|
||||||||||
|
|
|
|
|
|
|
|
|
|
|
|
||||||||||
|
|||
 |
 |
HP OpenVMS systems documentation |
|
| Previous | Contents | Index |
On VAX systems, CRD_CONTROL serves the function of CRDENABLE in earlier releases. On Alpha and I64 systems, CRD_CONTROL can be used to expand the function defined by CRDENABLE.
CRD_CONTROL is a bit mask for corrected read data (CRD) soft error control flags. These flags control the use of CRDERROR routines.
On VAX systems, the following bits are defined:
| Bit | Description |
|---|---|
| 0 | Enables CRD processing for all systems. |
| 1 | Enables scrubbing (rewriting) of the memory location that induced the CRD. |
| 2 | Enables page replacement of the pages that exhibit repeated CRD errors. |
| 3 | Forces all memory pages to be included in the PFN database. On systems that contain more than 512 megabytes of memory, all memory is mapped by the PFN database by default. This bit allows the mapping to occur on systems with less than 512 megabytes of memory. |
Default values are different for VAX, Alpha, and I64 systems. On VAX systems, the default is 7, which enables CRD processing, scrubbing, and page replacement.
On Alpha and I64 systems, the following bits are defined:
| Bit | Description |
|---|---|
| 0 | Enables CRD processing for all systems. |
| 1 | Enables scrubbing (rewriting) of the memory location that induced the CRD. |
| 2 | Enables page replacement of the pages that exhibit repeated CRD errors. |
| 3 | Forces all memory pages to be included in the PFN database. On systems that contain more than 512 megabytes of memory, all memory is mapped by the PFN database by default. This bit allows the mapping to occur on systems with less than 512 megabytes of memory. |
| 4 | Enables extended CRD handling, if available. |
| 5 | Enables loading of driver and process for handling server management events. Platform-specific code usually sets this bit if the required hardware and firmware support are available. |
| 6 | Disables CRD throttling. |
| 7 | Disables System Event Log (SEL) polling. |
| 16-31 | Reserved for platform-specific error-handling control. |
On Alpha and I64 systems, the default setting is 22, which enables scrubbing, page replacement, and extended CRD handling.
Beginning with OpenVMS Version 7.2, CRD_CONTROL can expand the function of CRDENABLE. (Refer to CRD_CONTROL.)
This special parameter is used by HP and is subject to change. Do not change this parameter unless HP recommends that you do so.
This special parameter is used by HP and is subject to change. Do not change this parameter unless HP recommends that you do so.
Prior to OpenVMS Version 7.3-1, if no process name was supplied, the system constructed a name by appending _n to the username, where n was the next available non-duplicate integer for any process currently in the system. For example, the first spawned process from user SYSTEM would be called SYSTEM_1, the second, SYSTEM_2, and so on. The next available number was chosen, as soon as a gap was found.
A problem with this technique is that determining the next available number is very expensive in terms of performance, because the mechanism attempts to create the process by incrementing names until one is found that unique. When several subprocesses already exist, the cost of creating the subprocess iteratively becomes even more expensive. When many processes are in the same OpenVMS group, the cost multiplies because process names must be unique throughout the group.
Beginning in OpenVMS Version 7.3-1, the default constructed process name for subprocesses has changed. Instead of incrementally searching for the next unique number, a random number is chosen to append to the username. Therefore, the first processes that are spawned from user SYSTEM might be SYSTEM_154, SYSTEM_42, SYSTEM_87, and so on. This procedure results in a very high probability of finding a unique name on the first try, because it is unlikely that the same number is already in use. This greatly reduces the cost of process creation, and applications that rely on spawned subprocesses might see a dramatic performance improvement due to this change.
However, some applications might rely on the previous method of assigning subprocess names. The DCL_CTLFLAGS parameter is available to allow you to configure the system as necessary.
Bit 0 of DCL_CTLFLAGS selects the behavior for assigning default subprocess names:
The value of DEFQUEPRI should be less than or equal to MAXQUEPRI.
DEFQUEPRI refers to relative queue scheduling priority, not the execution priority of the job. |
| Option | Description |
|---|---|
| 0 | Disable the exit handler functionality with $DELPRC. |
| 4 | Execute kernel mode exit handlers. |
| 5 (default) | Execute exec and more privileged mode exit handlers. |
| 6 | Execute supervisor and more privileged mode exit handlers. |
| 7 | Execute user and more privileged mode exit handlers. |
Following is the bit definition:
| Bit | Definition |
|---|---|
| 0 | If 1, enable new naming. |
| 1 | Must be 0. This bit is reserved for use by HP. |
| 2 | If 1, cloned device unit numbers wrap after 9999. |
For more information about port allocation classes, see HP OpenVMS Cluster Systems.
This special parameter is used by HP and is subject to change. Do not change this parameter unless HP recommends that you do so.
This parameter is not used on Alpha or I64 systems.
This special parameter is used by HP and is subject to change. Do not change this parameter unless HP recommends that you do so.
Increasing the value of DORMANTWAIT can increase the interval that a low priority process blocks a high priority process if that low priority process is holding a lock or resource that the higher priority process is waiting for.
DR_UNIT_BASE provides a way for unique RAID device numbers to be generated. DR devices are numbered starting with the value of DR_UNIT_BASE and then counting from there. For example, setting DR_UNIT_BASE to 10 produces device names such as $1$DRA10, $1$DRA11, and so on.
Setting DR_UNIT_BASE to appropriate, nonoverlapping values on all cluster members that share the same (nonzero) allocation class ensures that no two RAID devices are given the same name.
DUMPSTYLE is a 32-bit mask, with the following bits defined. Each bit can be set independently. The value of the system parameter is the sum of the values of the bits that have been set. Remaining or undefined values are reserved for HP use only.
| Bit | Mask | Description | |
|---|---|---|---|
| 0 | 00000001 | 0 = | Full dump (SYSGEN default). The entire contents of physical memory are written to the dump file. |
| 1 = | Selective dump. The contents of memory are written to the dump file selectively to maximize the usefulness of the dump file while conserving disk space. | ||
| 1 | 00000002 | 0 = | Minimal console output. |
| 1 = | Full console output (includes stack dump, register contents, and so on). | ||
| 2 | 00000004 | 0 = | Dump to system disk. |
| 1 = | Dump off system disk (DOSD) to an alternate disk. (Refer to the HP OpenVMS System Manager's Manual for details.) | ||
| 3 (Alpha and I64) 1 | 00000008 | 0 = | Do not compress. |
| 1 = | Compress. (See note below.) | ||
| 4 (Alpha and I64) 2 | 00000010 | 0 = | Dump shared memory. |
| 1 = | Do not dump shared memory. (See note below.) | ||
| 5 - 14 | Reserved for HP use only. | ||
| 15 (VAX only) 3 | 00008000 | 0 = | Disable use of bits 16 - 27. |
| 1 = | Enable use of bits 16 - 27. | ||
| 16 - 27 (VAX only) 2 | 0FFF0000 | Range of DOSD unit numbers. | |
| 28 - 31 | Reserved for HP use only. | ||
If you plan to enable the Volume Shadowing minimerge feature on an Alpha or I64 system disk, be sure to specify DOSD to an alternate disk.
On Alpha and I64 systems, you can save space on the system disk and, in the event of a crash, save time recording the system memory, by using the OpenVMS Alpha and I64 dump compression feature. Unless you override the default AUTOGEN calculations (by setting DUMPSTYLE in MODPARAMS.DAT), AUTOGEN uses the following algorithm:
|
The mask of 00000006 directs the system to send a full dump, with full console output, off the system disk (to the alternate disk).
For a VAX 7000, a mask of 00098006 directs the system to send a full dump with full console output to the DOSD whose unit number is 9.
On Alpha and I64 systems, the mask of 00000009 directs the system to compress a selective dump with minimal console output.
If you increase ERLBUFFERPAG_S2, you must either run AUTOGEN or manually increase the size of both the system dump file and the error log dump file.
If you increase ERRORLOGBUFF_S2, you must either run AUTOGEN or manually increase the size of both the system dump file and the error log dump file.
Starting in OpenVMS Version 7.2, FAST_PATH is enabled by default. In Versions 7.0 and 7.1, FAST_PATH was disabled by default.
For additional information, see FAST_PATH_PORTS and IO_PREFER_CPUS.
FAST_PATH_PORTS is a 32-bit mask, with a bit assigned for each Fast Path port driver. The following table describes the bit values:
| Bit Value | Description |
|---|---|
| 1 | Indicates that Fast Path is disabled for ports serviced by the corresponding driver. |
| 0 | Indicates that Fast Path is not disabled for ports serviced by the corresponding driver. |
Beginning in OpenVMS Version 7.3-1, values of specific bit positions are those described in the following table:
| Bit Position | Description |
|---|---|
| 0 | Controls Fast Path for PKQDRIVER (for parallel SCSI). |
| 1 | Controls Fast Path for FGEDRIVER (for Emulex LP7000, LP8000, LP9002, LP9802, LP10000 FibreChannel). |
| 2 | Controls Fast Path for PKADRIVER (for Adaptec AIC-78xx Ultra3 SCSI). |
| 3 | Controls Fast Path for PEDRIVER (for LAN). |
| 4 | Controls Fast Path for PKRDRIVER (for SMART Array 5300). |
| 5 | Controls Fast Path for PKMDRIVER, the LSI Logic LSI53C1030 SCSI port driver. |
| 6 | Controls Fast Path for PGQDRIVER, the Qlogic ISP23xx FibreChannel port driver. |
Currently, the default setting for FAST_PATH_PORTS is 0, which means that Fast Path is enabled for all drivers that appear in the table.
In addition, note the following:
For additional information, see FAST_PATH and IO_PREFER_CPUS. For an explanation of how to set the bits, see the HP OpenVMS I/O User's Reference Manual.
The system writes pages from the modified-page list, swaps out working sets, or reduces the size of the working sets to maintain the minimum count.
While the larger free-page list generally means less page I/O, it also means less space for the balance set, which tends to result in more swap I/O. You can monitor the size of the free-page list, the amount of page, and the amount of swap with the MONITOR IO command of the Monitor utility.
Specify one of the following:
| Value | Description |
|---|---|
| 0 | The default. Do not participate in a memory sharing. |
| 1 | Participate in a memory sharing sharing. |
When you set GALAXY to 1 in a hard partition, OpenVMS instances will share memory between soft partitions within that hard partition. (You can run more than two soft partitions in a hard partition, and you might might not want to share memory among all of them.) Note that GALAXY specifies only if a node uses shared memory. You do not need to use the parameter to run multiple cooperative instances of OpenVMS; you do this by console setup of the configuration tree that you want.
Users with CMKRNL privilege can change this parameter on a running system. Increasing the value of this parameter allows the global page table to expand, on demand, up to the maximum size.
The default value is sufficient for the images normally installed as shared in the system startup command procedures. Once the system is running and all global sections are created, you can examine the actual requirements with the /GLOBAL qualifier of the Install utility (INSTALL) and reduce the value of GBLPAGES accordingly. However, do not set the value of this parameter too low, because the page table entries use little permanently resident memory. If you plan to install many user images as shared, or if user programs are likely to create many global sections, you must increase the value of this parameter.
Global page-file sections are created with the Create and Map Section system services ($CREATE_GPFILE, $CRMPSC, and $CRMPSC_GPFILE_64) without an explicit disk file. These sections are used for the RMS global buffers required for shared files. Users of shared files should note that global page-file sections cause both the global page table and the default system page file (PAGEFILE.SYS) to be used. If the value of GBLPAGFIL is too small, $CRMPSC issues an error message when you attempt to create global page-file sections.
You must have scratch global sections if you use RMS global buffers. Each file using global buffers requires, in the system page file, the file's bucket size multiplied by the number of global buffers for that file. If the file's bucket size varies, as with RMS indexed files, use the maximum bucket size. For shared sequential files, use the multiblock count of the first stream to perform the $CONNECT service in place of the file's bucket size.
| Previous | Next | Contents | Index |