Purpose

Writes text to the blog file.

Syntax and Description

acfsutil [-h] blog

acfsutil blog {-t text | -u} mount_point

For information about running Oracle ACFS acfsutil commands, refer to About Using Oracle ACFS Command-Line Tools.

The following table contains the options available with the acfsutil blog command.

The acfsutil blog command enables you to write text to a blog file.

.

Examples

The following example illustrates how to run the acfsutil blog command. Running acfsutil blog with the —h option displays help.

Purpose

Collects internal Oracle ACFS state information for diagnosis by Oracle support.

Syntax and Description

acfsutil [-h] dumpstate 
acfsutil dumpstate {acfs_path | [-d] [-z] [acfs_path]}

acfsutil -h dumpstate displays help text and exits.

For information about running Oracle ACFS acfsutil commands, refer to About Using Oracle ACFS Command-Line Tools.

The following table contains the options available with the acfsutil dumpstate command.

The acfsutil dumpstate command collects internal Oracle ACFS state information for a specified file system. The state information is written to a binary incident file in a logging directory. The binary log incident file is specific to the file system mounted at the specified path. The acfs.dumpstats statistics file contains statistics for the entire Oracle ACFS kernel module.

Examples

The following example shows the use of the acfsutil dumpstate command.

Purpose

Display the trace entries for open files associated with the Oracle ACFS file system specified by the mount point.

Syntax and Description

acfsutil info ftrace -h
acfsutil info ftrace [-s] mount_point

acfsutil info ftrace —h displays help text and exits.

For information about running Oracle ACFS acfsutil commands, refer to About Using Oracle ACFS Command-Line Tools.

The following table contains the options available with the acfsutil info ftrace command.

The acfsutil info ftrace command displays a list of open files on a mounted Oracle ACFS file system.

The Oracle ACFS kernel driver keeps track of which files are loaded in memory. These files may not have an active open and may just be cached. The open file tracing is able to determine which of those cached File Control Blocks (FCBs) have an active open or reference. The intent of this command is to enable you to determine if, and which, files are still being referenced that could prevent a file system unmount from occuring.

When acfsutil info ftrace initially runs, the command attempts to purge any cached files that are no longer referenced. This operation may require some time to complete because the modified metadata and user data for each file has to be flushed to disk.

The following describes the output of the acfsutil info ftrace command. Note that a file can refer to a regular file or directory.

The basic format of the output is:

Fileid: %ID%, Pathname: %PATH%
          [%OP%] Pid:  %PID% Ppid:   %PPID% Elapsed time: %TIME% Cmd: %CMD%
          ...

The fields are described in the following list.

• %ID%: The numeric file identifier. This is the same number that is used with acfsutil info id. This value is also known as the inode number on Linux.

• %PATH%: The generated pathname for the file based on the %ID%. N/A may be displayed if it is not available.

• %OP%: The type of operation that accessed the file. The values may be the following:

  • LOOKUP: The specified process looked up this file via the pathname.

  • CREATE: The specified process created the file.

  • NFS: A NFS process has accessed the file on behalf of client.

  • OPEN: The specified process opened the file.

  • MAP: The specified process mapped the file into memory.

• %PID%: The process id %PPID%: The parent process id. This output item may not be available.

• %TIME%: The elapsed time from when the operation occurred. The format is: d (days), h (hours), m (minutes), s (seconds)

• %CMD%: The name of the process that performed the operation.

Each file listed may have more than one operation listed depending on the system workload. The amount of operations displayed is limited to conserve memory. The Oracle ACFS driver keeps a rotating log for each operation and the operation entries may wrap. As a result, the oldest operation may not be the first one displayed.

Examples

The following example shows the acfsutil info ftrace command run on the /mnt mount point.

Purpose

Displays lock contention statistics.

Syntax and Description

acfsutil lockstats lh -h
acfsutil lockstats lh [-b] [-e] [-z] [-t top_n] [-s sort_column]

acfsutil lockstats lh —h displays help text and exits.

For information about running Oracle ACFS acfsutil commands, refer to About Using Oracle ACFS Command-Line Tools.

The following table contains the options available with the acfsutil lockstats command.

Examples

The following example shows multiple ways to use of the acfsutil lockstats command.

Purpose

Retrieves memory diagnostic log files and manages debug settings.

Syntax and Description

acfsutil [-h] log

acfsutil log [-f filename] [-s] [-r n{K|M|G|T|P}] [-p {avd|ofs|oks}] [-l debuglevel] 
         [-n consolelevel] [-o wait_time] [-q] [-c debugcontext] [-T file_type] 
         [-m mount_point] [-a] [-C] [-t]

For information about running Oracle ACFS acfsutil commands, refer to About Using Oracle ACFS Command-Line Tools.

The following table contains the options available with the acfsutil log command.

The acfsutil log command enables you to manage memory diagnostic log files. With none of the options specified, the acfsutil log command retrieves and writes the ./oks.log memory log by default.

The -o option performs the following:

1. Sets the log size to 500M , the log level to 5, and the product to ofs (acfs) for the in-memory log on all nodes

2. Displays an informational message, such as Blocking for 180 seconds, reproduce problem now

3. After waiting for the specified number of seconds, then displays Dumping log on all nodes

4. Initiates a clusterwide dump of logs

5. Resets the log level to 2 and resets the log size to the default

The -o option can be combined with the -p, -l, and -r options if the default product, debug level, or log size settings should be changed.

You must be the root user or an Oracle ASM administrator user to run this command.

Examples

The following example shows various ways to run the acfsutil log command.

Purpose

Copies metadata from an Oracle ACFS file system into a separate output file.

Syntax and Description

acfsutil meta -h
acfsutil meta  [-v] 
               [-g] 
               [-g -O -C -S]
               [-O]
               [-C COW_filepath]
               [-S COW_size]
               [-q nn[K|M|G|T]]  
               [-l log_file_path] 
               [-o acfs_extent_offsets]      
               {-f record_oriented_metadata_output_file} [-a accel_device] volume_device

acfsutil meta  {-e record_oriented_metadata_input_file [-i]}
               {-f output_filesystem_meta_file_prefix_name}

acfsutil meta -h displays help text and exits.

For information about running Oracle ACFS acfsutil commands, refer to About Using Oracle ACFS Command-Line Tools.

The following table contains the options available with the acfsutil meta command. The options are shown in Linux format (also AIX and Solaris format). Commands run on Windows use a slash (/) instead of a hyphen (-) in front of an option. For example, use —f on Linux and /f on Windows.

The acfsutil meta command operates as a metadata collector to partially copy an Oracle ACFS file system into a separate specified record-oriented output file. The metadata collector reads the contents of the file system specified by the volume device name of an Oracle ACFS file system. This input file system is searched for Oracle ACFS metadata and then all metadata found is written into the specified record-oriented output file. The generated record-oriented output file can be easily transferred to a another system, where it can expanded for diagnostics and analysis, without impact to the original file system at the customer site.

The -g option collects only the known good metadata. The -g option should not be used with a corrupted file system because the -g option does not find lost metadata. Any lost metadata may be important in diagnosing a file system corruption. If the file system is in good repair, the -g option may collect the metadata much faster because it does not need to scan the entire physical volume looking for lost metadata blocks.

When the acfsutil meta -g command is run on Linux, the Oracle ACFS online checker (fsck) runs automatically. The Oracle ACFS online checker, running on behalf of acfsutil meta -g, transverses the Oracle ACFS file system metadata using the metadata on-disk pointers, and writes metadata that has been read into the acfsutil meta-g metadata collection file. For information about the online fsck command on Linux, refer to Oracle ACFS Command-Line Tools for Linux Environments.

To obtain the best copy of the file system with acfsutil meta, unmount the file system before running acfsutil meta. If it is not possible to unmount the file system, avoid modifying the contents or performing a volume resizing operation while acfsutil meta is running.

If the original file system is very large, then the output file can also be very large. Compress the output file when possible to reduce storage space and transmission time.

If the file system has an accelerator device associated with it, acfsutil meta also copies the accelerator device data into the record-oriented output file. This operation occurs automatically.

In most circumstances, acfsutil meta automatically copies the accelerator device into the record-oriented output file. However, if you have think that the meta collector is not able to find the accelerator device on its own, you can specify the name on the command line with the -a option. For example, this situation could occur if the file system is corrupt. Note that using the -a option overrides how the meta collector operates automatically, so -a should be used carefully.

The output file should not be placed on the Oracle ACFS device that is specified as the input device because the metadata command might process the output file also. The output file should be placed on a file system that can support an output file which is the size of the Oracle ACFS input volume device. The output file should not need all that storage unless the file system is full and contains all metadata and almost no user data, which is unlikely, but not impossible.

The -q flag should be used with caution. When -q is specified, the meta collector does not scan and copy the entire input file system. Instead, it only scans and copies a predetermined number of bytes and certain data structures which are considered important. The primary use for the -q flag is for situations where there is not sufficient time to run the full version of the metadata collector. The -q flag should not be used unless it is recommended by the support personnel investigating the problem.

Expanding the record-oriented output file should be performed on the system where diagnosis and analyzes is to be performed. For example, the following command expands record-oriented metadata file on another file system that has adequate storage space.

acfsutil meta -e record_oriented_metadata_input_file -f output_filesystem_meta_file_prefix_name

The output of the command provides sparse files suitable for use with fsck or acfschkdsk. If the record oriented metadata input file includes an accelerator volume, a second sparse output file is created using the same output file name prefix with .acc suffix appended. The file system used for the expanded files should support sparse files. Otherwise, the resulting expanded files could be extremely large containing useless zeros where sparse holes could be saving space.

The acfsutil meta expanded output file can be read by the fsck command in most cases. However, the Oracle ACFS specific fsck command on some OS platforms might not access the output file correctly or might not work with a specified flag. You can use a slightly modified fsck command form in these cases. For example:

• On Linux, run the command in this format if you are using the —x flag:

  /sbin/fsck.acfs -x filesystem_meta_file.acc filesystem_meta_file

• On Solaris, run the command in this format if you are using the —o x flag:

  /usr/lib/fs/acfs/fsck -o x=filesystem_meta_file.acc filesystem_meta_file

• On AIX, run the command in this format:

  /sbin/helpers/acfs/fsck filesystem_meta_file

• On Windows, run the command in this format:

  acfschkdsk filesystem_meta_file

Examples

Example 17-14 shows the use of the acfsutil meta command to copy and expand metadata into output files.

Purpose

Manages Oracle ACFS persistent logging configuration settings.

Syntax and Description

acfsutil plogconfig [-h] [-d persistent_log_directory] [-t] [-q ] [-i seconds] 
                    [-s buffer_size] [-l low_water_percent] [-u high_water_percent] 
                    [-m max_logfile_size] [-n max_logfile_number]

acfsutil -h plogconfig displays help and exits.

For information about running Oracle ACFS acfsutil commands, refer to About Using Oracle ACFS Command-Line Tools.

The following table contains the options available with the acfsutil plogconfig command.

The acfsutil plogconfig command provides a diagnostic tool to manage configuration settings for persistent logging.

All command arguments are optional, but at least one argument must be specified.

You must be the root user or an Oracle ASM administrator user to run this command.

Examples

The following example illustrates the use of the acfsutil plogconfig command to display the current configuration settings.