Document revision date: 15 July 2002 | |
![]() |
![]() ![]() ![]() ![]() |
![]() |
Previous | Contents | Index |
OpenVMS Galaxy offers the ability to share data between Galaxy instances, (provided by OpenVMS as Galaxy global sections). OpenVMS Version 7.3--1 adds support in the INSTALL utility and the image activator to place write shared image sections in Galaxy global sections. This feature allows existing applications that share data using writeable image sections to be extended to a Galaxy environment.
INSTALL/ADD WRITABLE=GALAXY places write shared image sections in Galaxy global sections. You can also use the IDENT keyword with GALAXY to include the image identity in the name of the Galaxy global section. Doing this lets you use multiple versions of an image simultaneously in a Galaxy system.
Refer to the OpenVMS Alpha Partitioning and Galaxy Guide for more details.
5.10 Larger Mailbox Buffer Quotas
Prior to Version 7.3--1, the maximum size for the mailbox buffer quota was 64,000 bytes. OpenVMS Alpha Version 7.3--1 supports the creation of mailboxes with larger buffer quotas for improved application scaling.
When creating a mailbox with the $CREMBX system service, you can explicitly supply a mailbox buffer quota with the BUFQUO parameter. If not, the mailbox buffer quota is implicitly supplied by the system parameter DEFMBXBUFQUO.
Note the following:
The following sections describe new RMS features added to the OpenVMS
Alpha Version 7.3--1 operating system that improve I/O performance and
enable easier porting from UNIX to OpenVMS Alpha.
5.11.1 Increase in Default I/O Transfer Size to Reduce I/O
The multiblock count for a sequential disk file sizes the RMS intermediate buffer for disk block transfers; this count serves as the I/O transfer size. The system default multiblock count is increased from 16 to 32. This increase reduces sequential file read or write I/O requests by a factor of 2. If the application does not assign the multiblock count explicitly (using the RAB$B_MBC setting) or as a process default (using the DCL command SET RMS_DEFAULT), the new system default multiblock count of 32 applies.
The process or system multiblock count default parameters (set with the SET RMS_DEFAULT/BLOCK_COUNT or SET RMS_DEFAULT/SYSTEM/BLOCK_COUNT command, or with or the SYSGEN parameter RMS_DFMBC) can dynamically restore the default value for a process or system to the former value of 16 blocks. You can also increase the multiblock count to greater than 32 at either the process or system level, after making appropriate changes to working set quotas. The maximum multiblock count is 127 blocks. Because a larger RMS buffer can lead to increased page faulting by a process with a small working set quota, the increase in the system default is limited to 32.
Because files opened as process-permanent files (PPF) by DCL are not affected by process or system multiblock defaults, DCL file opens are not affected by this improvement. However, note that the change to the system multiblock count default can affect any privileged application that uses the FAB$V_PPF (undocumented) option to open a process-permanent file, if the application does not assign the multiblock count explicitly in the application itself (using the RAB$B_MBC setting).
While this multiblock increase is compatible in a mixed cluster, it has an indirect effect on VAX or older Alpha versions. If a shared sequential file is first opened from the new release, the system multiblock default is 32 blocks. Because the first accessor determines the multiblock count for all subsequent accessors, 32 is the multiblock count for any subsequent accessors opening the file from an older system.
This new feature also indirectly changes the default extend quantity,
which uses 64 blocks (two times the multiblock count of 32) instead of
the current 32 blocks (two times the default multiblock count of 16).
5.11.2 Support for POSIX-Compliant File Timestamps
To enable easier porting of UNIX applications to OpenVMS Alpha, RMS adds support for POSIX-compliant file timestamps. The XABDAT structure has been extended to include the POSIX-compliant access date, attribute modification date, and data modification date ($OPEN and $DISPLAY). This mechanism can also be used to set all dates on a $CREATE, which is likely to happen in the case of a file restoration. Support for these dates is restricted to ODS-5 volumes and requires the volume to be enabled for access date support.
Updating the POSIX-compliant access date, attribute modification date, and data modification date values through RMS is supported using the XABITM structure.
XABITM$_NORECORD can be used to suppress update of access time.
5.11.3 New System RMS Write-Behind Performance Option
An existing user write-behind (RAB$V_WBH setting in RAB$L_ROP) option is available for unshared sequential files to request that RMS overlap writes to disk. Specifying the write-behind option ensures that at least two buffers are allocated by RMS at connect time. When one buffer has been filled with records, RMS switches to the second buffer and issues an asynchronous QIO to write the contents of the first buffer out to disk. RMS does not stall the process, but rather continues to process records in the second buffer while the QIO for the first buffer is completing. Overlapping I/O can be a big performance win for some applications.
OpenVMS Alpha Version 7.3--1 introduces a new system RMS write-behind performance option as a dynamic SYSGEN parameter (RMS_SEQFILE_WBH) to externally request RMS to implement the write-behind feature as a system default (refer to the OpenVMS System Management Utilities Reference Manual). If this system option is set, RMS implements the existing RMS user write-behind option as a system default (regardless of the RAB$L_ROP RAB$V_WBH setting) whenever write access is requested for an unshared sequential file that is opened for image I/O.
This feature is a system option. It is not the default, however, for the following reasons:
OpenVMS Alpha Version 7.3--1 performs a faster large memory dump that significantly reduces the time to write a system crash dump file.
There are also a number of new commands and qualifiers for the System
Dump Analyzer (SDA), including the new SDA Extended File Cache (XFC)
commands. For more information, refer to the OpenVMS Alpha System Analysis Tools Manual.
5.12.1 New VALIDATE TQE Command
The new VALIDATE TQE command validates all the data structures
associated with timer queue entries (TQEs).
5.12.2 New TQE Type
The SHOW TQE command displays timer queue entries (TQEs), including the following new TQE type:
Column | Symbol | Description |
---|---|---|
6 | N | TQE should not be deallocated at AST completion. |
-- | TQE should be deallocated at AST completion. |
Table 5-2 describes the new SDA command qualifiers in OpenVMS Version 7.3--1.
Command | Qualifier | Description |
---|---|---|
SHOW PROCESS | /POOL | Displays the dynamic storage pool of the process P0 (process) or P1 (control) region. The following qualifiers from the SHOW PROCESS command now also apply to the SHOW PROCESS/POOL command: /FREE, /HEADER, /MAXIMUM_BYTES, /STATISTICS, /SUBTYPE, /SUMMARY, /TYPE, and /UNUSED. |
/TQE | Displays all timer queue entries associated with the current process. | |
SHOW RESOURCES | /LIST | Displays summary information for each resource, followed by a list of all locks associated with the resource. |
The SDA Extended File Cache (XFC) commands along with their parameters and qualifiers are described in Table 5-3. These commands provide analysis tools for tuning the extended file cache to improve the performance of your programs.
Command | Function |
---|---|
EXIT | Exits the XFC SDA extensions. This command has no parameters or qualifiers. |
LOAD DSF | Loads the symbols in the specified debug symbol file (DSF) for use with the SDA> FORMAT command. This command has no parameters or qualifiers. |
SHOW CONTEXT |
Displays the contents of an XFC context structure (CTX). It has one
parameter,
address, which is the address of the CTX. If no
address is supplied, then all the context structures are displayed. It
has three qualifiers:
|
SHOW EXTENT | Displays the contents of an extent control block (ECB). It has one parameter, address, which is the address of the ECB. It has no qualifiers. |
SHOW FILE |
Displays the contents of the cache file block (CFB). It has one
parameter,
address, which is the address of the CFB. The /OPEN
and /CLOSED qualifiers, if present, are ignored. If no address is
supplied, then all the CFBs are displayed. It has eight qualifiers:
|
SHOW MEMORY |
Displays information about memory used by the cache. It has no
parameters. It has two qualifiers:
|
SHOW SUMMARY | Displays general information about the Extended File Cache. It has no parameters. It has one qualifier, /STATISTICS, which displays read and write activity arranged by I/O size. |
SHOW TABLES |
Displays both the extent hash table (EHT) and the file hash table
(FHT). It has no parameters. It has four qualifiers:
|
SHOW TRACE |
Displays all or selected portions of the XFC trace buffer, starting
with the most recent entry and moving backward in time. It has no
parameters. It has six qualifiers:
|
SHOW VOLUME |
Displays the contents of a cache volume block (CVB). It has one
parameter,
address, which is the address of a CVB. If no address
is supplied, then all volumes are displayed. It has four qualifiers:
|
The SHOW TRACE and SHOW COLLECT commands from the SDA Spinlock Tracing utility have the following new qualifiers:
Qualifier | Description |
---|---|
/RATES | Reports activity as a rate per second and hold/spin time as a percentage of time. This is the default. |
/TOTALS | Reports activity as a count and hold/spin time as cycles. |
This section describes new and updated system services for OpenVMS Version 7.3--1. The following table summarizes these changes.
For additional information, refer to the OpenVMS System Services Reference Manual.
System Service/Project | Documentation Update |
---|---|
DECdtm |
The following services have been added:
$ACK_EVENT, $ADD_BRANCH[W], $CREATE_UID, $DECLARE_RM[W], $END_BRANCH[W], $FORGET_RM[W], $GETDTI[W], $GET_DEFAULT_TRANS, $JOIN_RM[W], $SETDTI[W], $SET_DEFAULT_TRANS[W], $START_BRANCH[W], $TRANS_EVENT[W] The following services have been updated: $ABORT_TRANS[W], $END_TRANS[W], $START_TRANS[W] |
$ACM[W] | New services for Authentication and Credentials Management (ACM). |
$ASSIGN | A new status code, SS$_EXBYTLM, has been added. |
$CPU_TRANSITION | Two symbolic names, CST$K_CPU_POWER_OFF and CST$K_CPU_POWER_ON, have been added to the tran_code argument. |
$CREMBX | The text has been updated for the SS$_BADPARAM condition value and the bufquo and maxmsg arguments. Note that the maximum value limit of 60,000 bytes for the bufquo argument no longer applies. |
$DELPRC | A new argument, flag, has been added. |
$ERAPAT | The code example has been updated. |
$GETDVI |
Two new item codes, DVI$_MPDEV_CURRENT_PATH and DVI$_MULTIPATH, have
been added.
The following bits have been added to the DVI$_STS item code: UCB$V_ALTBSY, UCB$V_CLUTRAN, UCB$V_DISMOUNT, UCB$V_EXFUNC_SUPP, UCB$V_FAST_PATH, UCB$V_FP_HWINT, UCB$V_IOPOST_LOCAL, UCB$V_LCL_VALID, UCB$V_MNTVERPND, UCB$V_NO_ASSIGN, UCB$V_PATHVERIP, UCB$V_SNAPSHOT, UCB$V_SUPMVMSG, UCB$V_SVPN_END, UCB$V_WRTLOCKMV. |
$GETTIM | The Description section has been updated. |
$GETJPI |
The item code description for JPI$_ASTEN has been updated to include
all processes.
The following persona item codes have been added: JPI$_CLASSIFICATION, JPI$_PERSONA_ID, JPI$_PERSONA_AUTHPRIV, JPI$_PERSONA_PERMPRIV, JPI$_PERSONA_WORKPRIV, JPI$_IMAGE_WORKPRIV, JPI$_IMAGE_AUTHPRIV, JPI$_IMAGE_PERMPRIV, JPI$_PERSONA_RIGHTS, JPI$_SUBSYSTEM_RIGHTS, JPI$_INSTALL_RIGHTS, JPI$_PERSONA_RIGHTS_SIZE, JPI$_SUBSYSTEM_RIGHTS_SIZE, JPI$_SYSTEM_RIGHTS_SIZE, JPI$_INSTALL_RIGHTS_SIZE |
$GETQUI |
The following items have been added for NUMA RAD Batch Job Support:
The QUI$_TRANSLATE_QUEUE function code has been added to the QUI$_QUEUE_NAME item code list. The string length limit has been changed to 255 characters for the QUI$_DEVICE_NAME item code. |
$GETSYI |
The SYI$_NODE_HWVERS item code text has been updated.
A new error status, SS$_UNREACHABLE, has been added. The following item codes have been added: SYI$_ACTIVE_CPU_MASK, SYI$_AVAIL_CPU_MASK, SYI$_CPU_AUTOSTART, SYI$_CPU_FAILOVER, SYI$_CPUCAP_MASK, SYI$_GLX_FORMATION, SYI$_GLX_MAX_MEMBERS, SYI$_GLX_MBR_MEMBER, SYI$_GLX_MBR_NAME, SYI$_GLX_TERMINATION, SYI$_HP_ACTIVE_CPU_CNT, SYI$_HP_ACTIVE_SP_CNT, SYI$_HP_CONFIG_SBB_CNT, SYI$_HP_CONFIG_SP_CNT, SYI$_MAX_CPUS, SYI$_POTENTIAL_CPU_MASK, SYI$_POTENTIALCPU_CNT, SYI$_POWERED_CPU_MASK, SYI$_POWEREDCPU_CNT, SYI$_PRESENT_CPU_MASK, SYI$_PRESENTCPU_CNT, SYI$_PRIMARY_CPUID |
$GETUAI | The context argument restriction for VAX only has been removed. |
$ICC_CONNECT | A new status code, SS$_NOPRIV, has been added. |
$PROCESS_AFFINITY | A new item code, CAP$M_PURGE_WS_IF_NEW_RAD, has been added. |
$PROCESS_CAPABILITIES | A new item code, CAP$M_PURGE_WS_IF_NEW_RAD, has been added. |
$PURGWS | A new condition value, SS$_NOPRIV, has been added. |
$REGISTRY | The item code, REG$_VALUEDATASIZE, has been added. |
$SNDJBC |
The following item codes have been added for NUMA RAD Batch Job
Support: SJC$_RAD and SJC$_NO_RAD.
A new item code, SJC$_JOB_COMPLETION_STATUS, has been added. The string length limit has been changed to 255 characters for the SJC$_DEVICE_NAME item code. |
$SUSPND | A cluster-related note has been added to the description section. |
Previous | Next | Contents | Index |
![]() ![]() ![]() ![]() |
privacy and legal statement | ||
6657PRO_007.HTML |