Configuring an HP-UX Kernel

　　Kernel Commands A new set of kernel-related commands has been developed that have a common behavior. This section covers all the commands and provides examples of using some of them. kcmodule kcmodule queries and changes kernel modules in the currently running kernel configuration or a saved kernel configuration that you are staging for future use. Hundreds of modules are present in an HP-UX kernel that consists of device drivers, kernel subsystems, and other kernel code. kcmodule with no options provides the modules in your system and both their current state and the state on next boot if any changes are pending, as shown in this abbreviated example: # kcmodule Module State Cause Notes DeviceDriver unused KeyboardMUX unused LCentIf static best MouseMUX unused UsbBootKeyboard unused UsbBootMouse unused UsbHub unused UsbMiniBus unused UsbOhci unused acpi_node static best arp static depend asio0 static best audio static best autofsc static best azusa_psm static best beep static depend BTlan static best c460gx_psm static depend c8xx static best cachefsc static best ccio unused cdfs auto best auto-loadable, unloadable cec_hp static depend cell static best cifs static best clone static best consp1 unused diag2 static best dlpi static best dm_sample_fsid unused dmapi unused dmp static depend dmsample unused echo static best ehci unused fcd static best fcms static depend fcp static depend fcp_cdio static depend fcparray static depend fcpdev static depend fcpmux static depend fddi4 static best ffs static best framebuf unused gelan static best graph3 unused gvid unused gvid_core unused hcd unused hid unused hpstreams static best hsx static eXPlicit loadable, unloadable hub unused ia64_psm static best idds unused


. . . This abbreviated output shows some of the modules in the system without mUCh detail. We modify the last module shown, idds, in a later example. To get detailed information on a specific module, use the -v option, as shown for vxfs: # kcmodule -v vxfs Module vxfs [3F559170] Description Veritas Journal File System (VxFS) State static (best state) State at Next Boot static (best state) Capable static unused Depends On module libvxfs:0.0.0 interface HPUX_11_23:1.0 #This verbose output shows more information for the specific module that we specified, or would have shown verbose output for every module if I had not specified the name of a specific module. For every module in the verbose output, there is a name and description such as a name of vxfs, the version number in square brackets after the name, and a short description of the module in the example. It is possible for multiple versions to be listed if, for instance, the currently running kernel uses a different version than will be used on the next boot. The state of the module is relative to the currently running kernel (which is shown in the example), the next boot (both the currently running and next boot states are shown), or a saved configuration. The module in the example is for the currently running kernel, so static means that the module is statically bound into the kernel and changing this state would require relinking the kernel executable and rebooting. The module could also be in the unused state, which means it is installed but not used, the loaded state, which means it has been dynamically loaded into the kernel, or the auto state, which means it will be dynamically loaded when it is first needed but hasn't been loaded yet. The following list shows commonly used options to kcmodule: kcmodule command-line flags: -a Includes all modules in the output. -B Backs up the currently running configuration prior to changing it. -c config Specifies the saved configuration to manage. If none is specified, manage the currently running configuration. -C comment Includes a comment pertaining to this invocation of the command in the kernel configuration log file. -d Adds the description for each item. -D Displays elements for which there is a pending change at the next boot. -h Holds the specified change(s) for the next boot. -K Keeps the currently running configuration, but does not back it up. Keep the existing backup unmodified.

-P Parses using the specified output format. -S Displays the elements that have been set to something other than the default. -v Displays items using verbose output. Some of these options will be used in the upcoming example of updating the kernel. kctune kctune queries and changes the value of kernel tunable parameters in the currently running kernel configuration or a saved kernel configuration that you are staging for future use. kctune with no options provides the parameters in your system, as shown in the following abbreviated example: # kctune Tunable Value Expression Changes NSTREVENT 50 Default NSTRPUSH 16 Default NSTRSCHED 0 Default STRCTLSZ 1024 Default STRMSGSZ 0 Default acctresume 4 Default acctsuspend 2 Default aio_listio_max 256 Default Immed aio_max_ops 2048 Default Immed aio_monitor_run_sec 30 Default Immed aio_physmem_pct 10 Default aio_prio_delta_max 20 Default Immed aio_proc_thread_pct 70 Default Immed aio_proc_threads 1024 Default Immed aio_req_per_thread 1 Default Immed allocate_fs_swapmap 0 Default alwaysdump 0 Default Immed bufcache_hash_locks 128 Default chanq_hash_locks 256 Default core_addshmem_read 1 1 Immed core_addshmem_write 1 1 Immed create_fastlinks 0 Default dbc_max_pct 50 Default Immed dbc_min_pct 5 Default Immed default_disk_ir 0 Default disksort_seconds 0 Default dma32_pool_size 268435456 Default dmp_rootdev_is_vol 0 Default dmp_swapdev_is_vol 0 Default dnlc_hash_locks 512 Default dontdump 0 Default Immed dst 1 Default dump_compress_on 1 Default Immed enable_idds 0 Default Immed eqmemsize 15 Default executable_stack 0 Default Immed fs_async 0 Default fs_symlinks 20 Default Immed ftable_hash_locks 64 Default hp_hfs_mtra_enabled 1 Default io_ports_hash_locks 64 Default ksi_alloc_max 33600 Default Immed

ksi_send_max 32 Default max_acct_file_size 2560000 Default Immed max_async_ports 50 Default max_mem_window 0 Default max_thread_proc 256 Default Immed maxdsiz 1073741824 Default Immed maxdsiz_64bit 4294967296 Default Immed maxfiles 8192 8192 maxfiles_lim 8192 8192 Immed maxrsessiz 8388608 Default maxrsessiz_64bit 8388608 Default maxssiz 8388608 Default Immed maxssiz_64bit 268435456 Default Immed maxtsiz 100663296 Default Immed maxtsiz_64bit 1073741824 Default Immed maxuprc 256 Default Immed maxvgs 10 Default msgmap 1026 Default msgmax 8192 Default Immed msgmnb 16384 Default Immed msgmni 512 Default msgseg 8192 Default msgssz 96 Default msgtql 1024 Default ncdnode 150 Default Immed nclist 8292 Default ncsize 8976 Default nfile 65536 Default Auto nflocks 4096 Default Auto ninode 4880 Default . . . This output shows the tunable parameter, its current value, the expressions used to compute the value (which is the default in all cases in the example except for maxfiles), and changes to the value if any are pending. In an upcoming example, I modify the nproc tunable. Using the -d option, which also works with kcmodule, adds a description for each parameter, as shown in the following truncated example: # kctune -d Tunable Value Expression Changes Description NSTREVENT 50 Default Maximum number of concurrent Streams bufcalls NSTRPUSH 16 Default Maximum number of Streams modules in a stream NSTRSCHED 0 Default Number of Streams scheduler daemons to run (0 = automatic) STRCTLSZ 1024 Default Maximum size of the control portion of a Streams message (bytes) STRMSGSZ 0 Default Maximum size of the data portion of a Streams message (bytes; 0 = unlimited) acctresume 4 Default Relative percentage of free disk space required to resume accounting acctsuspend 2 Default

Relative percentage of free disk space below which accounting is suspended aio_listio_max 256 Default Immed Maximum number of async IO operations that can be specified in lio_list call aio_max_ops 2048 Default Immed Maximum number of async IO operations that can be queued at any time aio_monitor_run_sec 30 Default Immed Frequency of AIO Thread Pool Monitor Execution (in seconds) aio_physmem_pct 10 Default . . . Each module now has a more detailed description associated with it as a result of using the -d option. To group parameters based on the kernel module that defines the tunable, use the -g option, as shown in the following abbreviated example: # kctune -g Module Tunable Value Expression Changes cdfs ncdnode 150 Default Immed dump alwaysdump 0 Default Immed dump dontdump 0 Default Immed dump dump_compress_on 1 Default Immed fs bufcache_hash_locks 128 Default fs dbc_max_pct 50 Default Immed fs dbc_min_pct 5 Default Immed fs disksort_seconds 0 Default fs dnlc_hash_locks 512 Default fs fs_async 0 Default fs fs_symlinks 20 Default Immed fs ftable_hash_locks 64 Default fs maxfiles 8192 8192 fs maxfiles_lim 8192 8192 Immed fs ncsize 8976 Default fs nfile 65536 Default Auto fs nflocks 4096 Default Auto fs o_sync_is_o_dsync 0 Default fs sendfile_max 0 Default fs vnode_cd_hash_locks 128 Default fs vnode_hash_locks 128 Default hpstreams NSTREVENT 50 Default hpstreams NSTRPUSH 16 Default hpstreams NSTRSCHED 0 Default hpstreams STRCTLSZ 1024 Default hpstreams STRMSGSZ 0 Default hpstreams streampipes 0 Default idds enable_idds 0 Default Immed inet tcphashsz 2048 Default io aio_listio_max 256 Default Immed io aio_max_ops 2048 Default Immed

io aio_monitor_run_sec 30 Default Immed io aio_physmem_pct 10 Default io aio_prio_delta_max 20 Default Immed io aio_proc_thread_pct 70 Default Immed io aio_proc_threads 1024 Default Immed io aio_req_per_thread 1 Default Immed io io_ports_hash_locks 64 Default io max_async_ports 50 Default ite scroll_lines 100 Default Immed lvm maxvgs 10 Default pm acctresume 4 Default pm acctsuspend 2 Default pm chanq_hash_locks 256 Default pm dst 1 Default . . . This output shows all the tunables grouped with their kernel modules. You can see that, in the case of the fs module, for example, many tunables are associated with some modules. The -v output, as shown in the following abbreviated example, provides a lot of tunable-related information: # kctune -v Tunable NSTREVENT Description Maximum number of concurrent Streams bufcalls Module hpstreams Current Value 50 [Default] Value at Next Boot 50 [Default] Value at Last Boot 50 Default Value 50 Can Change At Next Boot Only Tunable NSTRPUSH Description Maximum number of Streams modules in a stream Module hpstreams Current Value 16 [Default] Value at Next Boot 16 [Default] Value at Last Boot 16 Default Value 16 Can Change At Next Boot Only Tunable NSTRSCHED Description Number of Streams scheduler daemons to run (0 = automatic) Module hpstreams Current Value 0 [Default] Value at Next Boot 0 [Default] Value at Last Boot 0 Default Value 0 Can Change At Next Boot Only Tunable STRCTLSZ Description Maximum size of the control portion of a Streams message (by tes) Module hpstreams Current Value 1024 [Default] Value at Next Boot 1024 [Default] Value at Last Boot 1024 Default Value 1024 Can Change At Next Boot Only Tunable STRMSGSZ Description Maximum size of the data portion of a Streams message (bytes ; 0 = unlimited) Module hpstreams Current Value 0 [Default] Value at Next Boot 0 [Default] Value at Last Boot 0

Default Value 0 Can Change At Next Boot Only Standard input . . . This output shows additional information, including the value the parameter will have upon the next boot. All tunables have manual pages. If, for example, you want to know more about a tunable, just issue the man for the tunable, as shown in the following example for nproc: # man nproc nproc(5) nproc(5) Tunable Kernel Parameters NAME nproc - limits the number of processes allowed to exist simultaneously VALUES Failsafe 4200 Default 4200 Allowed values 100 - 30000 This may be set higher, but more will not be used. Setting nproc below 110 will interfere with the systems ability to execute in multi-user mode. Some configurations may have a higher minimum. nproc must be greater than nkthread + 100. Standard input Only the beginning of the nproc output is shown, but you can see there is a lot of useful information in the man page, such as the default and allowed values. The following list shows commonly used options to kctune. kctune command-line flags: -a Includes all information in the output that is normally suppressed. -B Backs up the currently running configuration prior to changing it. -c config Specifies the saved configuration to manage. If none is specified, manages the currently running configuration. -C comment Includes a comment pertaining to this invocation of the command in the kernel configuration log file. -d Displays the descriptions of each item. -D Displays elements for which there is a pending change at the next boot. -g Groups related tunables in the output. -h Holds the specified change(s) for the next boot. -K Keeps the currently running configuration, but does not back it up. Keeps the existing backup unmodified. -P Parses using the specified output format. -S Displays the elements that have been set to something other than the default. -u Allows the creation of user-defined tunables. -v Displays items using verbose output. Some of these options are used in the upcoming kernel configuration example. kconfig kconfig manages kernel configurations. Running kconfig with no options shows you all the saved kernel configurations on your system. You can view the output using -a for all, -v for verbose, and -P for parse.

Changes to the kernel can be delayed until the next boot using the -n option to kconfig. (Next boot options are also available with the -h option for kcmodule and kctune.) To obtain a list of changes being held for the next boot, you would use the -D option to kconfig to show differences between the currently running kernel and what is planned for the next boot. You could also run kcmodule and kctune with the -D option to get this same list. If you don't want these changes to be applied at the next boot, use the -H option to kconfig, which "unholds" the settings. To obtain a list of non-default kernel values, use the -S option to kconfig. This too is a shortcut for running both kcmodule and kctune with the -S option. You can specify a saved configuration with no option to kconfig by just naming it on the command line or by using the -c option for kcmodule and kctune. If you have made changes to the currently running kernel and want to save, use the -s option to kconfig. You can load a saved configuration using the -l option to kconfig. If the configuration can be loaded without a reboot, the change takes effect immediately. If not, the change is held for the next reboot. You can specify that the change be applied at the next reboot with the -n option for kconfig. You can identify the configuration to be loaded a the next boot using the -w option to kconfig. Several of these options are used in the upcoming kernel reconfiguration example. The following is a list of commonly used kconfig options. kconfig command-line flags: -a Includes all information in the output that is normally suppressed. This provides a lot of information so you may want to pipe this to more. -B Backs up the currently running configuration prior to changing it. -C comment Includes a comment pertaining to this invocation of the command in the kernel configuration log file. -d config Deletes the specified kernel configuration. -D Displays elements for which there is a pending change at the next boot. -e Exports the saved configuration. -h Holds the specified change(s) for the next boot. -H Discards all changes being held for the next boot. -i config Imports the specified configuration. -K Keeps the currently running configuration, but does not back it up. Keeps the existing backup unmodified. -l config Loads the specified configuration. -P Parses using the specified output format. -S Displays the elements that have been set to something other than the default. -v Displays items using verbose output.

kclog kclog manages the kernel configuration log file. All the commands previously covered (kcmodule, kctune, and kconfig) update and maintain the kernel configuration plain text log file called /var/adm/kc.log. You can view this file directly to see the kernel-related commands that have been issued, which is the way that I view kernel-related changes, or issue the kclog command to view /var/adm/kc.log. kclog has the following commonly used command-line flags: kclog commonly used command line options: -a Prints all entries matching -c config Prints log file entries from the specified configuration. -C comment Includes the specified comment. kcusage kcusage shows the usage level of kernel resources. If you issue the kcusage command with no options, you get output for the currently running system, as shown in the following output: # kcusage Tunable Usage / Setting ============================================= dbc_max_pct 5 / 50 maxdsiz 37666816 / 1073741824 maxdsiz_64bit 7258112 / 4294967296 maxfiles_lim 56 / 8192 maxssiz 1179648 / 8388608 maxssiz_64bit 20480 / 268435456 maxtsiz 421888 / 100663296 maxtsiz_64bit 237568 / 1073741824 maxuprc 0 / 256 max_thread_proc 57 / 256 msgmni 2 / 512 msgseg 0 / 8192 msgtql 0 / 1024 nfile 586 / 65536 nflocks 20 / 4096 ninode 653 / 4880 nkthread 484 / 8416 nproc 150 / 4200 npty 0 / 60 nstrpty 0 / 60 nstrtel 0 / 60 semmni 23 / 2048 semmns 25 / 8192 shmmax 17906400 / 68719476736 shmmni 7 / 400 [rx8620b{root}:/roothome]> This is an idle system so the resources are used minimally. You can specify the time period over which data should be printed, including 24 hours, 31 days, and 52 weeks. Two interesting options to kcusage are -l, which prints a long format, and -t, which prints the top 5 users or processes that have consumed each resource. The following is an obviated listing of these two outputs: # kcusage -l Parameter: dbc_max_pct Usage: 5 Setting: 50 Percentage: 10.0 Parameter: maxdsiz Usage: 37666816 Setting: 1073741824 Percentage: 3.5 Parameter: maxdsiz_64bit