README VERSION : 1.1 README CREATION DATE : 2012-09-24 PATCH-ID : PHCO_43070 PATCH NAME : VRTSdbed 5.1 SP1RP2 BASE PACKAGE NAME : Veritas Storage Foundation for Oracle from Symantec 142631-06SP1RP3(patch-6) BASE PACKAGE VERSION : VRTSdbed 5.1SP1 SUPERSEDED PATCHES : NONE REQUIRED PATCHES : NONE INCOMPATIBLE PATCHES : NONE SUPPORTED PADV : hpux1131 (P-PLATFORM , A-ARCHITECTURE , D-DISTRIBUTION , V-VERSION) PATCH CATEGORY : PATCH CRITICALITY : OPTIONAL HAS KERNEL COMPONENT : NO ID : NONE REBOOT REQUIRED : NO PATCH INSTALLATION INSTRUCTIONS: -------------------------------- Please refer the release notes for installation instructions PATCH UNINSTALLATION INSTRUCTIONS: ---------------------------------- Please refer the release notes for un-installation instructions. SPECIAL INSTRUCTIONS: --------------------- NONE SUMMARY OF FIXED ISSUES: ----------------------------------------- 2848176 (2816482) vxdbd memory leak in build_function_header() 2848193 (2664050) vxdbd coredumps in build_function_header on malloc failure 2848204 (2664794) vxdbd sigseg in strcpy from dbed_ul_print_valist 2915205 (1957142) reverse_resync_abort and reverse_resync_commit fails SUMMARY OF KNOWN ISSUES: ----------------------------------------- KNOWN ISSUES : -------------- FIXED INCIDENTS: ---------------- PATCH ID:PHCO_43070 * INCIDENT NO:2848176 TRACKING ID:2816482 SYMPTOM: memory consumed by vxdbd continues to increase with time, even when the daemon is totally idle. DESCRIPTION: After a few minutes of running vxdbd , 2 memory leaks are observed in the code. This seems to occur even if the vxdbd is idle RESOLUTION: free the memory allocated in the above two scenarios. * INCIDENT NO:2848193 TRACKING ID:2664050 SYMPTOM: vxdbd core dump after running for about 5 hours DESCRIPTION: VXDBD was running fine for about 5 hours. (10:15 to 15:15) VXDBD calls "vxdctl -c mode" every minute to figure out if its the MASTER node or not If MASTER, the code tries to run scheduled jobs. After running for a few hours, malloc fails at this point and the return value is not checked leading to a failure. RESOLUTION: Check for the return value of malloc() and bail out if memory not allocated. * INCIDENT NO:2848204 TRACKING ID:2664794 SYMPTOM: Sometime after startup vxdbd process is core dumping due to segmentation fault. DESCRIPTION: In the function dbed_ul_print_valist() the return value from malloc() is not checked leading to the dailure. RESOLUTION: Added a check for malloc() failure in dbed_ul_print_valist(). * INCIDENT NO:2915205 TRACKING ID:1957142 SYMPTOM: Sometimes reverse_resync_abort and reverse_resync_commit operation fail with an error similar to SFORA dbed_vmsnap ERROR V-81-3004 An error occurred while shutting down the database instance. DESCRIPTION: Currently the database is shutdown within the script as "SHUTDOWN NORMAL", In NORMAL shutdown the Oracle database would wait for all connections to exit or complete any pending tasks. This is not really necessary since at times the connections are open with no activity. This causes the operation to stall for sometime before eventually quitting with an error. RESOLUTION: The Shutdown is done now, using SHUTDOWN IMMIDIATE. INCIDENTS FROM OLD PATCHES: --------------------------- Patch Id::PHCO_42093 * Incident no::2361363 Tracking ID ::2361361 Symptom::The qio scripts are not present in the DBED package Description::The scripts /opt/VRTSdbed/bin/qio_convertdbfiles ,/opt/VRTSdbed/bin/qio_convertdbfiles_ora ,/opt/VRTSdbed/bin/qio_getdbfiles ,/opt/VRTSdbed/bin/qio_getdbfiles_ora ,/opt/VRTSdbed/bin/qio_recreate ,/opt/VRTSdbed/bin/qio_recreate_ora are missing in the VRTSdbed package Resolution::The files have been added. * Incident no::2395173 Tracking ID ::2395171 Symptom::The vxdbd daemon uses Medium or Low ciphers Description::The vxdbd daemon and its client traditionally do not use any ciphering to communicate. This is not favorable in certain situations and hence there is a need to provide communication using configurable ciphers. Resolution::The latest VRTSat libraries are used which set a configurable parameter in the profile file /opt/VRTSdbed/eat/root/.VRTSat/profile/VRTSatlocal.conf. Specifically the "SSLCipherSuite" parameter needs to be set to appropriate ciphers to be used. To make use of this feature 1. The "SSLCipherSuite" should be set to the appropriate cipher strings in the /opt/VRTSdbed/eat/root/.VRTSat/profile/VRTSatlocal.conf file 2. The daemon should be restarted after setting the environment variable VXDBD_USE_ENCRYP=1 3. All the client side scripts/binaries need to be run with the environment variable VXDBD_USE_ENCRYP=1 set. Note: By default LOW ciphers are not supported once this feature is used. The default SSLCipherSuite string is SSLCipherSuite"="HIGH:MEDIUM:!eNULL:!aNULL:!SSLv2". An example for a SSLCipherSuite strings is , "SSLCipherSuite"="ALL:!aNULL:!SSLv2:!ADH:!eNULL:!LOW:!EXP:!MEDIUM:+HIGH" * Incident no::2395194 Tracking ID ::2395192 Symptom::The vxdbd daemon consumes excessive CPU. Description::The vxdbd daemon listens on TCP port 3233 for client connections. When a client connects, the server loops by design in vxdbd_handle_connection(). However, this looping is eventually supposed to end. In this case, for some connections and under some unknown conditions, it doesn't. For whatever reason, vxdbd threads are stuck in this loop continuing to read and write to these sockets while the number of looping threads grows over time. Meanwhile, the sockets are in an IDLE state which means the client is no longer active on the connection. Resolution::The read_n function was incorrectly returning a positive value in case of error, which caused its caller to loop indefinitely. The function was fixed to correctly return a negative value on error.