README VERSION : 1.1 README CREATION DATE : 2014-04-03 PATCH-ID : VRTSvcsag-6.0.500.000 PATCH NAME : VRTSvcsag 6.0.500.000 BASE PACKAGE NAME : VRTSvcsag BASE PACKAGE VERSION : 6.0.100.000 SUPERSEDED PATCHES : 6.0.300.000 REQUIRED PATCHES : NONE INCOMPATIBLE PATCHES : NONE SUPPORTED PADV : sol11_x64 (P-PLATFORM , A-ARCHITECTURE , D-DISTRIBUTION , V-VERSION) PATCH CATEGORY : CORE , CORRUPTION , PERFORMANCE PATCH CRITICALITY : CRITICAL HAS KERNEL COMPONENT : NO ID : NONE REBOOT REQUIRED : NO REQUIRE APPLICATION DOWNTIME : YES PATCH INSTALLATION INSTRUCTIONS: -------------------------------- Please refer to Install Guide for install instructions and Release Notes for other information. PATCH UNINSTALLATION INSTRUCTIONS: ---------------------------------- Please refer to Install Guide for uninstall instructions and Release Notes for other information. SPECIAL INSTRUCTIONS: --------------------- NONE SUMMARY OF FIXED ISSUES: ----------------------------------------- PATCH ID:VRTSvcsag-6.0.500.000 3326594 (3326591) The IPMultiNICB agent delays bringing IPv4 VIP online on Solaris version 10 by 5 seconds. 3343032 (3028760) The NFSRestart resource does not start the NFS processes such as statd and lockd, during the online or offline operation. 3343041 (3125918) The Asynchronous Monitoring Framework (AMF) driver causes a panic in the node when the vxconfigd process is unresponsive. 3343098 (3154104) For Application agent, an error message is logged when the StartProgram or StopProgram returns a non-zero value. This gives incorrect implication of the program failure. 3343104 (3153987) In the Application agent, the clean operation is reported successful even when the CleanProgram returns a non-zero value. 3345056 (2848020) When IP is unplumbed or network cable is pulled, the SambaShare agent fails to detect the fault. 3346418 (3076468) Incorrect configuration for one zone results in failure report for other zone resources. 3347538 (3347536) The Application agent may dump a core while registering the resource with Asynchronous Monitoring Framework (AMF). 3413193 (2979745) The MultiNICA agent is unable to detect the loss of network connectivity. 3424409 (3422904) The Zone agent does not handle a zone in unavailable state. PATCH ID:6.0.300.000 2933305 (2737653) Incorrect descriptions on RVGPrimary online script 2938979 (2980002) High availability support for logical domain when control domain is down with multiple I/O domains in Oracle VM (OVM) for SPARC environment. 2967539 (2967536) MonitorProgram of Application does not work correctly with C shell of User env. 2986205 (2979745) The MultiNICA agent is unable to detect the loss of network connectivity. 2999849 (2964772) If you take an NFSRestart resource offline, the NFSRestart agent may unexpectedly stop the NFS processes in a local container (Zones on Solaris or WPARs on AIX). SUMMARY OF KNOWN ISSUES: ----------------------------------------- NONE KNOWN ISSUES : -------------- NONE FIXED INCIDENTS: ---------------- PATCH ID:VRTSvcsag-6.0.500.000 * INCIDENT NO:3326594 TRACKING ID:3326591 SYMPTOM: The IPMultiNICB agent delays bringing IPv4 VIP online on Solaris version 10 by 5 seconds. DESCRIPTION: If the if_mpadm(1M) command fails with the File exists error message, then one or more IP addresses are left without an UP flag for few seconds. During this duration, if VCS attempts to bring the IPMultiNICB resource online, the online operation fails and the monitor reports the resource state as offline. To avoid the failure, the code was modified to introduce a sleep state of 5 seconds before the IPMultiNICB resource is brought online for both IPv4 and IPv6 addresses. However, the delay is observed only in IPv6 addresses. RESOLUTION: The code is modified to avoid sleep of 5 seconds for IPv4 addresses. * INCIDENT NO:3343032 TRACKING ID:3028760 SYMPTOM: The NFSRestart resource does not start the NFS processes such as statd and lockd, during the online or offline operation. DESCRIPTION: The NFSRestart agent uses an internal counter to synchronize with the resources configured in different service groups to start the NFS processes. If a resource is deleted in an online state, the counter value does not get adjusted correctly. Due to inconsistent counter value, the agent fails to start the NFS processes such as statd and lockd. RESOLUTION: The NFSRestart agent code is modified to adjust the counter value correctly when the NFSRestart resources are added and deleted. * INCIDENT NO:3343041 TRACKING ID:3125918 SYMPTOM: The Asynchronous Monitoring Framework (AMF) driver causes a panic in the node when the vxconfigd process is unresponsive. DESCRIPTION: The libusnp_vxnotify.so library, used for disk group notifications goes into an infinite loop when the vxconfigd process is unresponsive. This causes the AMF driver to panic the node. RESOLUTION: The code is modified for the AMF driver and libusnp_vxnotify.solibrary, to address this issue. * INCIDENT NO:3343098 TRACKING ID:3154104 SYMPTOM: For Application agent, if the program or script configured in StartProgram or StopProgram returns a non-zero value, an error message is logged in engine log. DESCRIPTION: If StartProgram or StopProgram returns a non-zero value, a message is logged in engine log. The severity level of this message is marked as "ERROR", which implies the resource fails to go online or offline. But in fact, the resource might go online or offline. RESOLUTION: The code has been modified, so that the severity level of the message is changed from "ERROR" to "INFO", which avoids the incorrect implication of program failure. * INCIDENT NO:3343104 TRACKING ID:3153987 SYMPTOM: In the Application agent, the clean operation is reported successful even when the CleanProgram returns a non-zero value. DESCRIPTION: When an application resource fails, VCS invokes CleanProgram. CleanProgram returns zero if the clean operation succeeds and a non-zero value if fails. Because the Application agent fails to check the returned status value of CleanProgram, the clean operation is reported successful. This is incorrect because the non-zero returned value of CleanProgram indicates a clean entry point failure. RESOLUTION: The issue is fixed by correctly handling the returned value of CleanProgram. * INCIDENT NO:3345056 TRACKING ID:2848020 SYMPTOM: When IP is unplumbed or network cable is pulled, the SambaShare agent fails to detect the fault. DESCRIPTION: When the IP is unplumbed or network cable is pulled, the SambaShare agent fails to handle the NT_STATUS_IO_TIMEOUT error. RESOLUTION: The code is modified to handle errors specific to timeout issues. * INCIDENT NO:3346418 TRACKING ID:3076468 SYMPTOM: Incorrect configuration for one zone results in failure report for other zone resources. DESCRIPTION: The Zone agent uses the zoneadm(1M) command to monitor the state of the configured zones. If one of the configured zones is configured incorrectly, the zoneadm(1M)command shows an error even when the command is run for a correctly configured zone. Because the monitor script is unable to process this output, it incorrectly reports other functional zones as offline. RESOLUTION: The code is modified to enable the Zone agent to correctly identify the state of zones which are correctly configured. * INCIDENT NO:3347538 TRACKING ID:3347536 SYMPTOM: The Application agent may dump a core while registering the resource with AMF. DESCRIPTION: While registering the resource with AMF, incorrect memory reference in imf_register entry point leads to core dump. RESOLUTION: The code is modified to resolve this issue. * INCIDENT NO:3413193 TRACKING ID:2979745 SYMPTOM: The MultiNICA agent is unable to detect the loss of network connectivity. DESCRIPTION: While monitoring the Network Interface Cards (NIC) configured with the MultiNICA resource, if the NIC is not able to ping the NetworkHosts, the Agent tries to ping the broadcast address to check whether the NIC is healthy. The agent checks the packet count for the NIC to check if the network is reachable. The packet count may increase even due to the NIC replying to itself, without the replies from the systems in the network. Thus, Veritas Cluster Server (VCS) may incorrectly report the NIC as healthy. RESOLUTION: The MultiNICA agent is modified to check whether the NIC replies to its own broadcast ping or whether the other systems in the network are replying to the NIC. If the replies to the broadcast ping are not from the systems in the network and the NetworkHosts are not reachable, the NIC is detected to have lost network Connectivity. * INCIDENT NO:3424409 TRACKING ID:3422904 SYMPTOM: The Zone agent fails to bring a zone online and logs the following error message: Zone [zone_name] state is [unavailable]. Zone should be either in configured or installed state before onlining the resource. DESCRIPTION: When a zones state is unavailable, the Zone agent fails to bring the resource online. RESOLUTION: The code is modified such as to handle the unavailable state of the zone and to bring the resource online. PATCH ID:6.0.300.000 * INCIDENT NO:2933305 TRACKING ID:2737653 SYMPTOM: If you override the OnlineTimeout attribute value for RVGPrimary resource, the agent does not take it in account. DESCRIPTION: If the RLink of the RVG is connected but is not up-to-date, the RVGPrimary agent waits for it to become up-to-date till the OnlineTimeout attribute value expires. When this attribute value is overridden for the RVGPrimary resource in picture, it is not taken into account. Instead, the agent takes into account the OnlineTimeout defined for RVGPrimary resource type. RESOLUTION: Symantec has modified the RVGPrimary Agent to take into account the OnlineTimeout value when overridden by the RVGPrimary resource. * INCIDENT NO:2938979 TRACKING ID:2980002 SYMPTOM: High availability support for logical domain when control domain is down with multiple I/O domains in Oracle VM (OVM) for SPARC environment. DESCRIPTION: When a logical domain is configured with I/O services from multiple I/O domains, the logical domain can survive the loss of any of the I/O domain. When a I/O domain fails the logical domain transparently uses the I/O services from other I/O domain. In OVM environment, the control domain is default I/O domain for any logical domain. The logical domain administration happens from control domain. When the control domain is shutdown or rebooted the logical domain cannot be managed from any other I/O domain. Veritas Cluster Server (VCS) which is running in the control domain manages the logical domain. By default, when the VCS node (primary/control domain) is down, VCS fails over the logical domain under its management, though the logical domain can continue to run. RESOLUTION: Symantec has enhanced its solution to support the high availability for the logical domain when the control domain is shutdown or rebooted. With the help of logical domain failure policy setting with its master(s) and VCS service group attributes, Symantec provided a mechanism to not to failover the logical domain resource when control domain is rebooted or shutdown. User can set logical domain master failure-policy for control domain to ignore and VCS service group attribute SysDownPolicy to AutoDisableNoFailover, so that VCS does not failover the logical domain. * INCIDENT NO:2967539 TRACKING ID:2967536 SYMPTOM: When Application resource is configured with MonitorProgram and non-default user (where default user is root), and if the user shell is C-shell, then the resource goes to UNKNOWN state. DESCRIPTION: Application agent invokes /usr/bin/test command to check whether the file configured under MonitorProgram is executable. When an Application resource is configured with a non-default user, Application agent invokes test command with "su - ". This does not work for C-shell as it requires the su command to be invoked with "-c" option (su - -c ). Thus, Application monitor reports UNKNOWN state for the resource. RESOLUTION: The Application agent is modified to ensure that the monitor reports correct state for the resource whenever the User configured is in C-shell environment. * INCIDENT NO:2986205 TRACKING ID:2979745 SYMPTOM: The MultiNICA agent is unable to detect the loss of network connectivity. DESCRIPTION: While monitoring the Network Interface Cards (NIC) configured with the MultiNICA resource, if the NIC is not able to ping the NetworkHosts, the Agent tries to ping the broadcast address to check whether the NIC is healthy. The agent checks the packet count for the NIC to check if the network is reachable. The packet count may increase even due to the NIC replying to itself, without the replies from the systems in the network. Thus, Veritas Cluster Server (VCS) may incorrectly report the NIC as healthy. RESOLUTION: The MultiNICA agent is modified to check whether the NIC replies to its own broadcast ping or whether the other systems in the network are replying to the NIC. If the replies to the broadcast ping are not from the systems in the network and the NetworkHosts are not reachable, the NIC is detected to have lost network Connectivity. * INCIDENT NO:2999849 TRACKING ID:2964772 SYMPTOM: If you take an NFSRestart resource offline, the NFSRestart agent may unexpectedly stop the NFS processes in a local container (Zones on Solaris or WPARs on AIX). DESCRIPTION: As a part of the offline operation, the NFSRestart agent is expected to stop NFS processes only in the global container. However, the agent does not perform a strict match of container IDs before terminating a process. As a result, the agent may also terminate NFS processes in a local container. RESOLUTION: The code has been modified such that the NFSRestart agent ensures a strict container ID match, before terminating NFS processes. INCIDENTS FROM OLD PATCHES: --------------------------- NONE