HP OpenVMS Systems Documentation

This chapter describes problems resolved by software changes made in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server). These software changes will also be included in future ECOs and in PATHWORKS V6.1 for OpenVMS (Advanced Server). The problems are describd in the sections indicated in the following table:

This section describes the file server problems corrected in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.1.1 PWRK$LMSRV Crashes During Client Operation Invoking NetLogon

Problem:

When a client performs an operation that invokes the NetLogon service (such as when a user attempts to log on to the server), and the client sends the wrong pipe name, the server crashes in module OPEN, routine file_exists. A traceback similar to the following would be seen in the PWRK$LMSRV log file:

...
%SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual
address=00000021, PC=002EDF6F, PSL=03C00000
%TRACE-F-TRACEBACK, symbolic stack dump follows
module name     routine name              ...   rel PC    abs PC

                                          ...   002EDF6F  002EDF6F
OPEN            file_exists               ...   000000C7  000872FF
OPEN            NT_openX_open2            ...   00000338  00088694
NTCREATE        smbntcreateX              ...   000001C4  0009F1B0
SMBWORK         smbwork                   ...   00000DE8  0007ECA4
                                          ...   002BB885  002BB885
MTS$MAIN        main                      ...   00000029  00029995

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.1.2 BDC Servers Experience High CPU Utilization and Replication Traffic

Problem:

The PWRK$LMSRV process on a backup domain controller (BDC) server might consume a high percentage of the CPU time during replication of domain security (SAM) databases. In this case, the replication traffic is excessive, and the server event log includes unexpected entries 5716, 5717, 5718, and 5719, depending on the replication process being performed.

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.1.3 Using OpenVMS Security Mode, READALL Privilege Is Ignored

Problem:

When the PATHWORKS server is configured to use the Advanced Server and OpenVMS security model, the OpenVMS READALL privilege is ignored.

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.1.4 Client Directory Listings Incomplete

Problem:

When viewed from various clients, such as Microsoft Windows Explorer, directory listings might fail to show all the files in the directory. In some cases, the missing contents are those files that would be at the end of the directory listing. The directory cache might inadvertently stop before the entire cache contents for the directory are processed. This seems to occur most often on heavily accessed directories. For example, the problem might occur when a substantial number of subdirectories and files are added to the same directory. The problem might also occur with directories that are accessed by multiple clients. After the contents of a directory that has been cached are changed (for example, from OpenVMS or another client, or after a ZIP file is unpacked), the directory listing might be incomplete.

This symptom is similar to that described in Section 4.2.11, Incomplete Directory Listings, but the underlying problem differs.

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server). Note that incomplete directory listings can also be caused by XQP problems. To address these problems, the latest XQP related patches must be installed. At a minimum, the following patch levels should be installed:

If your Advanced Server participates in an OpenVMS Cluster, these patches must be installed on all cluster members, even those that do not run the Advanced Server.

5.1.5 SAM Database Becomes Corrupted After Shutdown

Problem:

If the server shuts down while updates to the SAM database are in progress, the SAM database might become corrupted, especially when update activity is high, such as during a full replication or when using scripts to add or remove a large number of users.

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server). Various ECOs, patch kits, and recommendations from Compaq, for earlier releases of the Advanced Server, provided a fix for this problem that interlocked the PWRK$SHUTDOWN.COM procedure with stopping the NetLogon service. The new fix provided with the PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server) does not include the interlocking of NetLogon in PWRK$SHUTDOWN.COM. It is no longer necessary to interlock stopping the NetLogon Service with PWRK$SHUTDOWN.COM. Any such customized changes can be removed.

5.1.6 SAM Corruption Issues Addressed

Problem:

Previously, it was possible for the SAM database to become corrupt in other situations other than that discussed in the preceding note. The causes were many and various, but always resulted in similar symptoms, which included the following:

• SAM database replication would fail during partial synchronizations or full synchronizations, or both
• Attempts to issue any of the following ADMINISTER commands while the server is running fail, with the "LM-E-ERROR_GEN_FAILU general failure" error message being displayed:
  • SHOW COMPUTERS
  • SHOW GROUPS/FULL
  • SHOW TRUSTS
  • SHOW USERS/FULL
• When the server attempts to access any objects stored in the SAM database (such as computers, groups, trusts, and users), it crashes. A traceback similar to the following would be seen in the PWRK$LMSRV log file:

  image    module    routine   . . .   rel PC          abs PC
PWRK$LMRPCAPISHR  SERTL  RtlLengthSecurityDescriptor
                               . . . 0000000000000BD8 0000000000730978
PWRK$LMSRV  SECDESCR  PegSamrSetSecurityObject
                               . . . 0000000000001904 00000000001FDF24
PWRK$LMSRV  SAMRPC  SamrSetSecurityObject
                               . . . 00000000000000F8 00000000000D9AD8
 .
 .
 .

• Even if the database is corrupted, the preceding symptoms still might not appear. To test the database, enter the following SAMCHECK command while the server is not running. (Note that SAMCHECK is an unsupported tool.) This example shows the expected output for a sound database.

  $ SAMCHECK -s No corruption was detected in the SAM database.

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server). Note that it is still possible for SAM corruption to occur. The most common cause is improperly shutting down the Advanced Server. To avoid SAM corruption, always use the SYS$STARTUP:PWRK$SHUTDOWN.COM shutdown procedure whenever stopping the Advanced Server and also prior to shutting down the OpenVMS system.

Various ECOs, patch kits, and recommendations from Compaq, for earlier releases of the Advanced Server, provided a fix to prevent the SAM corruption problem resulting from improperly shutting down the Advanced Server. The fix interlocked the PWRK$SHUTDOWN.COM procedure with stopping the NetLogon service. The new fix provided with the Chapter 5 does not include the interlocking of NetLogon in PWRK$SHUTDOWN.COM. It is no longer necessary to interlock stopping the NetLogon Service with PWRK$SHUTDOWN.COM. Any such customized changes can be removed.

5.1.7 Server Crashes with Access Violation in DirRecycleDirectory

Problem:

The file server might crash due to an ACCVIO while a directory is being accessed by more than one client. For example, it might crash in the DIRRecycleDirectory routine of the ODS2$DIRCACHE module. A traceback similar to the following would be seen in the PWRK$LMSRV log file:

  image    module    routine    . . .   rel PC          abs PC
PWRK$FSLIB_ODS2  ODS2$DIRCACHE  DIRRecycleDirectory
                                     0000000000002928 000000000A9A9F8
PWRK$FSLIB_ODS2  ODS2$DIRCACHE  DIRMaintenance
                                     0000000000003F40 0000000000A9C010
 .
 .
 .

In addition, one or more "dce on purgatory queue without bit lit" messages might be logged in the PWRK$LMSRV log file.

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.1.8 Server Crashes in ODS2NewASUSecurityAce

Problem:

The server might crash due to an ACCVIO in module ODS2$LIB, routine ODS2NewASUSecurityAce. A traceback similar to the following would be seen in the PWRK$LMSRV log file:

  image    module    routine    . . .   rel PC          abs PC
PWRK$FSLIB_ODS2  ODS2$LIB       ODS2NewASUSecurityAce
                                     00000000000026AC 0000000000A0AF7C
PWRK$FSLIB_ODS2  ODS2$LIB       ODS2SetASUSecurity
                                     000000000000294C 0000000000A0B21C
 .
 .
 .

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.1.9 PWRK$LMSRV Crashes in Routine ssignon_gethostmap

Problem:

When using external authentication with the Advanced Server, the PWRK$LMSRV.EXE crashes. A traceback similar to the following would be seen in the PWRK$LMSRV log file:

  image    module    routine    line   abs PC
PWRK$LMSRV  SSIGNON_PROC  ssignon_gethostmap
                               88482   0000000000000984
PWRK$LMSRV  SSIGNON_PROC  process_acme_message
                               88828   00000000000019DC
PWRK$LMSRV  SSIGNON_COM   receive_comp_handler
                               23651   0000000000000A58
PWRK$LMSRV  SSIGNON_COM   ssignon_work   24150  0000000000001828
 .
 .
 .

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.1.10 Client Application Displays "Device-specific Error Has Occurred"

Problem:

Seagate Backup Exec gets the message "a device-specific error has occurred" when backing up to a PATHWORKS share. This problem occurs because an OpenVMS path specification, rather than an MS-DOS path specification, is being returned in response to the "Transact 2 Query File Info" message. This problem might also occur with other client applications.

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.1.11 PWRK$LMSRV Process crash in RPCBIND/rpc__binding_free

Problem:

The PWRK$LMSRV process might crash in image PWRK$LMRPCSHR with a process traceback that shows an ACCVIO in RPCBIND/rpc__binding_free, called from CNSRV AbortAssociation, following a series of IPC error messages regarding a full or closing pipe such as:

IPCerr\RPC-56B8\CLOSE.3129 :#373717 [invalid pipe end) failure writing
EOF
   Status = 2264 (0x8D8), %SYSTEM-W-MBFULL, mailbox is full
IPCerr\RPC-56B8\freepd.3332 can't free write IOSB 0x081CCC18
IPCerr\RPC-3778\SEND.2320 }#2929674 [invalid pipe end)  sendmsg error

These error messages might also be seen without a corresponding PWRK$LMSRV crash.

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.1.12 PWRK$LMSRV Process Crashes in Routine CheckMsgAuthenticator

Problem:

The server might crash with an ACCVIO in routine CheckMsgAuthenticator. A traceback similar to the following would be seen in the PWRK$LMSRV log file:

  image    module    routine   line    . . .   abs PC
PWRK$LMSRV  SSIGNON_PROC  CheckMsgAuthenticator
                               108897 0000000000004698
PWRK$LMSRV  SSIGNON_PROC  process_acme_message
                               106887 0000000000001490
PWRK$LMSRV  SSIGNON_COM   receive_comp_handler
                               23651   0000000000000A58
PWRK$LMSRV  SSIGNON_COM   ssignon_work   24150  0000000000001828
 .
 .
 .

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.1.13 BDC Servers Experience High CPU Utilization and Replication Traffic, or ADMINISTER Commands Fail with "RPC Server Is Unavailable" Errors

Problem:

One or both of two problems might occur:

• As reported in Section 5.1.2, BDC Servers Experience High CPU Utilization and Replication Traffic, the PWRK$LMSRV process on a backup domain controller (BDC) server might consume a high percentage of the CPU time during replication of domain security (SAM) databases. In this case, the replication traffic is excessive, and the server event log includes unexpected entries 5716, 5717, 5718, and 5719, depending on the replication process being performed.
• ADMINISTER commands fail with the following error message:

  -LM-E-RPC_S_SERVER_UN, The RPC server is unavailable

  
  Note that this error message does not necessarily indicate the problem being documented here; it might also appear when the RPC server is indeed unavailable.

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.1.14 PWRK$LMSRV Process Crashes in Module THREAD

Problem:

The Advanced Server might crash with the following error messages and traceback in the PWRK$LMSRV log file:

%SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual
address=0000000000000030, PC=000000000041B560, PS=0000001B
%TRACE-F-TRACEBACK, symbolic stack dump follows
image           module  routine          line   rel PC
PWRK$CSSHR_V7   THREAD  resume_0         16861 0000000000008520
PWRK$CSSHR_V7   THREAD  threadresumelast
                                         12759 0000000000003E0C
PWRK$CSSHR_V7   WORKQ   WorkQ_queue_work_safe
                                          9556 0000000000000DC8
PWRK$CSSHR_V7   WORKQ   WorkQ_queue_work
                                          9453 0000000000000B0C
PWRK$LMTASKSHR  QUEUE   queue_put        14706 00000000000000E4
PWRK$CSSHR_V7   LIBIPC  mbreaddoneast
                                         32735 0000000000008EC4

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.2 File Access/Printing Problems

This section describes file access and printing problems corrected in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.2.1 Windows 2000 Backup Fails to Advanced Server Share

Problem:

A Windows 2000 backup fails when attempting to write to an Advanced Server share.

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.2.2 Files Remain Open

Problem:

When a user with less than Full Control Access to a file views the security on that file, the file might be left in an "open" state. Subsequent attempts to delete or rename the file would fail.

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.3 Transport Problems

This section describes problems related to the use of transport protocols with the Advanced Server, corrected in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.3.1 Call to KNBNQ_UNSET_TIMER in the PWRK$LMMCP Process Might Cause OpenVMS to Crash

Problem:

While in PWRK$LMMCP, during a call to the KNBNQ_UNSET_TIMER routine, the server might cause OpenVMS to crash with the following system bugcheck information:

Bugcheck Type: INVEXCEPTN, Exception while above ASTDEL
Current Process: PWRK$LMMCP Current Image:
AXPA$DKC0:[SYS0.SYSCOMMON.][SYSEXE]PWRK$LMMCP.EXE
Failing PC: FFFFFFFF.884386D0 KNBNQ_UNSET_TIMER_C+00198
Failing PS: 00000000.00000804
Module: PWRK$STREAMSOS_V7 Offset: 0006E6D0

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.3.2 System Hangs or Crashes with a Spinwait Bugcheck

Problem:

When the server is not able to send out a message over its transport, it will start a retry mechanism. This retry mechanism could cause the OpenVMS operating system to become overloaded with high IPL fork requests, causing the system to hang or crash in PWRK$STREAMSOS with a CPU spinwait bugcheck.

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.3.3 System Crashes in Routine RcvUDSthand

Problem:

The system might crash due to an ACCVIO in routine RcvUDSthand of module STREAMSUB in the sharable library PWRK$SHREAMSSHR, with the following bugcheck:

Bugcheck Type: SSRVEXCEPT, Unexpected system service exception
VMS Version: V7.1-1H1
Current Process: PWRK$LMBROWSER
Current Image: $1$DKB0:[SYS0.SYSCOMMON.][SYSEXE]PWRK$LMBROWSER.EXE;5
Failing PC: FFFFFFFF.800086E8 OTS$MOVE_C+000B8
Failing PS: 10000000.00000003 Module:
SYS$BASE_IMAGE Offset: 000066E8

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.4 Event Logging Problems

This section describes problems related to event logging functions of the server, corrected in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.4.1 Client Event Logs Fill with Many Event 3012 Errors

Problem:

Some client applications may cause the event log on Windows NT clients to become full with Event 3012 entries when the user attempts to open files with names greater in length than the MS-DOS standard 8.3 format.

A status returned by PATHWORKS was more severe than necessary which caused the client to log an error in its event log.

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.5 Cluster-Related Problems

This section describes problems relating to servers participating in OpenVMS Clusters. These problems have been corrected in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.5.1 Multiple Servers Perform Replication in a Cluster

Problem:

In a cluster, only one cluster member should actually participate in replication of the SAM databases --- the server that holds the PWRK$DMN_UAS lock. In some circumstances, no cluster member holds the lock, which results in all cluster members participating in replication. When multiple members attempt to replicate the database simultaneously, the replication fails.

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server).

5.5.2 BDC Loses Connection to PDC; Unable to Participate in Replication

Problem:

In WAN environments in which the Advanced Server runs in a cluster as a Backup Domain Controller (BDC), and the primary domain controller (PDC) is on a different LAN segment than that of the cluster BDC, the remote PDC might resolve the cluster alias to the wrong cluster BDC member. As a result, the cluster BDC is unable to participate in replication of the SAM database.

The system manager might notice user account modifications are not being replicated to the cluster BDC. Users might notice password updates do not take effect. Partial synchronization messages for the cluster BDC might be missing from the event log.

The cause of the problem can be explained as follows: in such environments, prior to PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server), it was necessary to ensure that a WINS or LMHOSTS entry for the Advanced Server cluster alias points to the cluster member with the active daemon (typically the member where the server was first started; it is the member that holds the PWRK$DMN_UAS lock, and so is the only member that can process replication operations). If the WINS or LMHOSTS entry is not set to the cluster member with the active daemon, the remote PDC will resolve the cluster alias to the wrong cluster BDC member, preventing the cluster from being able to participate in replication.

Solution:

This problem is resolved in PATHWORKS V6.0C-ECO2 for OpenVMS (Advanced Server) and PATHWORKS V6.0D for OpenVMS (Advanced Server). It is still necessary to define an LMHOSTS entry for the Advanced Server cluster alias; however, that entry can point to any cluster member that is running the NetLogon service, whether the member has an active or inactive daemon. When the PDC announces SAM database updates (by means of a pulse message), if that announcement gets to a cluster node with an inactive daemon, the message is forwarded to the cluster member with the active daemon. Then, replication will take place appropriately.