InterSystems M Technologies
-----

Support

Support News Flash

2003, 2002, 2001, 2000, 1999 , 1998, 1997, 1996
Join our mailing lists and have the News Flashes emailed to you.

Support News Flash for 1998

December 15, 1998

MSM 4.4 Y2K Patches Available

November 25, 1998

3 DSM News Issues

  • DSM and Multi-CPU Alpha machines
    Some fast multi-CPU Alpha DSM configurations may crash with a CLUEXIT VMS bugcheck under certain conditions.

    A situation was recently encountered in which a customer site experienced cluster connection timeouts and some CLUEXIT crashes, which were traced to an unusually high amount of IPL8 activity preventing the VMScluster code from maintaining internode connections. This was the result of VMS Timer Queue Entries (TQEs) being queued at the rate of thousands per second and swamping the primary CPU.

    The catalyst is DSM code used in processes which are waiting for write-inhibited volume sets. Each process queues up waits at the rate of 50 per process per second on the Journal Master node.

    Sites running DSM 6.4 - 7.1.2, especially with cluster mounted database sets, are at risk. This problem is corrected in DSM V7.2.

  • DSM DDP between VMS and OSF
    DSM DDP does not work between DSM 7.* configurations on VMS and DSM for OSF (DEC Unix) configurations.

    The VMS configuration will report that the Unix configuration is unreachable and the Unix configuration will not be able to see the VMS configuration at all. The dsmddpdrv.log file on the Unix system will contain a message: "Received advertisement with unexpected length".

    InterSystems is continuing to investigate this issue and continues to encourage migration from DSM for OSF to ISM or Caché.

  • DSM under VMS 7.1
    InterSystems has diagnosed a problem under VMS 7.1 that causes a DSM process to hang and can lead to the environment hanging. All versions of DSM are susceptible to this problem.

    This will be seen as a process in CURRENT state at Priority 0.

    The solution to this problem is application of Compaq/DEC patches: ALPSYSA02_071, ALPBASE02_071 and ALPDCL01_071

November 25, 1998

Caché Terminal Host Connect Syntax

    Caché Terminal Host Connect Syntax
    In order to connect to a host with Caché Terminal from the command prompt, you will have to use the following syntax:

    cterm /console=cn_iptcp:<host>[<port>] {script file}

    where:
         <host> is the either the IP address or name of the host
         <port> is the telnet port (Caché default is 23)
         {script file} is the optional script file
         * <host> and <port> are required parameters

    Example:
    c:\cachesys\bin> cterm /console=cn_iptcp:123.4.5.6[23] cache.scr
    This will open Caché Terminal and connect to port 23 of machine 123.4.5.6 which will then run a connect script named cache.scr

November 25, 1998

ActiveX Support in MSM Workstation

    MSM Workstation provides the capability to work as an ActiveX Container to enhance the GUI capabilities of GUI Applications.

    There is a wide variety of ActiveX components available in the market that can be purchased as single components or that come with development environments like Visual Basic, Visual C++, Delphi, etc. These ActiveX components can be very complex and are sometimes tightly linked to a specific environment. E.g. there are ActiveX components which provide data binding requiring specific hooks into Visual Basic. Others require an ActiveX Container to be multi threaded to handle asynchronous events. Most of the ActiveX components come with a runtime license and request a development key for GUI design and verification.

    It is this complexity which makes it extremely difficult and time consuming to verify that individual ActiveX compo nents work correctly with MSM Workstation. Because of the architecture (Win32s, single-threaded) of the core engine of MSM Workstation most of them can not be made available.

    InterSystems is therefor dropping the verification of ActiveX components in all MSM Workstation products and it is up to the user to verify that a given ActiveX component correctly works with MSM Workstation.

November 13, 1998

MSM for Windows NT 4.4.0 and %GCH

    MSM for Windows NT 4.4.0 crashes when using %GCH to change the protections of a translated global. The problem only occurs when you have a "local global" (such as "^UTILITY") and a translated global (such as "^[AAA,VGA]UTILITY") hiding the local copy. When you select ^UTILITY within %GCH on the local machine, you will actually select (and edit) your local global--one that you can't access until translation is disabled. If you select the protection option for the local global and make some changes, then MSM will crash when rewriting the global directory block.
November 13, 1998

How to Avoid "tty hash table full" in [ISM] and Caché

    An increasing number of customers are encountering a condition where the terminal table becomes full, thereby preventing additional processes from using [ISM] or Caché. This is easily avoided by increasing the size of the TTY table, ttysiz in the configuration file, and restarting. Since this table is extremely small, there is no penalty for setting it's size to the 64kb (65532 bytes) maximum. Future releases of [ISM] and Caché will increase the default value from 8KB to 64kb to further safeguard against this condition.
November 13, 1998

MSM %DO backward compatibility

    Changes made to %DO in MSM 4.4.0 to show four digit dates did not provide a way for userapplications to maintain backward compatibility to previous versions of MSM. As a result of the changes, dates in the 20th Century (1900-1999 inclusive) converted by %DO and returned in %DS always contained a four digit year.

    There is now a modified version of %DO available on the InterSystems FTP site (ftp.intersys.com) to fix this problem. The version on the FTP site will return a four digit year for dates in the 20th Century (1900-1999 inclusive) only if Switch One is set to "YES" in %MODESET, or inif four digit dates are requested in the configuration setup. Note: It is possible using ^%MODESET to request this behavior on a per job basis.

    To obtain this change, log on to the InterSystems FTP site, and download the file /pub/msm/do.rs.

November 2, 1998

Inconsistencies between information reported by %GSIZE and INTEGRIT explained.

    Recently users have reported that the size of a global when displayed by %GSIZE does not always match the size of the same global displayed by INTEGRIT. Changes made to attempt to correct this problem have not been accurate, and this has added to confusion.

    The following description details what is actually reported.

    %GSIZE

    • %GSIZE reports the size of a global in blocks, and will also display the size in bytes, and the efficiency of the use of space when a detail report is requested.
    • The number of blocks that %GSIZE reports is not correct. %GSIZE reports the number of data blocks and blocks used to store long strings. It fails to include pointer blocks and bottom pointer blocks.
    • Each database block is a 2K (2048 byte) block. However, there are some overhead bytes in most blocks. Each block used to store a global has some overhead information in it. As a result, not all bytes in a block can be used to store data. The number of overhead bytes can vary depending on the use of the block (e.g. Pointer Block, Data Block, or Long String Data Block). When the size of a global is calculated in bytes, the space used in the block to store overhead information is not included in the reported size, however, the space used to store the global reference (indices, etc.) is included. As a result, the size of a global in bytes is not the same as the value obtained by doing a $LENGTH on the global, and in general, it is not possible to predict the number of blocks a global will occupy based solely on the number of bytes in the global value.
    • In calculating the efficiency of use of space by the global, %GSIZE excludes the overhead bytes in a block. This, coupled with the fact that the space used to store the global reference is included in the size of the global reported, cause the reported storage efficiency for the global to vary from what one might otherwise anticipate. The reported efficiency is based on the number of bytes in a block available for use to store the global and its associated global reference, and should be interpreted that way.

    INTEGRIT

    • For Caché 2.1.3, and all previous revisions of Caché, all versions of NextGen, and all versions of Open M back to and including 6.1
    • CHECKPNT, the routine called by INTEGRIT to calculate the size of a global, did not include any long string data in its calculations. As a result, the size of any global that contained long string data would be reported incorrectly.
    • Caché 2.1.6 (up to and including build 28) and all versions of Caché up to 3.1 (up to and including build 23)
    • A repair made to CHECKPNT in Caché 2.1.6 attempted to fix the problem described above. Unfortunately, the calculation was not done in all situations where it was required. Specifically, if the leftmost data block in a global was a type 8 data block (i.e. contained no long string data), the calculation would be skipped, and all long string data in the global would be ignored. When the calculation was done, it produced incorrect results.

    Users may find it useful to obtain a modified version of CHECKPNT from the InterSystems FTP site (ftp.intersys.com). The following versions of a fix are available:

October 30, 1998

MSM Flash - Memory Leak Problem With MSM 4.4.0 Running Under HP/UX 10.20

    There is a known memory leak in MSM-UNIX 4.4.0 under HP/UX 10.20. When a partition exits, a few kilobytes of memory are lost. This can cause MSM to rapidly exhaust all available system memory. In some cases, customers have experienced crashes due to this leak. In other cases, MSM is simply rendered non-functional. This memory can only be recovered by restarting MSM.

    We are currently working to release a version of MSM with the problem fixed. Until then, customers should only upgrade to MSM 4.4.0 for HP/UX if it is required for an operating system upgrade, or to eliminate a problem with a previous MSM version. If a customer upgrades to 4.4.0 they MUST install the patch described below.

    The patch, ftp://ftp.intersys.com/pub/msm/patches/unix/hp44mem.rs, recovers memory lost due to a memory leak in MSM-UNIX 4.4.0 (HP/UX 10.20 only). The patch file is an MSM %RS (routine save) format file containing a USEROPS routine. Those customers without a USEROPS routine present on their system may simply restore this routine into the manager UCI (usually MGR of the boot volume group.) If the customer currently has a USEROPS routine, insert the patch code at the label HP44FIX so it will execute as part of label ENDPART.

October 21, 1998

Caché 2.1.x and Digital UNIX 4.0D

    Caché hangs have been observed using Caché 2.1.x on DEC Unix 4.0D due to what is believed to be an OS related problem. InterSystems Corporation has identified the problem and has worked around it. If your DEC Unix 4.0D system is experiencing Caché hangs contact InterSystems Worldwide Response Center for a fix. The fix code for this problem is RAM739.
October 15, 1998

F.12 Service Pack 5 Release Info

    The release notes for the F.12 Service Pack 5 is now available. Service Pack 5 is a group of routines which should be installed on top of the released version of F.12. The installation is to simply load the routines from the Service Pack 5 file into the %SYS namespace (or the manager's UCI on DSM). Service Pack 5 includes all changes from Service Packs 1, 2, 3, & 4. If you install Service Pack 5, there is no need to install Service Packs 1, 2, 3, or 4.

    The list of changes in the Service Packs is described in the file. The file is in Microsoft Word 95 Format. Please download the release notes for more information on which service pack you will need.

    For Open M [ISM] 6.X UNIX Systems: ftp.intersys.com/pub/dbms/f12sp5_omunx_002.zip
    File should be loaded into %SYS with %RI.

    For DSM for OpenVMS Systems: ftp.intersys.com/pub/dbms/f12sp5_dsm_002.zip This file is to be used for DSM on OpenVMS systems. File has an autoload header, load by executing the first line in the file.

October 9. 1998

Update to DSM 7.x[.*] - SETs via DDP may corrupt remote volume sets

    A problem has been discovered with the fix available on our FTP server. A .com file used during the building of the image was not properly updated for VMS 7.*. The result of this is that the problem of data corruption is not corrected with images built using this procedure. The save sets on our FTP server have now been updated to correct this. The *.com file is only a problem for AlphaVMS 7.* systems. Save sets downloaded before 10:00AM today should be downloaded again if they will be used on AlphaVMS 7.* systems.
October 6, 1998

Update to DSM 7.x[.*] - SETs via DDP may corrupt remote volume sets

October 5, 1998

DSM 7.x[.*] - SETs via DDP may corrupt remote volume sets

    A serious problem with DSM V7.*[.*] releases that can lead to data corruption was detected this week and has been corrected as of today. The corruption takes the form of control and other characters being introduced into global data and in the After-Image Journal (AIJ) if configured. Since control characters are permitted in global data this problem cannot be detected via the ^IC or ^FASTIC integrity checkers, but will manifest itself as application errors or simple loss of data.

    The problem only occurs with SETs transmitted by DDP from DSM 7.x[.*] systems: if a DDP SET operation is attempted on a remote volume set that has been write-inhibited (even temporarily, as with an after-image journal file rollover) DSM will back off and then restart the operation. Starting with DSM V 7.*[.*] such a restart will result in DSM sending corrupted data to the remote system.

    A work around for some sites is to cluster-mount the problem volume set. This causes the SET to complete locally rather than via DDP. This is not a work around for volume sets that serve DSM V7.*[.*] systems that are not in the same VMS cluster. In such a case an alternate work around is to stop or suspend applications that access a remote volume set before write-inhibiting the remote volume set, including before any after-image journal file rollover.

    InterSystems engineers have built a variant of DSM V7.1.1 that corrects this problem. In addition a special utility is available to help identify global data nodes that may have been affected.

    Further information about obtaining the fix for your DSM V7.*[.*] system will be posted at here (http://www.intersys.com/services/support/flash/98announce.html) as it becomes available.

September 24, 1998

DSM 6.6D - Write Demon loops at COM 0 in a cluster.

    A problem has been discovered in DSM 6.6D that causes the write demon to loop in computable state at priority 0. This is only a problem in systems with cluster mounted database sets. This is a problem for 6.6D only and does not effect 7.x releases. A fix will be available soon.
September 17, 1998

<DATABASE> After <FILEFUL> Possible with Big Strings

    During random code inspection, InterSystems detected an opportunity for a <DATABASE> error to occur when attempting to set a long string (757-32K characters) into a global in a full database. We are not aware of this problem actually occuring at a customer site, but we do recommend that customers request the correction or avoid these conditions. A correction is available upon request to the Worldwide Response Center (support@intersys.com). Please be sure to supply the OrderNumber from the cache.key file or the complete version string.
September 17, 1998

Deleting registry keys & the Packet Driver

    If you have the Packet Driver installed on a Windows NT system, you CANNOT simply blow away the registry subkey for HKEY_LOCAL_MACHINE\Software\InterSystems in order to simulate a "clean system". If you do this, you'll get the "Registry subkey exists" error when you try to install it again.

    When it is installed, the Packet Driver also gets added to the network bindings and thus must be removed properly. If you need to remove the packet driver, always use the Network Control Panel applet to remove the driver. If you want to simulate a "clean system", either first remove the packet driver, or delete all subkeys in HKEY_LOCAL_MACHINE\Software\InterSystems EXCEPT for ISPKTDRV.

September 14, 1998

MSM - Unlock a protected global

    Title: How To: Unlock a protected global
    Platform: All MSM
    Author: SC - CM

    If all access to a global has been denied (which some utilities do - GBMAINT is an example), the database manager has to change the protection back to something reasonable using the %GCH utility from the boot manager UCI, using extended global reference to specify the global name.

    For example, if the global ^TEST, located in APP,SYS needs to be unlocked: 

     [MGR,CCM]>d ^%GCH

          MSM - Global Characteristics Utility

    Global selector: ^|APP,SYS|TEST

    Global selector: ^

    Select OPTION

     1    Collating Sequence
     2    Global Growth
     3    Protection
     4    Journaling
     5    Allow/Prevent KILL

Enter option for ^["APP","SYS"]TEST: 3  Protection

Current Status -->  System = None   World = None      Group = None      User = None

Select PROTECTION STATUS


     1    N   - None
     2    R   - Read
     3    RW  - Read/Write
     4    RWD - Read/Write/Delete

Enter protection for ^["APP","SYS"]TEST:
    Once you change the protection to something suitable to your configuartion, you should then be able to access the global.
August 28, 1998

Change in WRC for today

    Please be advised that InterSystems Worldwide Response Center (WRC) will be affected by a planned telephone switch hardware upgrade associated with a planned move within our building between approximately 1900 Hours Eastern Daylight Time (EDT) and 2000 Hours EDT on Friday, 28 August 1998.

    During this time period only, if you require assistance, please call the following alternative telephone number which will ring into the WRC's answering service: (617)-556-2514.

    You can continue to reach the WRC at all other times for all other calls at our normal telephone number of (617)-621-0700.

    Customers calling the Worldwide Response Center line (617)-621-0700 directly or through InterSystems Europe after 1700 EDT on 28 August 1998, other than the time of planned hardware outage discussed above, will encounter either: a WRC telephone line being answered by either the WRs Automated Call Director which, upon a message being left, will page a Customer Support Specialist to respond to your call; or, the WRS answering service which will then page and dispatch a Customer Support Specialist. This is our normal operations procedure.

    It is our expectation, other than the arrangements discussed above, that WRC service will be otherwise unaffected.

August 28, 1998

MSM 4.4.0x critical patch available

    P12765

    This patch corrects three errors in MSM-Server, Version 4.4.0 and 4.4.0a. Two of them involve adding database volumes to an existing volume group. First, if the file name of the volume being created was longer than 64 characters, it was not handled properly, leading to errors during creation and mounting (PR012729). The second, and more significant issue, resulted in <DKHER> errors in the newly created volume. In addition, in some cases, one block in an existing volume could be corrupted (PR012765). Finally, if the SysAdmin tool was used to expand a volume group with BIJ enabled, the expand would work, but a <NOPGM> error would occur at the end, and BIJ would be disabled (PR012785).

    We strongly recommend that this patch be installed on ALL MSM Version 4.4.0 and 4.4.0a systems.

    The fix is located in:

    ftp://ftp.intersys.com/pub/msm/patches/DB40FIX.zip

August 27, 1998

DSM V7.0, V7.1 and V7.1.1 - Problem enabling the "Disable Process Deletion" Option.

    DSM V7.0 introduced a new option that allows you to prevent the deletion of DSM processes by use of the DCL STOP/ID command. This feature can be enabled via the ^CONFIG utility at the following prompt:

    Disable process deletion by STOP/ID [Y OR N] ?

    InterSystems recommends that this new feature should NOT be enabled until further notice (I.E. you should answer 'N').

    It has been discovered that under certain circumstances, it is possible to get DSM processes going into a COM 0 state consuming high levels of CPU time when this feature is enabled for a configuration. InterSystems is currently investigating this issue and will update you when the problem is corrected.

August 6, 1998

SysAdmin Tool 1.0 for MSM Update

    There is an issue with the new SysAdmin tool v1.0 distributed from the MSM web page and on the CD. If you attempt to look at or modify (or just view without changing) a configuration on a server and the configuration has a tied terminal table; that table will be deleted.

    The routine(s) required to correct the problem are MSACONNT.RTN (for Windows-NT & 95) and MSACONAIX.RTN (for all UNIX platforms). The files are located in the MSM directory on the FTP server.

    ftp://ftp.intersys.com/pub/msm/

    The routine is in %RS (routine save) format and is to be restored onto each SERVER that is to be administered using the SysAdmin tool.

June 17, 1998

DSM and OpenVMS 7.1 with patch ALPSYS12_071 Update

    In a previous newsflash InterSystems recommended installation of DEC VMS patch ALPSYS12_071 for those customers running VMS 7.1 and DSM. The current version of the patch ALPSYSA_071 fixes the access violation problem. It is available now from Digital.
May 12, 1998

Caché and Digital UNIX 4.0b Issue

    InterSystems Corporation has uncovered a Digital Unix bug with asynchronous I/O operations on Alpha platforms. This bug was discovered on a system running DEC Unix 4.0B and has been fixed in version 4.0D. The effect on Caché has been Write Daemon hangs in version 2.1.3 and possibly other versions. InterSystems recommends all sites running Alpha and Digital UNIX to upgrade to version 4.0D as soon as possible. The DEC tracking number for this problem is C980408-2331.
May 11, 1998

MSQL Service Pack 1 for F.12

    Service Pack 1 is a group of routines which should be installed on top of the released version of F.12. The installation is to simply load the routines from the Service Pack 1 file into the %SYS namespace (or the manager's UCI on DSM).

    If you would like to download the service pack, it can be found on our ftp server at:

    ftp://ftp.intersys.com/pub/dbms/f12sp1_dsm_004.rtn for DSM systems

    ftp://ftp.intersys.com/pub/dbms/f12sp1_wnt_004.rtn for Caché systems

    Below is a list of changes that comprise the F.12 Service Pack 1. If any part of the change modifies a database object like a table or form, there will be a Y in the OC column of the table. This indicates that part of this change may only be accessible via predefined entry points described below.

    Changes in F.12 Service Pack 1 Build 004 [08 May 1998] 

    
Description                                    			OC


SERVER: Optimize Relational Server For DSM 
SERVER: Vendorize TCP Benchmark Tool
SERVER: Alter Benchmark For TR Bug...
SERVER: Benchmark for Old/New Relational Server
Fix SELECT V1.Alias, V2.Alias Join Problem
Fix Problem with GENERAL VIEW ALIAS...
Return Proper FileMan Date/Time Value to the Client
Fix Directory Reference Problems 
Fix Problem with Microsoft Access INSERTs
Fix Server Debugging Code...
Fix Minor Problem with OUTER JOIN..
Patch COMMIT for Omnis7 & TP to Work Properly                Y see # 1 below
Fix problems with NULL comparison & OR 
Reinstate a MRP700 change for IN as per original design
Fix redundant CR LF for Multi-Line Fields  
Allow Exclusion of Fields from SELECT * 	  	     Y see # 2 below
Fix FileMan Date/Time Stamp for Client/Server...
Add Encryption & Decryption Algorithms to New Server
Fix Multi-Condition Problem for Index Option Fld
Extensions to DDL for Omnis7 support 
Handle Integers better through DDL for ODBC
Allow fields with name of table in DDL
Fix bug in Data Import Utility w/User field order
Support Owner.Catalog.Schema.TableName in DDL parser
Fix Dropping of 4 Characters from Generated Query text
Fix  run time in query with ALPHAUP(name) LIKE var
Fix Client returning dates in 1800's
Fix SQL INSERT/UPDATE to not consider ODBC excluded columns  Y see #3 below
FIX  %qaperr ERROR REPORTING

The following entry points are interactive methods to either initialize 
database objects with new values, or to simulate user interface actions 
to perform specific actions:

#1  >d sp1^%qarsp1             ; Should be executed in the  Manager's namespace

# 2  >s ok=$$nostar^%qarsp1(ifn)   ; Returns a 1 for success or 0 for
                                           failure. Simulates the Field 
                                           Definition question: Exclude 
                                           Field from SELECT * ?

# 3  >s ok=$$nolookup^%qarsp1(ifn) ; Returns a 1 for success or 0 for
                                           failure.  Simulates the  Field
                                           Definition question: Do not
                                           include in lookup lists?
May 8, 1998

Multiple SQL joins with Crystal Report Writer 6.0

    SeaGate Software has made a patch available for Crystal Report Writer 6.0 so that it will recognize The correct level of SQL support on the server. After applying this patch it is possible to write reports that join more than two tables when you are querying a Caché database.

    The file can be downloaded from SeaGate's web site:

    http://www.seagatesoftware.com/crystalreports/updates/

    Click on patches section and then select the zip file pdsodbc.zip. Unzip this file on your PC that has Crystal install and place pdsodbc.dll and p2sodbc.dll in the windows/system directory, overwriting the dll that where loaded by the Crystal install.

    If you have questions and problem with this please contact InterSystems Technical Support at (617) 621-0700 or support@intersys.com

May 7, 1998

PowerBuilder connections to Caché via ODBC

    Several issues have been discovered and corrected involving PowerBuilder and Caché. If you try to connect PowerBuilder 6.0 via ODBC to Caché 2.1.5 or 2.1.6 the connection would fail. These problems have been corrected in the latest version of the Relational Client (2.2-0012)and the Caché SQL Server (Version 6).

    The Relational Client is for any Window 95 or Intel NT system.

    The Caché SQL Server is for F.12 running on Caché only.

    If you would like to connect PowerBuilder 6.0 to Caché via ODBC you will have to down load two files from out ftp server:

    ftp://ftp.intersys.com/pub/cache/qsnF12V6.zip

    ftp://ftp.intersys.com/pub/cache/client/RC22_12.zip

    qsnF12V6.zip contains a read me file that has directions on what you need to do to load the qsnstart server and get it running.

    RC22_12.zip contains the Relational Client EXE that you need to load on the client PC.

    If you have any question or problems, please contact InterSystems Technical Support at: 617 621-0700 or support@intersys.com

May 5, 1998

DSM and OpenVMS 7.1 with patch ALPSYS12_071

    In a previous newsflash InterSystems recommended installation of DEC VMS patch ALPSYS12_071 for those customers running VMS 7.1 and DSM. This patch is currently on engineering hold and has been superseded by ALPSYS16_071 which will possibly soon be superseded by ALPSYS18_071. From what we have been able to gather, neither of these patches contain the relevant fix as described in the ECO KIT SUMMARY for ALPSYS12_071:

    "The symptom of this problem is an access violation at EXE$AST_RETURN,
    in an outer mode - most typically user mode - but with the FP pointing
    to the kernel stack. The access violation occurs trying to access data
    on the kernel stack from user mode. This doesn't crash the system, but
    causes the user image to exit."

    The symptom seen from a DSM point of view is a process takes an access violation and goes into a computable loop possibly while holding exclusive access to a volume set. The result is that any process trying to access that volume set will hang.

    In most cases there will be no evidence in DSMCRASH.LOG since the acc vio takes place in VMS not DSM code.

    Any DSM version compatible with VMS 7.1 is at risk of this problem (DSM 6.6B and above).

May 4, 1998

OpenVMS 7.1 patch needed for DSM

    There is a bug in OpenVMS 7.1 that can cause DSM to access violate which in turn could hang DSM.

    It is advised to if you are using OpenVMS 7.1 to apply the patch ALPSYS12_071 which is available from Digital. Be aware that this patch may be superseded by more recent ones.

April 28, 1998

DSM 6.6D

    DSM 6.6D will be available on May 4, 1998.
    This release fixes three important bugs with DSM.

    1) DSM may hang when trying to inhibit a volume set with ^VOLINH.
    Introduced in DSM 6.5

    2) DSM may fail to properly roll back BIJ transactions during recovery leading to database degrade. See March 24 newsflash.
    Introduced in DSM 6.5.

    3) A DSM process may go into HIB state and never return while holding exclusive access to a volume set. This results in the volume set being hung.
    This is a rare but long-standing problem.

April 27, 1998

DSM 7.0 Issue

    Using greater than 11,000 global buffers can lead to access violations and truncation of data being read from globals.
    This is fixed in DSM 7.1. 

    Example:

>S $P(Y,"O",300)="" 

>F X=1:1:20 S ^ZZZ(2,X)=Y

>F X=1:1:20 W !,$L(^ZZZ(2,X))

299
299
255
299
299
255
299
299
255
299
299
255
299
299
255
299
299
255
299
299
April 10, 1998

Caché and insufficient routine buffers issue

    On systems which are configured for greater than 255 processes, and less than 256 routine buffers, Caché hangs have occured under special circumstances.

    In such configurations, if the system runs out of routine buffers, more unique routines are accessed at once than the number of routine buffers, Caché hangs could result.

    This problem will be fixed in the future releases of Caché.

March 27, 1998

DSM Updates through Digital

    Attention users who receive DSM Updates through Digital's Consolidated Distribution (ConDist) CD's. There is a TCP/IP bug in the release of DSM 6.6C on the ConDist CD's. Please contact Intersystems Order/Processing to receive the corrected copy. The problem is that with TCP/IP communications, a fixed length read greater than 1 will not be recognized. The read will instead read the entire buffer.
March 24, 1998

DSM 6.5 and above Write Demon Issue

    InterSystems has discovered a issue with DSM 6.5 and above that could leave DSM databases in a degraded state should recovery be necessary. A 6.6D release will be available in the near future, in the meantime a corrected 6.6C image, routines and installation instructions are available at ftp.intersys.com/pub/openm/dsm/recovfix.bck (you may need to hold the Shift Key down while clicking the link to force a save file). A description of the problem follows:

    When the DSM Write Demon process is under extreme load and DSM is shutdown in an abnormal way such as a VMS crash or a Stop/ID of the Write Demon, recovery may fail to rollback BIJ transactions. This presents itself as the message that 0 BIJ transactions have been recovered. If a recovery situation is encountered where the 0 BIJ transactions are recovered and it is suspected that the system should have had something to rollback then a ^FASTIC is in order to determine if the database has been compromised.

February 10, 1998

DSM 7.0 installation issue

    A directory SYS$SPECIFIC:[DSM$LICENSE] with protection:

    s:rwe,o:rwe,g:re,w:re

    must exist prior starting the DSM 7.0 installation.

February 5, 1998

$ZDATE changed

    $ZDATE changed in a subtle way as Sliding-Window Date support was introduced in [ISM] 6.3-F.11 such that $ZDATE(null) and $ZDATE(0) return "12/31/1840" rather than null-string as in prior releases. Some customers have found this to be undesirable. A software modification is available to customers that require the previous behavior. Please contact your supplier of [ISM] to request this modification. [ISM] release 6.4-F.12 will restore the prior behavior such that $ZDATE("") and $ZD(0) return the null-string.
January 9, 1998

Fix available for <MAXARRAY> error encountered when using ^BLKDIST in directory with large numbers of globals.

January 7, 1998

Home | M Technologies | Support | Company | Contact

MOVE TO CACHÉ

© Copyright 1996-2000 InterSystems Corporation. All Rights Reserved.
email: wwwadmin@intersys.com