acf(1M) acf(1M)
NAME
acf - array configuration
SYNOPSIS
acf -i -z hw-addr
acf -c -z hw-addr -canned-config-key-letter
acf -l -z hw-addr [-f file]
acf -u -z hw-addr logicalunitconfigurationoptions
acf -e -z hw-addr errorrecoveryconfigurationoptions
acf -d -z hw-addr arraydriveconfigurationoptions
acf -f -z hw-addr
acf -R -z hw-addr
DESCRIPTION
acf is used to perform initial configuration of the disk array when
installing a disk array on a system. It is also used to reconfigure
the disk array whenever reconfiguration is required (e.g., addition of
more disk drives, redefinition of LUNs, RAID levels, etc.). The first
command option is taken as a function key letter, identifying the
desired function:
-i Display device inquiry information
-c Execute a "canned" configuration sequence
-l Download array controller firmware
-u Set or display array logical unit parameters
-e Set or display array error recovery parameters
-d Set or display array drive parameters
-f Format a logical unit of a disk array
-R Establish the configuration of the array controller or the speci-
fied logical drive again.
Page 1 Reliant UNIX 5.44 Printed 11/98
acf(1M) acf(1M)
Option -z:
-z The argument hw-addr is a string of characters giving the
hardware address of the "object" to which the command is directed
- this can be a controller, a logical unit, or a physical drive.
hw-addr has exactly one of the following forms:
iosb/sraidppu
iosb/sraidppull
iosb/sraidppu:bp
The first form specifies a controller, the second designates a logical
unit, and the third one refers to a physical drive.
The naming conventions are as follows:
b RM400: 0 (on-board iobus) or 1 (peripheral bus)
RM600: always 0
pp RM400: the SCSI Host Adapter port number
RM600: the logical controller number
u the device's unit number
l the logical unit name; this may range from 0 to 7
: separator between array controller name and backend drive parts
of hw-addr
After the separator, the naming conventions are as follows:
b specifies the SCSI bus of the array controller
p specifies the target ID of the array controller
Some examples of valid hardware address notation:
ios0/sraid041
system bus 0, host controller 04, unit number 1
ios0/sraid041l0
system bus 0, host controller 04, unit number 1, LUN 0
ios0/sraid041:15
system bus 0, host controller 04, unit number 1, physical drive
on SCSI bus 1 and target id 5 of the array controller
Page 2 Reliant UNIX 5.44 Printed 11/98
acf(1M) acf(1M)
OPTIONS
Options for canned configuration invocation:
-canned-config-key-letter
canned-config-key-letter is a letter that identifies an entry
in a file of "canned" array controller configurations
(/etc/default/acf). acf reads the file of canned configura-
tions, locates the entry identified by canned-config-key-letter
and invokes an associated sequence of configuration commands,
passing hw-addr as an input to the sequence. (The FILES section
below gives details on the organization of the /etc/default/acf
file).
Options for array controller download:
-f file
Name of the file containing the disk array controller firmware,
or else a list of files, each of which contains firmware to be
downloaded. The object files must be in Motorola S-record for-
mat. If this option is not present, the controller firmware
file is read from standard input.
Options for logical unit configuration:
(For all options listed below, a "d" instead of a value following the
option letter will cause the current setting of the option to be
displayed to standard out in the format "nameofparameter =
valueofparameter".)
-B d Number of blocks in the LUN indicated by hw-addr (display only,
unchangeable).
-b n Set the logical block size for the LUN indicated by hw-addr to
n.
-U {a | x | r | d}
If "a" is present, the logical unit indicated by hw-addr is
added; if "x" is present, the logical unit is deleted; if "r"
is present, the status of the logical unit is revived; if "d"
is present, the status of the logical unit is displayed.
-d drive-list
The drives in the drive-list are associated with the logical
unit indicated by hw-addr. The drive list consists of drive
identifiers (following the notation of the -z option), sepa-
rated by commas. A drive identifier uses only the SCSI bus and
target ID components of the hardware address to identify a
drive in the array. (An example of a drive list representing
the first rank of drives is: 10,20,30,40,50.)
Page 3 Reliant UNIX 5.44 Printed 11/98
acf(1M) acf(1M)
Two groups of drive (disk ranks) must be specified for RAID-
Level 1, where "ff" is used to separate the groups, e. g.
-d11,21,31,ff,12,22,32. RAID-Level 1 is set implicitly with
this format!
-r n Set the RAID level for the LUN indicated by hw-addr to n.
RAID-Level 1 is already set with the -d option (see above), if
the corresponding drive list format is the default!
-S n Set the segment size to n for the LUN indicated by hw-addr.
-Z n Set the segment zero size to n for the LUN indicated by
hw-addr.
-s d Sector size for the LUN indicated by hw-addr (display only,
unchangeable).
-A {on|off}
This option controls disabling of automatic drive-replacement-
detection for the LUN indicated by hw-addr. on will disable
automatic drive-replacement-detection, and off will enable it.
When enabled (i.e., set to off), the controller will automati-
cally detect a drive replacement and automatically begin recon-
structing the drive.
-F freq
This option is the reconstruction frequency, or the number of
tenths of a second to delay between instances of reconstruction
activity.
-a amount
This option is the reconstruction amount, or the number of
blocks to reconstruct once reconstruction activity is ini-
tiated.
-t {numeric-mode-page | symbolic-mode-page}
This option controls the type of mode pages that the array con-
troller will sense when issuing a MODE SENSE command to the
array drives. If the value following the -t option is a number,
it is interpreted as a hexadecimal value; otherwise it is a
symbolic page type. Choices for symbolic page types are:
default sense default pages
current sense current pages
saved sense saved pages
Page 4 Reliant UNIX 5.44 Printed 11/98
acf(1M) acf(1M)
-M {numeric-status | symbolic-status}
This option switches the specified controller to an active role
and the alternative controller to a passive role or gives both
controllers an active status.
If a number is entered after -M, it is interpreted as a hexade-
cimal status value.
Alternatively, a symbolic status can be specified:
active Switches the specified controllers to an active
status.
dual Switches both controllers to an active status.
If the "dual" status is activated, you have to specify a list
of the logical drives, which are to be assigned to the speci-
fied controller. All other logical drives are assigned to the
alternative controller. If the controllers are already in
"dual" status, the specified logical drives are also assigned
to this controller the next time this option is invoked.
The drive list can have the following format, for example:
-Mdual:1,3,6 or -M2:1,3,6
-m {numeric-status | symbolic-status}
This option controls the redundant controller operational mode.
If the value following the -m option is a number, it is inter-
preted as the hexadecimal value for the status; otherwise it is
a symbolic status. Choices for the symbolic status and associ-
ated meanings are:
release release failed controller from reset condition
fail fail alternate controller
Note: the order of the above options can be important in some
cases as all changes to the configuration are done exactly in
the order in which the options are given.
-h {d | a## | x##}
This option controls the configuration of hot spare drives:
-hd Displays the list of configured hot spare drives.
-ha## Defines this drive (##, same format as with -d) as the
hot spare drive.
-hx## Marks the original hot spare drive as "free".
Page 5 Reliant UNIX 5.44 Printed 11/98
acf(1M) acf(1M)
-w {d | a | x}
This option controls the setting of the "write cache" mode; it
is valid for the specified logical drive.
-wd Displays the current setting (switched on/off).
-wa Switches on "write cache" mode.
-wx Switches off "write cache" mode.
-i {d | a | x}
This option controls the setting of the cache mirroring mode;
it is valid for the specified logical drive.
-id Displays the current setting (switched on/off).
-ia Switches on "cache mirroring" mode. Note: This mode is
only effective if the write cache is also switched on.
-ix Switches off "cache mirroring" mode.
Options for array error recovery configuration:
(For all options listed below, a d instead of a value following the
option letter will cause the current setting of the option to be
displayed to standard out in the format "nameofparameter =
valueofparameter".)
-r n Set the read retry count to n for the controller indicated by
hw-addr.
-w n Set the write retry count to n for the controller indicated by
hw-addr.
-W {on|off}
Enable (on) or disable (off) automatic write reallocation for
the controller indicated by hw-addr (display only, unchange-
able).
-R {on|off}
Enable (on) or disable (off) automatic read reallocation for
the controller indicated by hw-addr (display only, unchange-
able).
-P {on|off}
Enable (on) or disable (off) reporting of recovered errors for
the controller indicated by hw-addr.
Page 6 Reliant UNIX 5.44 Printed 11/98
acf(1M) acf(1M)
Options for drive configuration:
(For all options listed below, a d instead of a value following the
option letter will cause the current setting of the option to be
displayed to standard out in the format "nameofparameter =
valueofparameter".)
-c {numeric-status | symbolic-status}
Set the drive status. If the value following the -c option is a
number, it is interpreted as the hexadecimal value for the
status; otherwise it is a symbolic status. Choices for the sym-
bolic status and associated meanings are:
add add drive
delete mark drive as nonexistent
fail mark drive as failed
replace mark drive as replaced and begin reconstruction
replacenf mark drive as replaced and begin reconstruction
(without forcing a drive format)
Note: if replace is used, the drive is always formatted; if
replacenf (no format) is used, the drive is formatted only if
the controller determines that a format is necessary.
FILES
The file /etc/default/acf is an ASCII file containing a series of
identically-formed entries. Each entry associates three items of
information: (1) a unique key letter for the entry (used to select a
particular entry on the acf command line), (2) a text string describ-
ing the particular disk array configuration this entry corresponds to,
and (3) the full pathname for a command that, when executed, will
carry out the steps needed to set up the configuration. This command
will typically be a (Bourne) shell script and must follow the conven-
tion of being able to accept exactly one argument, the hardware
address in a form that matches the hw-addr argument to acf. A colon is
used to delimit fields within an entry, as in the following example:
a:RAID level 0, 2 drives:/etc/setupR0-2
b:RAID level 5, 3 drives:/etc/setupR5-3
c:RAID level 0, 5 drives:/etc/setupR0-5
d:RAID level 5, 5 drives:/etc/setupR5-5
etc.
Each entry is contained on one line and may not span multiple lines.
Page 7 Reliant UNIX 5.44 Printed 11/98
acf(1M) acf(1M)
EXIT STATUS
0 = Error-free termination
1 = Command line syntax error
2 = Argument not in acceptable range
3 = Error opening a file or device
4 = SCSI-bus or SCSI-device-related error condition
5 = Non-specific error
6 = Insufficient memory
NOTES
In some cases, acf will require exclusive ownership of the hardware
modules affected by the operations being performed, meaning, for exam-
ple, that a logical unit of the array could not be accessed as a file
system at the same time that its configuration was being fundamentally
altered. acf does try to be as liberal as possible in terms of permit-
ting those configuration actions that can be performed safely without
exclusive ownership. If the program cannot acquire exclusive ownership
when needed, it displays a notice to that effect and terminates.
acf -l will not work correctly if it encounters a Motorola S-record
file having multiple end records.
Invoking acf with an unrecognized function keyword, as in
acf ?
will cause usage information to be displayed.
A current limitation of acf is that it is unable to configure a RAID 3
logical unit.
The ability to submit a list of files to be downloaded is not a sup-
ported feature.
With release 3.0 of the array controller software, the add argument to
the -c option letter under drive configuration can be used to restore
a drive in a failed or warning state back to the optimal state without
reconstructing the drive. This function is intended for careful usage
in multiple-drive-down situations and can result in loss of data and
inconsistent array parity if the data on the logical unit has changed
since the drive in question was placed in the failed or warning state.
After adding a LUN, a dksetup -R must be executed (unless the system
is rebooted).
Page 8 Reliant UNIX 5.44 Printed 11/98
acf(1M) acf(1M)
Switching the passive controller into the active role may take some
time.
SEE ALSO
apc(1M), apr(1M), arc(1M), can(1M), lad(1M).
Page 9 Reliant UNIX 5.44 Printed 11/98