Patch Name: PHSS_30994 Patch Description: s700_800 11.X OV ITO7.1X Msg/Act Agt HPUX 11 IA A.07.25 Creation Date: 04/07/09 Post Date: 04/07/12 Hardware Platforms - OS Releases: s700: 11.00 11.11 s800: 11.00 11.11 Products: OpenView Operations 7.1 OpenView Operations 8.0 Filesets: OVOPC-CLT.OVOPC-UXIA-CLT,fr=A.07.10,fa=HP-UX_B.11.00_32/64,v=HP OVOPC-CLT.OVOPC-UXIA-CLT,fr=A.07.10,fa=HP-UX_B.11.11_32/64,v=HP Automatic Reboot?: No Status: General Release Critical: Yes PHSS_30994: ABORT PHSS_30169: HANG ABORT CORRUPTION MEMORY_LEAK PHSS_29354: HANG ABORT CORRUPTION MEMORY_LEAK PHSS_28960: HANG ABORT CORRUPTION MEMORY_LEAK Category Tags: defect_repair general_release critical halts_system corruption memory_leak Path Name: /hp-ux_patches/s700_800/11.X/PHSS_30994 Symptoms: PHSS_30994: Change Request: 8606353730 opcmsga may log OpC20-184 warning in opcerror while starting the agent at anytime (even if there was no message removed): 03/03/04 17:36:47 WARNING opcmsga (Message Agent)(2592) [genmsga.c:789]: Removing messages with lower severity from buffer file successful. (OpC20-184) Change Request: H555012193 Logging to opcerror file may fail if a process has more than 256 file descriptors open due to a bug in operating system. Change Request: 8606346413 The manual agent installation with opc_inst does not update the component version numbers in the agents opcinfo file and does not set the management server name automatically. Change Request: H555011117, H555013386 OVO8 HTTPS based agent are overwritten by OVO7 agent installation. Change Request: H555011875 opcctla core dumps during agent registration (opcagtreg -add regfile.reg) if AGENT ID is the same as the one of already registered agent. Change Request: 8606348297 opcnetchk is designed to test basic types of communication like TCP, ICMP and SNMP in one step without having the possibility to select/test each type of communication individually. Change Request: B555030766 If an agent is running as a non-root user (changed with opcswitchuser.sh), opceca and other MSI programs that are running as agent user cannot connect to the MSI. Change Request: B555025991 Applications which are started as opc_op user are started with LANG=C per default. E.g. if you start X-OVw from a japanese OVO MotifGUI then an english ovw is started instead of a japanese one. Change Request: R555026293 In case the agent is running in cluster aware mode, the distributed ServiceDiscovery policies do not get updated during runtime, but it is necessary to restart the agent to activate the policies. Change Request: B555032445 Usually, all sub-agents are terminated if the node type is changed from 'controlled' to 'unmanaged'. All the sub-agents are not restarted if the "unmanaged" node is changed back to "controlled". Change Request: B555028360 opcapm reports error opc30-3287 with an empty cluster name. Change Request: H555011381 The Message Agent may occasionally abort on a fast multiprocessor system. Change Request: B555029983 There is a typo within error message OpC30-1139 (operaition instead operation). Change Request: 8606355743 opctemplate exits with this error on nodes using the RPC_DCE_UDP communication type: Can't convert location: Protocol sequence not supported (dce / rpc). (OpC20-122) Change Request: B555022224 State-Based Advanced Monitoring of CODA values does not send Msgs to reflect the current state. Change Request: B555032246 The message storm circuit may not detect a second message storm and later on, you may see that opcecm even aborts. Change Request: B555017581 The monitor agent may produce very large error messages each time policies are redeployed. Change Request: B555023880 Some Coda Policies generate an operator initiated action to show the Fileystem graph. This action sometimes fails, and displays: Error in generating report No data found in the selected date range Change Request: B555028060 opcmona places the OVO paths at end of $PATH instead of in front which causes a delay in application starts if the PATH contains NFS mounted directories. Change Request: B555027814 The agent could not send proxied messages for switchable addresses in a cluster environment. Change Request: 8606359109 opcnetchk -p does not display info for each configured port in opcinfo Change Request: H555011688 Messages are sent with wrong ip address from agent to mgmt_sv. Messages are forwarded with the ip address specified in /etc/hosts instead of using the value in OPC_IP_ADDRESS. PHSS_30169: Change Request: R555026193 An Advanced Monitor only sends one message per threshold the first time the threshold is reached in a monitor template with multiple conditions. Change Request: H555010571 The Message agent could die when sending messages in a NAT environment. Change Request: R555026123 The monitor agent may not do threshold checking on multiple source policies in case there are no instance names that match and the policy is configured to process all instances separately. Change Request: 8606321177 The size of the buffer file msgagtdf is not set to zero, nor is msgagtdf deleted, after connection to the management server is re-established and agents are stopped and restarted. Change Request: 8606310908 If a logfile size is increasing rapidly, faster than opcle can process it, duplicate messages may not be suppressed. Change Request: B555015967 The control agent may restart the service discovery or llbserver sub-agent endless in case they abort. Change Request: B555026668 <$MSG_OBJECT> is never replaced with the Object set in the template condition that is producing the message. Change Request: B555015548 The built-in ECS annotation server (opcecaas/opcecmas) may start looping or even abort. Change Request: B555021669, B555024574 opctrapi/opcevti may fall behind during storms of traps, and as a result it is disconnected from pmd. Change Request: B555020612 Duplicate suppressing is inconsistent, when using a mix of counter/timer or when default settings are set and more settings are added in the conditions. Change Request: B555022945 Distribution of monitors fails with OpC20-65/66 if system resources are low and memory allocation fails for the monitor file. Change Request: 8606298307 If large amounts of messages that have a long original message text are forwarded to OVOW (for example messages from WMI: WINOSSPI-ADS_DomainChange), then the OVOW Console memory usage grows. Change Request: H555011662 A forked process might abort if tracing is enabled although OPC_TRACE_CHILD is NOT set in the opcinfo file. Change Request: B555018732 You may receive the error "Illegal value( function ) (OpC10-13)" in the opcerror file reported by the process opcecaas or opcecmas. Change Request: 8606322495 opctemplate returns always exit code zero, even if it fails or a template does not exist. Change Request: 8606323486 opcswitchuser.sh only works if the specified group is the primary group of the specified user. All other groups are rejected: "ERROR: Group ... does not exist". Change Request: R555022489 opcle may generate garbled messages in Japanese environments. Change Request: B555018987 opcagt -status and opctemplate may erroneously report: Error opctemplate(24470) : opcctla not running on node . (OpC30-3002) Cannot connect to RPC service at system 'ncacn_ip_tcp:[]'. Local port configuration has been consulted - rpcd/llbd on remote system not queried. (OpC20-186) Change Request: 8606325445, B555021772 opctemplate help information is incorrect. opctemplate - enable is shown in the help, but the correct command is opctemplate -e. Change Request: B555023264 <$MSG_OBJECT> is never replaced with the Object set in the template condition that is producing the message. Change Request: 8606333207, 8606355920 The Logfile Encapsulator may lose messages when monitoring files that periodically get renamed. Change Request: H555010925 The following error message appears in opcerror file: >> Cannot retrieve value of type 'MDT_UINT32' in column '4' of interval '0'(OpC30-762) << Change Request: B555018953 In trap conditions, $MSG_NODE_NAME is replaced by the trap sender's name, not the trap's agent address name. These names can differ in a proxied environment. Change Request: H555008553 itochecker was only able to collect information from the management server. There was no possibility to collect information from the managed nodes. Checking the kernel parameters did not return the correct values. No information about opcerror was gathered by itochecker. Change Request: B555018506 Want a command to return the communication type of agent. Change Request: B555025523 Enhance opcagt for coexistence with future versions of components. Change Request: B555023808, B555027712 opcmona does not replace <$MSG_OBJECT> properly. Change Request: H555011234 opcacta often reports a mutex error (OpC20-1055) and a wait(2) error (OpC20-413). After the error, opcacta is running and can execute the request. But it cannot return the action result to the requester. Therefore OVO operators cannot receive the action result. Change Request: R555025398 The logfile encapsulator loses one file handle per configured interval, if all of the following conditions are met: - the logfile name is dynamically evaluated via a script - the script returns file names that cannot be opened - the logfile template defines a 'file to be read' Under these conditions the 'file to be read' will be opened over and over. Change Request: B555019025 opcswitchuser.sh allows for a non existent group to be assigned. Change Request: B555025674, 8606341274 The OVO agent does not support HP-UX 11.23. Change Request: H555009905 If the agent is manually installed on the node, opc_inst does not cleanup after itself once the installation is completed. The installation packages remain in the installation directory as does the opc_inst script. PHSS_29354: Change Request: B555017407 The process opcmona has a memory leaks in different areas: 1. When using WMI sources in the Measurement Threshold Policies; 2. When monitoring of CODA sources fails and needs a reinitialization; 3. When either using the Policy.SourceEx method of scripting or upgrading policies very often. Change Request: H555010422 opcmsga binds to the NAT translated IP address causing long delays before messages are sent. Change Request: B555016456 After using opcswitchuser to change identity under which the OVO agent runs the coda process does not start anymore and writes the following error to the logfile. CODA database access denied, check permissions Change Request: 8606304320, 8606356454 The <$LOGFILE> variable is not expanded if used in the message key field. The correlation will not work as expected. Change Request: B555014691 In case the node attribute in automatic and operator initiated actions in msgi templates are empty, the value of <$MSG_NODE_NAME> is used, which is misleading. <$MSG_GEN_NODE_NAME> is used instead. Change Request: 8606301574 Text defined for the default message is not expanded for logfile, opcmsg and monitor policies. Change Request: B555015592 Child processes of opcmona may hang indefinitely. The message OpC30-610 "Can't start monitor script for object ... Script still running." is generated. Change Request: 8606313801, B555017924 If all the server entries in mgrconf are not resolvable then the agent restart will fail during initialization. Change Request: 8606300991 opcdcode of Logfile Encapsulator policy shows version number is not updated after a new policy has been deployed, even though the policy is working using the latest version. Change Request: B555017592 An application that uses the opcmsg() or opcmon() API calls and that is linked with the -Bimmidate flag will abort, reporting that there is an unresolved symbol GetLastError(). Change Request: B555018498, R555026201, B555019473 A measurement threshold policy with a single external source that is deployed from an OVO Windows Management Server may miss the first value sent by an opcmon call. Change Request: B555018312 opcmsg aborts with an error if called as non-root user because the nodeinfo file was not readable for non-root users. Change Request: H555010496 The control agent may abort when doing changing agent registration files and loading them with opcagtreg -add. Change Request: H555010468 opc(r)agt does not have an option to remove queue and temporary files. Change Request: B555016324 Coda does not restart after using opcswitchuser. Change Request: R555022806, R555021214 The Monitor Agent may abort when using Visual Basic or Perl scripts to write data into CODA. Change Request: H555010231 When the Application Package Monitor (opcapm) is configured and running on the node, the deployment of some policies may cause opcctla to abort. Change Request: B555006064 Enhancement to opc(r)agt to allow a removal of all agent's temporary files before a restart. PHSS_28960: Change Request: B555015785 opcecaas may report that it runs out of free process slots and therefore can't start anymore process. Change Request: B555008674 The opcagt and the opcragt commands have a new option '-version'. In both man pages, this was not documented. The new option was not part of the usage strings of opcagt and opcragt. Change Request: 8606189243, B555016692 Utility for checking basic types of communication between OVO server and agents is missing. Change Request: B555014591 When the OPC_INT_MSG_FLT is set to TRUE then the filtered message is received corrupted on the server in a Japanese environment. Change Request: B555017068 A.07.12 changed the behavior of opcle: Newly created logfiles that are discovered by the dynamic logfile discovery functionality are now read from the beginning. But often this is not what is desired, for example if using DBSPI with new instances that need to be monitored. Change Request: H555009399 opcmona cores if an invalid string tries to be converted. Change Request: B555015479 When executing opcagt -kill there sometimes still remains an entry for the Control Agent (opcctla) in the pids file. This should not happen because the opcctla isn't running anymore. Change Request: H555010095 Default location for GSS API library on HP-UX 11.22 agent is set to /opt/OV/lib instead of /opt/OV/lib/hpux32. Change Request: B555016565 The Agent APIs and the Agent API example programs on HP-UX 11.22 do not work. Change Request: B555015626 opcif_read() doesn't return data if the signal pipe is empty, but there is still more data in the queue file. This can for example happen if the maximum pipe size of 8192 bytes was reached and therefore no more signal bytes could be written into the signal pipe. Change Request: B555015496 Queue file handling was inefficient for larger queue files. Change Request: 8606290675, B555015230 Agent should report 'agent start' into the opcerror file. Change Request: R555018185 If the agent is started in Simplified Chinese codeset, monitor agent, logfile encapsulator and message interceptor processes will fail. Change Request: B555015449 Monitor agent may abort while receiving templates/policies. Change Request: B555015155 Errors in the opcerror file are not written to the trace file. This makes it difficult to find the place in the trace file where the error occurred. Change Request: B555015047 opcecaas (Annotation Server) might log error "Illegal NULL parameter (function ) (OpC10-11)" with many similar lines. Change Request: B555015557 The current implementation of the control agent does not allow to restart aborted sub agent processes automatically. Change Request: R555018044 The monitor scheduled command policy will be executed only once. Change Request: B555015199 If OPC_IP_ADDRESS (nodeinfo) and OPC_NAMESRV_LOCAL_NAME (opcinfo) were used in combination the entry in OPC_IP_ADDRESS was not used in the right way. Change Request: 8606282247, 8606284806, 8606310913, 8606286654 Logfile Encapsulator does not perform variable replacement for all Message Defaults fields. Change Request: B555015712 The trap/event interceptor forwards traps with the source address 127.0.0.1 unchanged, therefore they are discarded by the management server due to the unknown source address. Change Request: B555015331 The monitor agent, opcmona, may report wrong results of executed monitor scripts or programs when using many 'advanced monitors' such as OVPERF. In some of these cases opcmona might even abort. Change Request: B555015758 opcmsgi aborts if one of the set attributes has an unmatched '<'. Change Request: 8606232431 VPO tries to resolve node names that only contain blanks because of typo in template definition or variable assignment. This leads to a lot of unnecessary DNS traffic. Change Request: B555013121, B555016444, B555013474 When distributing agent software after the installation of an agent patch, all agent components (opc_pkg.Z, comm_pkg.Z and perf_pkg.Z) are distributed and installed, although only one of the components was changed. Change Request: B555015349 The monitor agent might abort while using templates/policies that have an external source specified. Change Request: B555015325 opcmsga wrongly generates a message OpC30-3002 'opcctla not running on node localhost.' when receiving a reconfigure signal, or at exit time. Change Request: 8606300119 The logfile encapsulator does not return the last line of a logfile immediately, if there is no carriage return in this line. Change Request: 8606297998 The ECS engine (opcecm & opceca) might have problems when a circuit is using global dictionaries. (This is the case for the MessageStorm detection ECS circuit). Defect Description: PHSS_30994: Change Request: 8606353730 The agent may report OpC20-184 during startup or after a management server reboot. This message can be safely ignored. This message indicates, that the opcmsga performed a cleanup of the msgagtdf file to remove empty entries, but it doesn't remove any messages. Change Request: H555012193 A workaround has been added to the code by using regular write calls instead of streams related ones if fdopen() fails. Change Request: 8606346413 The agent component versions have to be updated in the opcinfo file and the user has to get a note that the agent has to be activated with opcactivate(1m) to set the name of the management server. Change Request: H555011117, H555013386 Agent installation will fail if OVO8 HTTPS based agent is already installed on the managed node. Change Request: H555011875 A NULL pointer is present in string comparison when executing opcagtreg -add reg, AGENT_ID is the same as already registered agent and certain fields in reg file are missing. SOLUTION: Added check for NULL pointer and set problematic string to "". Change Request: 8606348297 It was not possible to test communication types individually. -t option has been added in order to allow the user to choose either tcp, icmp, or snmp. Change Request: B555030766 MSI now allows access for non-root programs that run as the same user as the agent and opcswitchuser.sh will change the permissions on some more directories that are required for ECS. Change Request: B555025991 The opc_op user gets a LANG=C line in its .profile upon installation. This has been changed, but in order to get this fix you have to remove this user then reinstall the agent (or remove this 'LANG=C' line manually). Change Request: R555026293 The control agent doesn't inform the ServiceDiscovery subagent about newly installed/updated policies in case the agent is configured in cluster aware mode. This is caused by an invalid handling of those policies by the control agent. At the time those policies should be updated, the control agent lost track which of the existing policies have been changed and treats them all as unchanged. Change Request: B555032445 opcctla only attempts to restart subagents with the id of 1. Now it attempts to restart all subagents whenever a configuration is distributed to the agents. If the user does not want the additional subagents to be restarted, then they should deregister the subagent using opcagtreg. Change Request: B555028360 opcapm reports error OpC30-3287. This error message is missleading, because actually it means, that this is either no cluster node or the cluster service is not running. opcapm reports a new error message OpC30-3288, which should is more meaningful, telling that it is not possible to connect to the cluster node. Change Request: H555011381 Nonreentrant library calls were used. The potential for problems is often not caught until the code is placed under heavy load on a multiprocessor machine. Calls such as ctime, gmtime, readdir, localtime... can corrupt the process' heap and cause crashes in unrelated areas. Change Request: B555029983 Typo was removed. Change Request: 8606355743 opctemplate only attempted to connect to opcctla via TCP. Now it checks nodeinfo in order to decided which protocol to use. Change Request: B555022224 Advanced Monitoring only sends a msg the first time a state is entered. Currently if a template has three thresholds of 75,50, and 25 and the values of 33, 55, 88, 60, 30, and 10 are given, msgs are only sent for 33, 55, and 88. A reset msg is sent for 10 which attempts to acknowledge the 25 threshold which is no longer in the browser. Now, messages are sent for 33, 55, 88, 60, and 30. The reset message then acknowledges the correct message. OPC_MONA_MSG_PER_STATE has been added to reflect this behaviour. TRUE (default) reflects the correct behaviour. FALSE has opcmona return to it's previous behaviour of only sending one message per state. Change Request: B555032246 The message storm circuit may not detect a second message storm and later on, you may see that opcecm even aborts. Change Request: B555017581 The monitor agent may produce very large error messages each time policies are redeployed. Change Request: B555023880 Some unsigned integers are printed as signed. Change Request: B555028060 opcmona places the OVO paths at end of $PATH instead of in front which causes a delay in application starts if the PATH contains NFS mounted directories. Change Request: B555027814 Added an opcinfo variable, OPC_SET_PROXY_FLAG_FOR_IP_ADDRESSES where is a comma-separated list of local addresses that will be used to generate 'proxied' messages instead of local ones. Change Request: 8606359109 opcnetchk did not parse the file correctly. It now correctly parses the port configurations. Change Request: H555011688 When local IP address is specified in /etc/hosts file and OPC_IP_ADDRESS is resolvable by dns but not specified in /etc/hosts, wrong IP address (from /etc/hosts instead from OPC_IP_ADDRESS)is appended to sent messages. SOLUTION: Added check for node name change after user variable replacement. If node name didn't change, there is no need for IP reset. PHSS_30169: Change Request: R555026193 An Advanced Monitor only sent a message the first time a threshold is reached. For example, if a template has three thresholds of 75, 50, and 25 and the values of 33, 55, 88, 60, 30, and 10 were given, messages were only sent for 33, 55, and 88. A reset msg is sent for 10 which attempts to acknowledge the 25 threshold which is no longer in the browser. Now, messages are sent for 33, 55, 88, 60, and 30. The reset message then acknowledges the correct message. The opcinfo variable OPC_MONA_MSG_PER_STATE has been added to reflect this behaviour: TRUE (default) reflects the new and correct behavior. FALSE has opcmona return to it's previous behaviour of only sending one message per state. Change Request: H555010571 When a message is sent and the agent is configured for a NAT environment, a NULL pointer could be used in strlen() causing a crash of the message agent. An extra check has been added to the code to prevent this. Change Request: R555026123 The monitor agent had a problem that it does thresholding against multiple source policies even when they are configured to process each instance separately. This threshold checking was done for the first set of instances found, but not for all. This behavior has been corrected, but the result is that policies that rely on this behavior will no longer work. In order to prevent this, the old behavior has been reintroduced. Two opcinfo variables exist to configure the monitor agent behavior. If no instance is available in all sources the script is called once. By setting the nodeinfo variable POLICY_NO_INST_CALL_SCRIPT to FALSE it is possible to define that the script is not called if no matching instances are available. If the nodeinfo variable POLICY_NO_INST_SEND_MSG is set to TRUE a warning message is created if no matching instances are available. The message contains also the available instances for all sources. The default is that no message is created. Change Request: 8606321177 The message agent does remove the free entries from the double buffer file (msgagtdf) regularly. Change Request: 8606310908 Added OPC_LE_MAX_LINES_READ variable for opcinfo. If it is not set, then the default of 50 lines are read with each interval. If it is set, then the number of lines that is specified will be read. If it is set to 0, then an unlimited number of lines will be read. Change Request: B555015967 The control agent may restart the service discovery or llbserver sub-agent endless in case they abort very quickly after being started. Change Request: B555026668 <$MSG_OBJECT> is now replaced with the Object set in the template condition if it is set and OPC_MON_MSGOBJ_COND_FIRST is set to TRUE in opcinfo. Change Request: B555015548 opcecaas can abort or exit in situations when there are several parallel annotate nodes running. Change Request: B555021669, B555024574 1. The original message text is now evaluated once per trap (no longer once per trap and template). This change can speed up opctrapi up to 50% when many trap templates are used. 2. opctrapi will now evaluate an opcinfo variable OPC_SEND_MAX_ONE_MSG_PER_TRAP. If set to TRUE, opctrapi will skip processing further templates for a trap after one template generated a message out of this trap. NOTE: Disconnection from pmd is still possible but will happen less frequently. Change Request: B555020612 Default suppressing setting is now used, if the setting is not made in the condition itself. The mix of counter/timer works like it was described before: First the timer is checked - all messages that go through this go into the counter check afterwards. Change Request: B555022945 Distribution of monitors fails with OpC20-65/66 if system resources are low and memory allocation fails for the monitor file. Change Request: 8606298307 An opcinfo variable was introduced to truncate the original message text to a maximum length: OPC_TRUNCATE_ORIG_TEXT -1: no truncation done (Default) 0: no original message sent at all : truncation done after the n'th character. This can be configured per managed node. Change Request: H555011662 The trace file mutex shouldn't be reinitialized if tracing of child process is not enabled. Change Request: B555018732 The error "Illegal value (function ) (OpC10-13)" is accidentally reported by opcecaas/opcecmas when using the built in ECS annotate server as well as a custom annotate server. The built in annotate server misinterprets the annotate requests for the custom annotate server and logs an error instead of ignoring those requests. Now it ignores those requests and only logs a note to tracing about the rejected request. Change Request: 8606322495 opctemplate returns always exit code zero, even if it fails or a template does not exist. Change Request: 8606323486 The check was modified to collect all groups, not just the primary group. Change Request: R555022489 This defect was caused by freeing memory twice. Change Request: B555018987 When a port range is specified for the control agent, and the first port of that range is occupied during startup, opcctla registers on a different port within the range. Clients connecting to opcctla however, expected opcctla to always be on the first port of the range, and failed to connect in this case. Change Request: 8606325445, B555021772 The help text for opctemplate has been corrected. Change Request: B555023264 <$MSG_OBJECT> is now replaced with the Object set in the template condition for threshold monitors if it is set and the opcinfo variable OPC_MON_MSGOBJ_COND_FIRST is set to TRUE. Change Request: 8606333207, 8606355920 opcle has been modified to check to see if the file name is still the same, if the file was not modified for a few intervals. If the file name has changed, opcle attempts to reopen the logfile, reading it from begin. This only works reliably if the 'close after read' option is switched off. Change Request: H555010925 FS metrics were retrieved from coda for some filesystems, for which coda values were not yet valid. Solution: Check whether the data for a specific instance is valid, before further processing the instance. Change Request: B555018953 <$MSG_NODE_NAME> is now set correctly to the trap override node name. Change Request: H555008553 itochecker_agt and its configuration file itochecker_agt.conf were introduced. Checking the kernel parameters now returns the correct values. An additional option (8) was added which gets the opcerror file on the management server. Change Request: B555018506 opcagt -type will now return the communication type. Change Request: B555025523 Enhance opcagt for coexistence with future versions of components. Change Request: B555023808, B555027712 <$MSG_OBJECT> in threshold monitor templates is now replaced consistently according to this priority: 1. Set Condition, if OPC_MON_MSGOBJ_COND_FIRST is TRUE 2. object from opcmon 3. Template Default 4. Empty String Change Request: H555011234 The "Process table mutex" was initialized too late. Change Request: R555025398 opcle now checks if the file was already opened, and closes it if necessary before reopening it. Change Request: B555019025 The opcswitchuser.sh utility checked whether the group exists, but did not detect if the group was only a substring of an existing group. Change Request: B555025674, 8606341274 Support of HP-UX 11.23 has been added for the Message/ Action component of the agent through this patch. Please install following patches for support of the other agent components: PHSS_29442 OV ITO7.1X Comm Agt HPUX 11.22 A.07.13 PHSS_29553 OV ITO7.1X Perf Agent HPUX 11 IA A.07.20 Change Request: H555009905 A query whether to remove the installation packages was added to the opc_inst script. PHSS_29354: Change Request: B555017407 The process opcmona has a memory leaks in different areas: 1. When using WMI sources in the Measurement Threshold Policies; 2. When monitoring of CODA sources fails and needs a reinitialization; 3. When either using the Policy.SourceEx method of scripting or upgrading policies very often. Change Request: H555010422 opcmsga used the address specified via the OPC_IP_ADDRESS option instead of a local IP address. Only local RPC calls were affected. Change Request: B555016456 After using opcswitchuser to change identity under which the OVO agent runs the coda process does not start anymore and writes the following error to the logfile. CODA database access denied, check permissions Change Request: 8606304320, 8606356454 The <$LOGFILE> variable is not expanded if used in the message key field. The correlation will not work as expected. Change Request: B555014691 In case the node attribute in automatic and operator initiated actions in msgi templates are empty, the value of <$MSG_NODE_NAME> is used, which is misleading. <$MSG_GEN_NODE_NAME> is used instead. Change Request: 8606301574 Text defined for the default message is not expanded for logfile, opcmsg and monitor policies. Change Request: B555015592 Child processes of opcmona may hang indefinitely. The message OpC30-610 "Can't start monitor script for object ... Script still running." is generated. Change Request: 8606313801, B555017924 If all the server entries in mgrconf are not resolvable then the agent restart will fail during initialization. Change Request: 8606300991 opcdcode of Logfile Encapsulator policy shows version number is not updated after a new policy has been deployed, even though the policy is working using the latest version. Change Request: B555017592 An application that uses the opcmsg() or opcmon() API calls and that is linked with the -Bimmidate flag will abort, reporting that there is an unresolved symbol GetLastError(). Change Request: B555018498, R555026201, B555019473 A measurement threshold policy with a single external source that is deployed from an OVO Windows Management Server may miss the first value sent by an opcmon call. Change Request: B555018312 The agent installation will now create the nodeinfo file with the correct permissions. Change Request: H555010496 When changing subagent registration data, opcagtreg informs the Control Agent of the change (if it is running). The Control Agent then reads the new registration file and stops/starts the processes based on the new data. During comparison of old and new data, a NULL pointer could be used in a strcmp() call which caused crashes. Change Request: H555010468 A new option (-cleanstart) has been added to the opc(r)agt command. It stops the agent, removes all files in the agent's temporary directory and restarts the agent. Trace and pids files are not removed. Change Request: B555016324 Coda does not restart after using opcswitchuser because the /var/opt/OV/log/coda.log permissions were not changed and the new user could not write into the logfile. Change Request: R555022806, R555021214 The Monitor Agent abort is caused by the usage of a static buffer to split the CODA data path into the pieces. When two or more threads split path names at the same time, one could overwrite the other's data. Change Request: H555010231 When the Application Package Monitor (opcapm) is configured and running on the node, the deployment of some policies may cause opcctla to abort. Change Request: B555006064 A new option (-cleanstart) has been added to the opc(r)agt commands. It stops the agent, removes all files in the agent's temporary directory and restarts the agent. The trace and pids files are not removed. PHSS_28960: Change Request: B555015785 One possible cause for opcecaas to report that there are no more free process slots is, that all slots are in use by applications that are running very long or might even hang. In order to have control on this, the opcecaas retrieves the timeout set for the "Annotate Node" in the ECS circuit and kills the process in case it is beyond this timeout. Change Request: B555008674 The man pages for opcagt and opcragt now document the new option '-version'. The message catalog was updated to show the '-version' option in the usage string of the opcagt and opcragt commands. Change Request: 8606189243, B555016692 The support tool /opt/OV/contrib/OpC/opcnetchk was introduced to allow a basic ICMP check, TCP check and SNMP check. Change Request: B555014591 The defect was caused by the double conversion from the server code set to the internal code set, once on the agents side when it sent the internal message to opcmsga and once by opcmsgi when it forwarded the message again. Now, the message is converted back from the internal code set to the server code set in opcmsga before sending the message to the opcmsgi queue. The management server will get the message through opcmsga in the internal code set, and it will convert it into the server code set. The conversion is made only if the internal code set is different from the server code set. Change Request: B555017068 The default behavior is now again the behavior of A.07.10: If a new logfile is returned by the logfile discovery program, only new lines of the new logfile are processed. If it is required that all lines of newly added logfiles are processed, add following line to the opcinfo file: OPC_NEW_LOGFILE_FROM_BEGIN TRUE Change Request: H555009399 opcmona cores if an invalid string tries to be converted. Change Request: B555015479 When executing opcagt -kill there sometimes still remains an entry for the Control Agent (opcctla) in the pids file. This should not happen because the opcctla isn't running anymore. Change Request: H555010095 Default location for GSS API library on HP-UX 11.22 agent is set to /opt/OV/lib instead of /opt/OV/lib/hpux32. Change Request: B555016565 The Agent APIs and the Agent API example programs on HP-UX 11.22 do not work. Change Request: B555015626 opcif_read() doesn't return data if the signal pipe is empty, but there is still more data in the queue file. This can for example happen if the maximum pipe size of 8192 bytes was reached and therefore no more signal bytes could be written into the signal pipe. Change Request: B555015496 So far a queue file garbage collection was done, if more than 256 Kbyte were unused. Now the unused space has to be more than 256 Kbyte and more than a quarter of the queue file size. This drastically reduces file I/O when handling large queue files. Change Request: 8606290675, B555015230 Agent should report 'agent start' into the opcerror file. Change Request: R555018185 Simplified Chinese codeset is now mapped as a valid codeset. Change Request: B555015449 Monitor agent may abort while receiving templates/policies. Change Request: B555015155 Whenever an error is added to the internal error list, a trace line with the DEBUG area ERRLIST is written to the trace file. When the error list is written to opcerror, another trace line with the DEBUG area ERROR is written. ERROR and ERRLIST are distinguished, since some errors are added to the error list, but then later ignored and therefore never appear in the opcerror file. Change Request: B555015047 opcecaas (Annotation Server) might log error "Illegal NULL parameter (function ) (OpC10-11)" with many similar lines. Change Request: B555015557 The control agent has been changed in order to make the restart of an aborted sub agent process configurable. The control agent can be configured to restart aborted sub agent processes. Furthermore it can be defined how often a process should be restarted in a certain time interval. To configure this, you can use the following variables in the opcinfo file: OPC_RESTART_SUBAGENT If set to TRUE, the control agent tries to restart aborted sub agent processes. The restart is done a defined number of times (OPC_RESTART_COUNT) in a specified period of time (OPC_RESTART_MINIMUM_RUN_TIME). In case the process aborts more often, it wont be restarted again. Type/Unit : TRUE|FALSE Default : TRUE OPC_RESTART_COUNT Defines how often an aborted sub agent process should be restarted within the specified minimum runtime. In case a process stops more often it wont be restarted. (See OPC_RESTART_SUBAGENT) Type/Unit : integer Default : 5 OPC_RESTART_DELAY Defines the time the control agent waits before it restarts an aborted sub agent process. The time is specified in seconds. Type/Unit : integer Default : 10 OPC_RESTART_MINIMUM_RUN_TIME Defines the time frame a sub agent process should run without being restarted more than specified by OPC_RESTART_COUNT. The time is specified in minutes. Type/Unit : integer (minutes) Default : 60 Change Request: R555018044 Internal flag is restored after monitor schedule policy execution. Change Request: B555015199 If OPC_IP_ADDRESS (nodeinfo) and OPC_NAMESRV_LOCAL_NAME (opcinfo) were used in combination the entry in OPC_IP_ADDRESS was not used in the right way. Change Request: 8606282247, 8606284806, 8606310913, 8606286654 Variable replacement is now performed for all Message Defaults fields. Change Request: B555015712 In order to be able to correctly handle traps that have a source address 127.0.0.1 the trap/event interceptor is now able to replace the localhost address (127.0.0.1) with the IP address of the node processing the trap. In order to enable this, you need to add the following line to the opcinfo file on your managed node: OPC_RESOLVE_TRAP_LOCALHOST TRUE Change Request: B555015331 opcmona holds a central table for all subprocess related information. Advanced monitors are executed in separate threads and could access this table in parallel, thus overwriting each other's data. The table accesses are now serialized by a mutex. Change Request: B555015758 opcmsgi aborts if one of the set attributes has an unmatched '<'. Change Request: 8606232431 VPO now ignores node names that contain only white space characters without contacting the name service. Change Request: B555013121, B555016444, B555013474 A new tool has been introduced, that will be called when an agent patch is installed to set the software flag in the database for all nodes of that platform to MODIFIED (node needs new agent software). Thus, it is no longer necessary to use force update to install the agent software to nodes of that platform. The second change will check what component versions are already installed on the agent and only distribute and install the newer agent packages, if force update is NOT used. Change Request: B555015349 The monitor agent aborts when using templates/policies with external sources as soon as it receives a value from the external source during checking for the threshold with a previously received value. In this case the received value is stored temporary and as soon as the monitor agent tries to process this value it aborts. The root cause is that the monitor agent tries to free up already freed memory. Change Request: B555015325 During startup a timing issue prevented a connection from opcmsga to opcctla, which generated the message. This message was not immediately reported, but only after receiving a signal. Retries will now prevent the connection failure. If it still fails because opcctla is really not running, the error will be reported immediately. Change Request: 8606300119 The logfile encapsulator does not return the last line of a logfile immediately, if there is no carriage return in this line. Change Request: 8606297998 The ECS engine (opcecm & opceca) might have problems when a circuit is using global dictionaries. (This is the case for the MessageStorm detection ECS circuit). Enhancement: No SR: R555026293 R555026201 R555026193 R555026123 R555025398 R555022806 R555022489 R555021214 R555018185 R555018044 H555013386 H555012193 H555011875 H555011688 H555011662 H555011381 H555011234 H555011117 H555010925 H555010571 H555010496 H555010468 H555010422 H555010231 H555010095 H555009905 H555009399 H555008553 B555032445 B555032246 B555030766 B555029983 B555028360 B555028060 B555027814 B555027712 B555026668 B555025991 B555025674 B555025523 B555024574 B555023880 B555023808 B555023264 B555022945 B555022224 B555021772 B555021669 B555020612 B555019473 B555019025 B555018987 B555018953 B555018732 B555018506 B555018498 B555018312 B555017924 B555017592 B555017581 B555017407 B555017068 B555016692 B555016565 B555016456 B555016444 B555016324 B555015967 B555015785 B555015758 B555015712 B555015626 B555015592 B555015557 B555015548 B555015496 B555015479 B555015449 B555015349 B555015331 B555015325 B555015230 B555015199 B555015155 B555015047 B555014691 B555014591 B555013474 B555013121 B555008674 B555006064 8606359109 8606356454 8606355920 8606355743 8606353730 8606348297 8606346413 8606341274 8606333207 8606325445 8606323486 8606322495 8606321177 8606313801 8606310913 8606310908 8606304320 8606301574 8606300991 8606300119 8606298307 8606297998 8606290675 8606286654 8606284806 8606282247 8606232431 8606189243 Patch Files: OVOPC-CLT.OVOPC-UXIA-CLT,fr=A.07.10,fa=HP-UX_B.11.00_32/64, v=HP: OVOPC-CLT.OVOPC-UXIA-CLT,fr=A.07.10,fa=HP-UX_B.11.11_32/64, v=HP: /var/opt/OV/share/databases/OpC/mgd_node/vendor/hp/ia64/ hp-ux11_32/A.07.10/RPC_DCE_TCP/opc_pkg.Z /var/opt/OV/share/databases/OpC/mgd_node/vendor/hp/ia64/ hp-ux11_32/A.07.10/RPC_DCE_TCP/opc_version /var/opt/OV/share/databases/OpC/mgd_node/vendor/hp/ia64/ hp-ux11_32/A.07.10/RPC_DCE_TCP/install/opcrinst /var/opt/OV/share/databases/OpC/mgd_node/vendor/hp/ia64/ hp-ux11_32/A.07.10/RPC_DCE_TCP/install/opcrverchk /var/opt/OV/share/databases/OpC/mgd_node/vendor/hp/ia64/ hp-ux11_32/A.07.10/RPC_DCE_TCP/opc_inst /opt/OV/OpC/examples/progs/Makef.hpuxIA32 /var/opt/OV/share/databases/OpC/mgd_node/vendor/hp/ia64/ hp-ux11_32/AgentPlatform what(1) Output: OVOPC-CLT.OVOPC-UXIA-CLT,fr=A.07.10,fa=HP-UX_B.11.00_32/64, v=HP: /var/opt/OV/share/databases/OpC/mgd_node/vendor/hp/ia64/ hp-ux11_32/A.07.10/RPC_DCE_TCP/opc_pkg.Z: None /var/opt/OV/share/databases/OpC/mgd_node/vendor/hp/ia64/ hp-ux11_32/A.07.10/RPC_DCE_TCP/opc_version: None /var/opt/OV/share/databases/OpC/mgd_node/vendor/hp/ia64/ hp-ux11_32/A.07.10/RPC_DCE_TCP/install/opcrinst: HP OpenView Operations A.07.25 (06/22/04) /var/opt/OV/share/databases/OpC/mgd_node/vendor/hp/ia64/ hp-ux11_32/A.07.10/RPC_DCE_TCP/install/opcrverchk: HP OpenView Operations A.07.25 (06/22/04) /var/opt/OV/share/databases/OpC/mgd_node/vendor/hp/ia64/ hp-ux11_32/A.07.10/RPC_DCE_TCP/opc_inst: HP OpenView Operations A.07.25 (06/22/04) /opt/OV/OpC/examples/progs/Makef.hpuxIA32: None /var/opt/OV/share/databases/OpC/mgd_node/vendor/hp/ia64/ hp-ux11_32/AgentPlatform: None cksum(1) Output: OVOPC-CLT.OVOPC-UXIA-CLT,fr=A.07.10,fa=HP-UX_B.11.00_32/64, v=HP: 3736050936 27324810 /var/opt/OV/share/databases/OpC/ mgd_node/vendor/hp/ia64/hp-ux11_32/A.07.10/ RPC_DCE_TCP/opc_pkg.Z 4136466889 8 /var/opt/OV/share/databases/OpC/mgd_node/ vendor/hp/ia64/hp-ux11_32/A.07.10/RPC_DCE_TCP/ opc_version 3938590254 123063 /var/opt/OV/share/databases/OpC/mgd_node/ vendor/hp/ia64/hp-ux11_32/A.07.10/RPC_DCE_TCP/ install/opcrinst 1483465024 33076 /var/opt/OV/share/databases/OpC/mgd_node/ vendor/hp/ia64/hp-ux11_32/A.07.10/RPC_DCE_TCP/ install/opcrverchk 3112192309 3599 /var/opt/OV/share/databases/OpC/mgd_node/ vendor/hp/ia64/hp-ux11_32/A.07.10/RPC_DCE_TCP/ opc_inst 3240931622 1571 /opt/OV/OpC/examples/progs/Makef.hpuxIA32 3742178381 6509 /var/opt/OV/share/databases/OpC/mgd_node/ vendor/hp/ia64/hp-ux11_32/AgentPlatform Patch Conflicts: None Patch Dependencies: None Hardware Dependencies: None Other Dependencies: None Supersedes: PHSS_30169 PHSS_29354 PHSS_28960 Equivalent Patches: ITOSOL_00322: sparcSOL: 2.7 2.8 2.9 Patch Package Size: 25800 KBytes Installation Instructions: Please review all instructions and the Hewlett-Packard SupportLine User Guide or your Hewlett-Packard support terms and conditions for precautions, scope of license, restrictions, and, limitation of liability and warranties, before installing this patch. ------------------------------------------------------------ 1. Back up your system before installing a patch. 2. Login as root. 3. Copy the patch to the /tmp directory. 4. Move to the /tmp directory and unshar the patch: cd /tmp sh PHSS_30994 5. Run swinstall to install the patch: swinstall -x autoreboot=true -x patch_match_target=true \ -s /tmp/PHSS_30994.depot By default swinstall will archive the original software in /var/adm/sw/save/PHSS_30994. If you do not wish to retain a copy of the original software, include the patch_save_files option in the swinstall command above: -x patch_save_files=false WARNING: If patch_save_files is false when a patch is installed, the patch cannot be deinstalled. Please be careful when using this feature. For future reference, the contents of the PHSS_30994.text file is available in the product readme: swlist -l product -a readme -d @ /tmp/PHSS_30994.depot To put this patch on a magnetic tape and install from the tape drive, use the command: dd if=/tmp/PHSS_30994.depot of=/dev/rmt/0m bs=2k Special Installation Instructions: BEFORE LOADING THIS PATCH... This patch applies to the HPUX 11.22 Agent fileset OVOPC-CLT.OVOPC-UXIA-CLT that was introduced after the base product was released. A depot containing the base HPUX 11.22 Agent fileset can be downloaded from: ftp://ovweb.external.hp.com/pub/cpe/ito/OVOPC-UXIA-CLT (A) Patch Installation Instructions ------------------------------- (A1) Install the patch, following the standard installation instructions. For backing up the system before installing a patch, you may use opc_backup(1m) NOTE: MAKE SURE THAT NO AGENT OF THE PLATFORM ADDRESSED BY THIS PATCH IS DISTRIBUTED (either from the VPO Administrator's GUI or from command line using inst.sh) WHILE RUNNING SWINSTALL. NOTE: This patch must be installed on the VPO Management Server system, NOT on an VPO Managed Node directly. Changes will take effect on managed nodes by means of VPO Software Distribution. See chapter 2 of the VPO Administrator's Reference manual for more information. NOTE: The VPO Agent consists of several components that are patched individually. This patch updates only the Event/Action component. Therefore the software distribution will tell, that the agent software is updated to A.07.10, not to A.07.25. Also the version of the ITOAgent bundle will still be A.07.10. You can verify the installed version of the components on the agent system using opcragt -agent_version. For example: opcragt -agent_version hpbbln8 Node hpbbln8.bbn.hp.com: OPC_INSTALLED_VERSION = A.07.25 PERF_INSTALLED_VERSION = A.07.15 COMM_INSTALLED_VERSION = 2.6.4.0 Done. (B) Patch Deinstallation Instructions --------------------------------- (B1) To deinstall the patch PHSS_30994 run swremove: NOTE: MAKE SURE THAT NO AGENT OF THE PLATFORM ADDRESSED BY THIS PATCH IS DISTRIBUTED (either from the ITO Administrator's GUI or from command line using inst.sh) WHILE RUNNING SWREMOVE. # swremove PHSS_30994