-number is link-count but should be corrected-link-count. The owner UID, mode file-mode, size file-size, and modification time modification-time of inode inode-number are displayed. If the -o p option is specified, the link count is adjusted unless the number of references is increasing. This condition does not occur unless there is a hardware failure. When the number of references is increasing during preening, fsck displays this message and exits: LINK COUNT INCREASING

Action

To replace the link count of directory or file inode inode-number with corrected-link-count, type y at the ADJUST prompt. To ignore this error condition, type n.

lost+found IS NOT A DIRECTORY (REALLOCATE)

Cause

The entry for lost+found is not a directory.

Action

To allocate a directory inode and change the lost+found directory to reference it, type y at the REALLOCATE prompt. The previous inode reference by the lost+found directory is not cleared. It will either be reclaimed as an unreferenced inode or have its link count adjusted later in this phase. Inability to create a lost+found directory displays this message: SORRY. CANNOT CREATE lost+found DIRECTORY and aborts the attempt to link up the lost inode. This error generates the UNREF error message later in phase 4. To abort the attempt to link up the lost inode, type n.

NO lost+found DIRECTORY (CREATE)

Cause

There is no lost+found directory in the root directory of the file system. When preening, fsck tries to create a lost+found directory.

Action

To create a lost+found directory in the root of the file system, type y at the CREATE prompt. If the lost+found directory cannot be created, fsck displays the message: SORRY. CANNOT CREATE lost+found DIRECTORY and aborts the attempt to link up the lost inode. This error in turn generates the UNREF error message later in phase 4. To abort the attempt to link up the lost inode, type n.

NO SPACE LEFT IN / lost+found (EXPAND)

Cause

There is no space to add another entry to the lost+found directory in the root directory of the file system. When preening, fsck expands the lost+found directory.

Action

To expand the lost+found directory to make room for the new entry, type y at the EXPAND prompt. If the attempted expansion fails, fsck displays the message: SORRY. NO SPACE IN lost+found DIRECTORY and aborts the request to link a file to the lost+found directory. This error generates the UNREF error message later in phase 4. Delete any unnecessary entries in the lost+found directory. This error terminates fsck when preening (-o p option) is in effect. To abort the attempt to link up the lost inode, type n.

UNREF FILE I=inode-number OWNER=UID MODE=file-mode SIZE=file-size 
MTIME=modification-time (RECONNECT)

Cause

File inode inode-number was not connected to a directory entry when the file system was traversed. The owner UID, mode file-mode, size file-size, and modification time modification-time of inode inode-number are displayed. When fsck is preening, the file is cleared if either its size or its link count is zero; otherwise, it is reconnected.

Action

To reconnect inode inode-number to the file system in the lost+found directory, type y. This error might generate the lost+found error message in phase 4 if there are problems connecting inode inode-number to the lost+found directory. To ignore this error condition, type n. This error always invokes the CLEAR error condition in phase 4.

UNREF type I=inode-number OWNER=UID MODE=file-mode SIZE=file-size 
MTIME=modification-time (CLEAR)

Cause

Inode inode-number (whose type is directory or file) was not connected to a directory entry when the file system was traversed. The owner UID, mode file-mode, size file-size, and modification time modification-time of inode inode-number are displayed. When fsck is preening, the file is cleared if either its size or its link count is zero; otherwise, it is reconnected.

Action

To deallocate inode inode-number by zeroing its contents, type y at the CLEAR prompt. To ignore this error condition, type n.

ZERO LENGTH DIRECTORY I=inode-number OWNER=UID MODE=file-mode 
SIZE=file-size MTIME=modification-time(CLEAR)

Cause

A directory entry filename has a size file-size that is zero. The owner UID, mode file-mode, size file-size, modification time modification-time, and directory name filename are displayed.

Action

To deallocate the directory inode inode-number by zeroing out its contents, type y. To ignore the error condition, type n.

Phase 5: Check Cylinder Groups Messages

This section contains phase 5 fsck messages in the current Solaris release.

This phase checks the free-fragment and used-inode maps. It reports error conditions resulting from:

• Allocated inodes missing from used-inode maps

• Free fragments missing from free-fragment maps

• Free inodes in the used-inode maps

• Incorrect total free-fragment count

• Incorrect total used inode count

These messages (in alphabetical order) might occur in phase 5:

FRAG BITMAP WRONG (CORRECTED)

Cause

A cylinder group fragment map is missing some free fragments. During preening, fsck reconstructs the maps.

Action

To reconstruct the free-fragment map, type y at the SALVAGE prompt. To ignore this error condition, type n.

CG cg-number: BAD MAGIC NUMBER

Cause

The magic number of cylinder group cg-number is wrong. This error usually indicates that the cylinder group maps have been destroyed. When running interactively, the cylinder group is marked as needing reconstruction. fsck terminates if the file system is being preened.

Action

If this occurs, contact your local service provider or another qualified person.

CORRECT GLOBAL SUMMARY (SALVAGE)

Cause

The summary information is incorrect. When preening, fsck recomputes the summary information.

Action

To reconstruct the summary information, type y at the SALVAGE prompt. To ignore this error condition, type n.

Phase 5: Check Cylinder Groups Messages

This sections contains phase 5 fsck messages in the Solaris 10 initial 3/05 release.

This phase checks the free-block and used-inode maps. It reports error conditions resulting from:

• Allocated inodes missing from used-inode maps

• Free blocks missing from free-block maps

• Free inodes in the used-inode maps

• Incorrect total free-block count

• Incorrect total used inode count

These messages (in alphabetical order) might occur in phase 5:

BLK(S) MISSING IN BIT MAPS (SALVAGE)

Cause

A cylinder group block map is missing some free blocks. During preening, fsck reconstructs the maps.

Action

To reconstruct the free-block map, type y at the SALVAGE prompt. To ignore this error condition, type n.

CG character-for-command-option: BAD MAGIC NUMBER

Cause

The magic number of cylinder group character-for-command-option is wrong. This error usually indicates that the cylinder group maps have been destroyed. When running interactively, the cylinder group is marked as needing reconstruction. fsck terminates if the file system is being preened.

Action

If this occurs, contact your local service provider or another qualified person.

FREE BLK COUNT(S) WRONG IN SUPERBLK (SALVAGE)

Cause

The actual count of free blocks does not match the count of free blocks in the superblock of the file system. If the -o p option was specified, the free-block count in the superblock is fixed automatically.

Action

To reconstruct the superblock free-block information, type y at the SALVAGE prompt. To ignore this error condition, type n.

SUMMARY INFORMATION BAD (SALVAGE)

Cause

The summary information is incorrect. When preening, fsck recomputes the summary information.

Action

To reconstruct the summary information, type y at the SALVAGE prompt. To ignore this error condition, type n.

fsck Summary Messages

This section contains fsck summary messages in the current Solaris release. If you are not running at least the , Solaris 10 6/06 release, these messages are displayed in the cleanup phase. For more information, see Cleanup Phase Messages.

Once a file system has been checked, a few summary messages are displayed.

number-of files, number-of-files
used, number-of-files free (number-of frags, number-of blocks, 
percent fragmentation)

This message indicates that the file system checked contains number-of files using number-of fragment-sized blocks, and that there are number-of fragment-sized blocks free in the file system. The numbers in parentheses break the free count down into number-of free fragments, number-of free full-sized blocks, and the percent fragmentation.

***** FILE SYSTEM WAS MODIFIED *****

This message indicates that the file system was modified by fsck. There is no need to rerun fsck if you see this message. This message is just informational about fsck's corrective actions.

Cleanup Phase Messages

This section contains fsck cleanup phase messages in the Solaris 10 release. In this Solaris release, similar messages can be found in the fsck summary phase. See fsck Summary Messages for more information.

Once a file system has been checked, a few cleanup functions are performed. The cleanup phase displays the following status messages.

number-of files, number-of-files
used, number-of-files free (number-of frags, number-of blocks, 
percent fragmentation)

This message indicates that the file system checked contains number-of files using number-of fragment-sized blocks, and that there are number-of fragment-sized blocks free in the file system. The numbers in parentheses break the free count down into number-of free fragments, number-of free full-sized blocks, and the percent fragmentation.

***** FILE SYSTEM WAS MODIFIED *****

This message indicates that the file system was modified by fsck. If this file system is mounted or is the current root (/) file system, reboot. If the file system is mounted, you might need to unmount it and run fsck again; otherwise, the work done by fsck might be undone by the in-core copies of tables.

filename FILE SYSTEM STATE SET TO OKAY

This message indicates that file system filename was marked as stable. Use the fsck -m command to determine if the file system needs checking.

filename FILE SYSTEM STATE NOT SET TO OKAY

This message indicates that file system filename was not marked as stable. Use the fsck -m command to determine if the file system needs checking.