Chapter 1

Kernel Configuration

The config(1M) commands are provided for constructing the system. The SUPER-UX config command creates an information file that coincides with your system and reconstructs the kernel.

In accordance to the config command, the kernel can be optimized.

This chapter explains how to construct the system.

1.1 SYSTEM CONSTRUCTING IN THE BASIC CONFIGURATION

System constructing requires system constants that determine the system configuration. Initially, system constants have default values which can be used to construct a standard configuration system. Using default values creates a basic configuration kernel.

First, construct a basic configuration system not including the hardware. For the hardware configuration (number of processors or number of channels) you must create a kernel according to the hardware configuration of your site.

The config command reads the hardware configuration data when creating a kernel, and automatically reflects it on the kernel. You can create a kernel that reflects the hardware configuration of your system site by executing the kernel creation function of the config command.

For information on using the config command, refer to Section 1.3.

Home

1.2 SYSTEM CONSTRUCTING IN THE NEW CONFIGURATION

1.2.1 Changing the System Configuration

When the basic configuration kernel does not satisfy the needs of your site or you need to create a system of your own, you must change the system configuration. To do so, use the config command as follows:

For information on how to use the config command, refer to Section 1.3.

1.2.2 New Kernel Startup

1.2.2.1 KERNEL REPLACEMENT

To start up the created kernel, execute the shutdown command (see the SUPER-UX Operation Guide) and change to single-user mode.

Next, copy the created kernel into the root directory (the created kernel has a different name from the presently activated kernel).

The system restart is done after repeatedly executing the sync command and rewriting the superblock. In accordance with the new kernel name specification corresponding to the Initial System Loader (ISL), the new kernel is loaded. When it is confirmed that there are no problems in activating the new kernel and if the name given is "unix," there will be no need to specify the kernel name in future startups. It is recommended that you save the previous "unix" file with a different name.

1.2.2.2 PROCESS WITHOUT SYSTEM STARTUP

If a new kernel startup fails, terminate the system, restart it, and specify the previous kernel name corresponding to ISL.

Home

1.3 config COMMAND OPERATION

1.3.1 Execution Methods Of The config Command

The config command is executed in super-user mode. The format is as follows.

/usr/sbin/config [-m mfile] [-d devfile] [-o ofile] [-k kfile] [dfile]

When the config command is executed, it reads into the specified mfile, devfile, and dfile. When the mfile is not specified, the default value becomes /etc/master.d/master. The devfile contains hardware configuration data. When devfile is not specified, the command reads hardware configuration data from the default file (/etc/master.d/sxsginfo.siox).

The default file contains the hardware configuration data on each user site. To create a hardware configuration that is different from the actual configuration, specify the file that contains configuration data for it. There is no default value for the other options. If there is a mistake in the descriptions of the file being read, the config command generates an error message and asks the user for a process. When the terminal cannot produce the screen, system structure is modified using only dfile.

-m mfile The specified mfile is the master file applicable for this kernel configuration. When mfile is omitted, /etc/master.d/master is taken. For details on the master file, refer to the SUPER-UX Programmer's Reference Manual. -d devfile Hardware configuration data file name. When devfile is omitted, /etc/master.d/sxsginfo.siox is taken. For details on the sxsginfo.siox, refer to the Installation Guide. -o ofile Master file name. -k kfile The file name specified in kfile is for kernel creation.

ofile and kfile can be input later on the config command executed. If the ofile is omitted, it cannot create a master file. If kfile is omitted, it cannot create a kernel.

1.3.2 Terminals that Can Produce a Screen

If the config command executed, the terminal displays the following menu. 

CONFIG COMMAND MAIN MENU
       ************************************************
       * 1. Display/Change tunable parameters         *
       *                                              *
       * 2. Display/Change system identification      *
       *                                              *
       * 3. Write out configuration files             *
       *                                              *
       * 4. Make kernel                               *
       *                                              *
       * 5. Exit this program                         *
       ************************************************
                               Enter Operation No. [  ]

The config Command Main Menu choices are defined as follows:

You can display and modify the contents of every item interactively with config. To overwrite a numeric value, use the following operations:

Home

The following sections explain the process items.

1.3.2.1 CONSTANTS DISPLAY/CHANGE

If you select number 1 from the config Command Main Menu, the following screen appears. 

     ***** 1. Display/Change tunable parameters **********
     *                                                    *
     * 1. Process Management                              *
     * 2. Memory Management                               *
     * 3. File Management                                 *
     * 4. Network Management                              *
     * 5. Account Management                              *
     * 6. Detail Tuning                                   *
     * 7. Exit this menu                                  *
     ******************************************************
                                   Enter Operation No. [   ]

When you select 1 through 5, one of the following displays appears. 

*** 1-1. Display/Change tunable parameters (Process Management) (1/1) ***

         name         value  default     range                memo
     -----------------------------------------------------------------------
       1 NPROC       [1024]    1024    128 - 8192             Max. number>
       2 MAXUP        [ 50]      50     20 - 2048             Max. number>
       3 NGROUPS_MAX   [16]      16      0 - 32               Maximum num>
       4 SLAVEPRI     [120]     120      0 - 0x7fffffff       family sche>
       5 CHKPOINT       [0]       0      0 - 1                Use CHKPOIN>
       6 CPUUSER        [0]       0      0 - 1                Determine C>
     ------------------------------------------------------------------------
     exit(out):CTL-O exit:CTL-W abort:DEL undo:CTL-U memo:CTL-G

Home 

*** 1-2. Display/Change tunable parameters (Memory Management) (1/1) ***

        name        value      default     range               memo
    ------------------------------------------------------------------------
      1 SHRETEXT        [1]         1      0 - 1               Flag for sh>
      2 SPMEM         [ 64]        64     10 - 512             Number of 1>
      3 MAXUMEM    [262144]    262144    160 - 16777216        Upper limit>
      4 MAXUMEM32    [1024]      1024    224 - 16384           Upper limit>
    -------------------------------------------------------------------------
    
exit(out):CTL-O exit:CTL-W abort:DEL undo:CTL-U memo:CTL-G 

*** 1-3. Display/Change tunable parameters (File Management) (1/1) ***

        name        value     default       range                memo
  ------------------------------------------------------------------------------
    1 NSBUF          [1000]      1000     100 - 9000             Number of s>
    2 SFSHCNT          [ 0]         0       0 - 500              Number of S>
    3 ULIMIT   [ 0x1000000] 0x1000000  0x2000 - 0x3fffffffffffff System defa>
    4 NINODE        [ 1200]      1200     100 - 10000            Number of i>
    5 NFILE         [ 1200]      1200     100 - 10000            Number of f>
    6 NMOUNT          [100]       100      10 - 500              Number of f>
    7 FLCKCREC        [100]       100      10 - 10000            Number of f>
    8 BDFLUSHR        [ 30]        30       1 - 120              Syncdaemon >
    9 NAIOREQ         [ 64]        64       1 - 256              Number of a>
   10 NTMPDATA    [ 131072]    131072       0 - 1048576          Data size o>
   11 RSTCHOWN          [0]         0       0 - 1                Enable to r>
   12 NQUOTA         [ 256]       256     128 - 60000            Number of q>
   13 NDQUOT        [ 6400]      6400     256 - 120000           Number of d>
   14 NQUOTAG        [ 128]       128      32 - 32767            Number of q>
   15 NDQUOTG       [ 2048]      2048      64 - 65535            Number of d>
   16 AMCACHE        [1024]      1024    1024 - 8192             MM cache si>
   17 CLSTH             [0]         0       0 - 1                Flag which >
   18 VVTH           [ 128]       128       0 - 1024             Reallocatio>
  ------------------------------------------------------------------------------
    exit(out):CTL-O exit:CTL-W abort:DEL undo:CTL-U memo:CTL-G 

*** 1-4. Display/Change tunable parameters (Network Management) (1/1) ***

           name        value   default    range        memo
    ---------------------------------------------------------------
      1 LHIPMAX       [ 1]      0     0 - 32        Max. number>
      2 PTYMAX       [160]    160     1 - 160       Max. number>
      3 IPFORWARDING   [1]      1     0 - 1         Receive pac>
      4 NSTREAM      [320]    320    60 - 1500      Number of s>
      5 NBUF         [500]    500    50 - 4500      Number of s>
      6 NFS_RNUM     [ 50]     50    50 - 1300      Number of N>
    ---------------------------------------------------------------
                       
      exit(out):CTL-O exit:CTL-W abort:DEL undo:CTL-U memo:CTL-G

If the system includes the program product MLS, PYTMAX is changed as follows. 

          name       value    default   range      memo

           2 PTYMAX     [32]      32          1 - 32   Max. number>

          name       value    default   range      memo

           2 PTYMAX     [64]      64          1 - 64   Max. number>
 
         name       value    default    range      memo

           2 PTYMAX     [128]    128          1 - 128  Max. number>

Home 

*** 1-5.Display/Change tunable parameters (Account Management) (1/1) ***

      name        value   default    range           memo
  ------------------------------------------------------------------------
    1 ACCTFSIZE   [ 100]    100     20 - 0x7fffffff  Max. size o>
    2 ACCTPSARGS     [0]      0      0 - 1          Switch to e>
    3 CPURESTM       [0]      0      0 - 1          Switch to e>
  ------------------------------------------------------------------------
  exit(out):CTL-O exit:CTL-W abort:DEL undo:CTL-U memo:CTL-G

The value column shows the present values of corresponding system constants. You can modify these values by moving the cursor in this column and overwriting the numerical values.

The default column shows the default value during system release.

The range column shows the range of specifiable values in the system constants. When changing the values, the new values specified must fall within this range. If the values specified fall out of this range, config outputs an error message.

The memo column shows the description of each system constant. To display the description to the end, press Ctrl-G. The description is displayed at the bottom.

Pressing Ctrl-O is equivalent to selecting number 3 from the config Command Main Menu.

For number 6 from the config Command Submenu, refer to Chapter 2.

If you select number 7 from the config Command Submenu, you return to the config Command Main Menu.

Home
1.3.2.2 NAMES DISPLAY/CHANGE

If you select number 2 from the config Command Main Menu, the current system name is displayed. The values shown next are used during the system release. 


    ***2. Display/Change system identification                         (1/1)***
    ----------------------------------------------------------------------------
       System Name     ............ SUPER-UX
       System Nodename ............  [ unix]
       System Release  ............    10.1 
       System Version  ............
       Machine Name    ............    SX-5






    ----------------------------------------------------------------------------
    exit(out):CTL-O exit:CTL-W abort:DEL undo:CTL-U

Only the system node name can be changed.

Pressing Ctrl-O is equivalent to selecting number 3 from the config Command Main Menu.

Home
1.3.2.3 MASTER FILE OUTPUT

If you select number 3 from the config Command Main Menu, the master file required in the creation of kernel is output. The master file is output to the file specified by ofile when starting the config command. If ofile is not specified, config asks for a new master file name.

1.3.2.4 CREATE NEW KERNEL

If you select number 4 from the config Command Main Menu, a kernel reflecting the specified structural information is created. The kernel is created by the specified name in kfile when starting the config command under /etc/master. If kfile is not specified, config asks for a new kernel name.

1.3.2.5 PROGRAM TERMINATION

If you select number 5 from the config Command Main Menu, the config command terminates.

After modifications are done in the config command, if this number is chosen without having selected 3 or 4 first, the following message appears:

If you select y, the master file output is performed and the program terminates.

If the structural information is modified by the config command, output the file and create a kernel before terminating the program.

1.3.3 Terminals that Cannot Produce a Screen

The config command cannot be used on a terminal that cannot display a screen. Terminals that cannot display a screen fit one or more of the following criteria:

In the above cases, the function is selected from the following simplified menu. However, the system configuration information can be changed only when the dfile option is specified to the config command. 

    ***  CONFIG COMMAND MAIN MENU ***

         1. Write out configuration files
         2. Make kernel
         3. Exit this program
      Enter Operation No. >

Select number 1 to reflect the value appointed to config and write the system configuration information. Select number 2 to reflect the value appointed to config and form the newly structured kernel. Select number 3 to exit the config command. These functions are the same as those described in Sections 1.3.2.3 through 1.3.2.5.

Home

The dfile has the following features:

The dfile structure rules are as follows:

Because information on dfile is not grouped, information classification is done via checks on the name found in field 1 of each line. After the classification of information, the corresponding data contents are changed to the specified contents.

The following shows the dfile format:

Home

1.4 SYSTEM CONSTANTS

This section describes the system constants.

1.4.1 Process Management

To optimize the kernel, use Process Management functions and modify the following system constants.

Table 1-1 Optimizing with Process Management Functions

Name Default Minimum Maximum Explanation
NPROC 1024 128 8192 Maximum number of processes in the system. In NPROC, the number of system processes and various types of daemon processes are added together. The number of system processes and the number of daemons when activating in the multi-user mode are as follows:

Number of processes: 7 + number of processes
Number of daemons: Approx. 20

Consider the usage structure (such as actual memory storage, network environment, and system operation form, etc.) and estimate a suitable value according to the above values.
MAXUP 50 20 2048 Maximum number of processes corresponding to one user. Consider the usage structure (such as the number of users executing at the same time, the number specified to NPROC, etc.) and estimate a suitable value.

Home

Table 1-1 Optimizing with Process Management Functions (cont'd)

Name Default Minimum Maximum Explanation
NGROUPS_MAX 16 0 32 Maximum number of supplementary group IDs. Estimate a suitable value for the site.
SLAVEPRI 120 0 0x7fffffff SLAVEPRI sets the slave priority for interactive processes and is a difference from the current priority. By setting SLAVEPRI, tasks in the microtasking group execute at the same time more easily. A more positive value of SLAVEPRI results in higher family scheduling priority. Estimate a suitable value for the system site.
CHKPOINT 0 0 1 Flag that indicates whether or not to use the checkpoint/restart function. Set CHECKPOINT to 1 to use this function.
CPUUSER 0 0 1 CPUUSER is a flag of CPU limitation methods. When the CPU limit function is based on the total of user CPU time and system CPU time, indicate 0. When the CPU limit function is based on user CPU time alone, indicate 1.

1.4.2 Memory Management

The kernel can be optimized for Memory Management by modifications to the following system constants.

Table 1-2 Optimizing with Memory Management Functions

Name Default Minimum Maximum Explanation
SPMEM 64 10 512 SPMEM is the size that can be allocated for small-pages. Set the number of large-pages to SPMEM because small-pages are managed by dividing large-pages. Estimate a suitable value for the site (such as total number of processes and actual memory storage.) For details, refer to Section 3.4.1.
MAXUMEM 262144 160 16777216 MAXUMEM is the upper limit on a large-page process size (in 32KB units.) Estimate a suitable value by considering the usage structure of the system site. For details, refer to Section 3.4.2.

Home

Table 1-2 Optimizing with Memory Management Functions (cont'd)

Name Default Minimum Maximum Explanation
MAXUMEM32 1024 224 16384 MAXUMEM32 is the upper limit on a small-page process size (in 32KB units.) Estimate a suitable value by considering the usage structure of the system site. For details, refer to Section 3.4.2.
SHARETEXT 1 0 1 SHARETEXT is the flag used to specify whether or not to share the text segment of large-page processes. To share the text segment, indicate 1. Otherwise, indicate 0. SHARETEXT cannot be changed during operation. For details, refer to Section 3.4.1.

1.4.3 File Management

The kernel can be optimized for the File Management function, depending on the modifications made to the following system constants.

Table 1-3 Optimizing with File Management Functions

Name Default Min. Max. Explanation
NSBUF 1000 100 10000 NSBUF is the number of system buffers used to access a file on the SFS file system. As the system buffers increase in number, it is possible to speed up the input/output process since the data to be accessed is likely in the system buffers.When the beginning/end of data for performing input/output is not on the 4KB boundary of the file, the input/output of the beginning/end of block is carried out via the system buffer. The number of system buffers is meaningless when transmitting only bufferless data.

The sar(1) command allows you to check the cache hit ratio, the number of data transfers between the buffer and the disk, and the number of accesses. When the cache hit ratio is extremely low because of too many access requests, NSBUF tuning is necessary because the buffer may be out of space. For more information on the sar command, refer to the SUPER-UX User's Reference Manual.

Home

Table 1-3 Optimizing with File Management Functions (cont'd)

Name Default Minimum Maximum Explanation
NINODE 1200 100 10000 NINODE is the maximum number of files that can be simultaneously opened in the system. There is no problem when there are more files in the system than you need. The value must be larger than the maximum number of processes in the system (NPROC). A higher NINODE value allows more data to reside in memory, thus increasing the input/output speed. When you receive the following message frequently, you should increase the NINODE value:

WARNING: iget - inide table overflow
WARNING: SFSinide table overflow

The sar(1) command allows you to check the cache hit ratio, the number of data transfers between the buffer and the disk, and the number of accesses. For more information on the sar command, refer to the SUPER-UX User's Reference Manual.
NFILE 1200 100 10000 NFILE is the number of files that can be opened at the same time. NFILE must be larger than NPROC and less than or equal to NINODE. NFILE is determined depending on NPROC and NINODE. When you receive the following message, you should increase the NFILE value:

NOTICE: File table overflow

NMOUNT 100 10 500 NMOUNT is the number of files to be mounted including files mounted by NFS. Depending on the lowest limit required for the NMOUNT value, you can increase the processing rate efficiently. Estimate a suitable value for the system site.
FLCKREC 100 10 10000 FLCKREC is the number of tables used by commands or system calls using the record lock function. One table is used each time a record is locked by the record lock function. When the fcntl(2) system call is used to lock a record with F_SETLK or F_SETLKW specified, this table is created. For more information on the fcntl system call, refer to the SUPER-UX Programmer's Reference Manual. The sar(1) command allows you to check the usage status of the record lock function. For more information on the sar command, refer to the SUPER-UX User's Reference Manual.

Home

Table 1-3 Optimizing with File Management Functions (cont'd)

Name Default Minimum Maximum Explanation
NQUOTA 256 128 60000 NQUOTA is the number of a management table for the disc quotas function that is allocated to each user. NQUOTA limits the number of users who can log into the system. It is recommended that the maximum be the number of users defined on the system. Normally, specify number of concurrent users * 2. A higher NQUOTA value allows more data to reside in memory, thus increasing the processing efficiency. When you receive the following message frequently, you should increase the NQUOTA value:

WARNING: quota: table is full

Estimate an appropriate value according to the usage status of the quota function and the number of users.
NDQUOT 6400 256 120000 NDQUOT is the number of a management table for the disc quotas function that is
allocated to each user and each file system. It is recommended that the maximum be NQUOTA* NMOUNT. Normally, specify NQUOTA* number of file systems using the quota function. When you receive the following message frequently, you should increase the NDQUOT value:

WARNING: dquota: table is full

Estimate an appropriate value according to the usage status of the quota function and the number of users.

NQUOTAG 128 32 32767 NQUOTAG is the number of a management table for the disc quotas function that is allocated to each group. NQUOTAG limits the number of groups who can log into the system. Normally, specify number of concurrent groups * 2. A higher NQUOTAG value allows more data to reside in memory, thus increasing the processing efficiency. When you receive the following message frequently, you should increase the NQOUTAG value:

WARNING: quota for group: table is full

Estimate an appropriate value according to the usage status of the quota function and the number of groups.

Home

Table 1-3 Optimizing with File Management Functions (cont'd)

Name Default Minimum Maximum Explanation
NDQUOTG 2048 64 65535 NDQUOTG is the number of a management table for the disc quotas function that is allo cated to each group and each file system. It is recommended that the maximum be NQUOTAG * NMOUNT. Normally, specify NQUOTAG * number of file systems using the quota function. When you receive the following message frequently, you should increase the NDQUOTG value:

WARNING: dquota for group: table is full

Estimate an appropriate value according to the usage status of the quota function and the number of users.
BDFLUSHR 30 1 120 BDFLUSHR is the starting interval (in seconds) of the sync daemon that periodically modifies the file system. The sync daemon updates the superblock (information peculiar to the system) that is modified on a memory, inode (file information), or system buffer that is being delayed to write onto the disk. Increasing this value reduces the load on the system, but can lower reliability.
NAIOREQ 64 1 256 NAIOREQ is the maximum number of asynchronous I/O requests that are allowed in pending (not yet completed) status when the asynchronous I/O function is used. When you receive the following message, you should increase the NAIOREQ value:

NOTICE: Asynchronous I/O table over flow

Define this value according to your expected frequency of asynchronous operations.
ULIMIT 0x10000000 0x2000 0x3fffffffffffff ULIMIT is the maximum size (in 512-byte blocks) of a file that can be created on SFS.
SFSHCNT 0 0 500 SFSHCNT specifies the number of SFS/H file systems created on a system. When an SFS/H file system is not used, set SFSHCNT to 0. Otherwise, set the number of file systems.
RSTCHOWN 0 1 1 RSTCHOWN is a handling flag that specifies whether chown (file owner modification) use is restricted to the super-user. If this value is 1, only the super-user can use chown. If this value is 0, the owner of the file and super-user can use chown.

Home

Table 1-3 Optimizing with File Management Functions (cont'd)

Name Default Minimum Maximum Explanation
NTMPDATA 131072 0 1048576 NTMPDATA is the temporary file information data area size (/dev/temp) in bytes. This area is required to use the temporary file system. The default is about 130 KB, which means about 1300 requests may be issued at the same time. To disable the temporary file function, specify 0. When you receive the following message frequently, you should increase the NTMPDATA value:

WARNING: temp-write: /dev/temp over flow

AMCACHE 1024 1024 8192 Specify the number of blocks for the capacity of AM cache on the main memory.
CLSTH 0 0 1 Specify the flag which decides the number of cache clsth table. Specifying 0 holds down the memory consumption and reserves clsth. Specifying 1 reserves clsth by giving priority to the cache hit. Specifying 1 needs twice as much memory for clsth as specifying 0.
Specify 1 for the site that uses a lot of small-cluster SFS or the 128KB-cluster SFS.
VVTH 128 0 1024 VVTH is the number of the table that manages space information on the reallocation facility. Estimate a suitable value depending on the amount of the file system capacity on the reallocation facility. When you receive the following message, you should increase the VVTH value:

NOTICE: get VVTOC - VVTH table overflow

1.4.4 Network Management

The kernel can be optimized for the Network function, depending on the modifications made to the following system constants.

Table 1-4 Optimizing with Network Functions

Name Default Minimum Maximum Explanation
LHIPMAX 1 0 32 LHIPMAX shows the number of HIPPI network interfaces that use the HIPPI channel as a sending/receiving channel. The interface requires one HIPPI channel for sending and one HIPPI channel for receiving. Because the maximum number of the HIPPI channels is 64, the system can have up to 32 interfaces. There is no need to specify more interfaces than the maximum number of HIPPI channel pairs that can be connected (one pair consists of two channels). To use system resources more efficiently, specify the actual number of supported HIPPI network interfaces.

Home

Table 1-4 Optimizing with Network Functions (cont'd)

Name Default Minimum Maximum Explanation
PTYMAX 160 1 160 PTYMAX shows the number of users that can log in at the same time. The number of registered users is sufficient for PTYMAX. Estimate a suitable value depending on the number of users. However if the system includes the program product MLS, PTYMAX is changed as follows:
  • MLS1
    Default: 32 Minimum: 1 Maximum: 32
  • MLS2
    Default: 64 Minimum: 1 Maximum: 64
  • MLS3
    Default: 128 Minimum: 1 Maximum: 128
The current value is automatically modified to the maximum value if the current value is more than the maximum value which is changed because of applying of the MLS package.
The current value is not modified if it is within the range of the value which is changed.
Please confirm the current value before making the kernel in case that the MLS package is applied.

Refer to pty(7) of the SUPER-UX System Administrator's Reference Manual for more information.
NSTREAM 320 60 1500 NSTREAM is the total number of STREAMS that may be opened at one time in a system. Because tty is installed as STREAMS, a value equal to or larger than PTYMAX * 2 must be specified. When you receive the following message frequently, you should increase the NSTREAM value:

NOTICE: stropen: out of streams

The netstat(1) command or the strstat command, a subcommand of the crash(1M) command, allows you to check these system constant values or statistical data. For more information on the netstat command, refer to the SUPER-UX User's Reference Manual. For more information on the crash command, refer to the SUPER-UX System Administrator's Reference Manual.
IPFORWARDING 1 0 1 IPFORWARDING is the flag used to specify whether or not the relay of the received packet is effective. Specify 0 to deactivate the relay function intentionally based on the formation of the network or condition of the traffic.

Home

Table 1-4 Optimizing with Network Functions (cont'd)

Name Default Minimum Maximum Explanation
NBUF 500 50 4500 NBUF is the number of system buffers used to carry out block-type I/O by using NFS. As the system buffers increase in number, it is possible to speed up the input/output process since the data to be accessed is likely in the system buffers. The sar(1) command allows you to check the cache hit ratio, the number of data transfers between the buffer and the disk, and the number of accesses. When the cache hit ratio is extremely low because of too many access requests, NSBUF tuning is necessary because the buffer may be out of space. For more information on the sar command, refer to the SUPER-UX User's Reference Manual.
NFS_RNUM 50 50 1300 NFS_RNUM is the number of remote files in the system that can be accessed at the same time. If an attempt is made to access a number of remote files exceeding this number, the following message is shown and the NFS_RNUM value should be increased:

WARNING: nfs_iget: rnode table overflow

1.4.5 Accounting Management

The kernel can be optimized for the Accounting Management depending on the modifications made to the following system constants.

Table 1-5 Optimizing with Accounting Management Functions

Name Default Minimum Maximum Explanation
ACCTPSARGS 0 0 1 ACCTPSARGS is the flag used to specify whether or not to save the command-line image of a process. The maximum size of the image is 80 bytes. To save the image, specify 1. You cannot change the value during operation.
ACCTFSIZE 100 20 0x7fffffff ACCTFSIZE is the maximum size (in MB) of the process accounting file. Estimate a suitable value in consideration of the system operation of the site.
CPURESTM 0 0 1 CPURESTM is the flag used to specify whether or not to save the resident time for every number of CPUs. To save the image specify 1.

Home

Contents Next Chapter Index