Hardware Sentry KM for PATROL
Release Notes v1.8.01
•Hardware Sentry KM for PATROL is now officially supported on Windows Server 2012.
•The KM now supports multi-tier authentication schemes for WBEM connections, that is connections to a VMware ESXi host with an authentication through VMware vCenter, for example.
•It is now possible to disable the validation tests of the JRE used by the KM to run Java code and therefore force the KM to use a non-Sun or non-Oracle JRE. To disable the validation test, you need to set the /SENTRY/MATSYA/disableJRECheck variable in the PATROL Agent's configuration accordingly. This is notably useful on IBM AIX and HP-UX systems where Sun/Oracle’s JRE is not available. Warning: Apache JRE is not supported. Note: A valid JRE is required by the KM only to perform certain tasks, like WBEM queries and remote SSH connections.
•Missing Parameters: It is now possible to modify the number of times Missing parameters thresholds can be breached before an alert is triggered. The default N times value for the “Missing” parameter can be customized by setting the related configuration variable, that is missingMaxOCC.
•The ZeroBufferCreditPercent parameter was added to the MS_HW_NETWORK application class to report the percentage of zero buffer credits over the last polling interval for FC ports.
•New Configuration Variable: findWbemNamespaceTimeout defines the number of seconds after which a remote Hardware Sentry KM for PATROL stops trying to get a server namespace via WBEM. Default value is set to 1200s.
•Dell Servers with iDRAC7 Management Cards: The KM is now able to monitor Dell PowerEdge Servers out of band via its iDRAC management card.
•Brocade SAN Switches: The connector now supports the monitoring of WWN Cards.
•HP-UX Servers via WBEM: The connector now enables the monitoring of newer HP Integrity i2 servers running HP-UX: HP Integrity BL860c i2, BL870c i2 & BL890c i2 Server Blades, rx2800 i2 and HP Integrity Superdome 2).
•Sun ILOM SSH Connector for Blades: The connector now supports the monitoring of Discrete Temperature and Voltage sensors.
Changes and Enhancements
•It is now possible to fully manage Hardware Sentry alert thresholds through BMC PATROL standard administration tools (Event Management KM, PCM, etc.). Hardware Sentry KM for PATROL no longer resets the alert thresholds automatically for all application classes at each discovery. The default alert thresholds are now set only once when the KM first runs.
•The overall performance and scalability of the KM has been dramatically improved so that the product is now able to monitor several hundreds remote VMware ESX hosts from one Agent.
•The overall performance of the KM discovery process has been greatly improved. The number PATROL Agent configuration changes has been drastically reduced leading to lesser I/O resource consumption.
•The KM no longer automatically switches from one threshold mechanism to another. Threshold mechanism can be selected through the Selecting the Threshold Mechanism KM Command.
•Data Domain Storage Systems Monitoring: Temperature thresholds are now set for Ambient and CPU sensors.
•Monitoring IBM DS3000, DS4000, DS5000 Series: Disk Drawers are now fully monitored.
•The default threshold for the DegreesBelowWarning parameter has been removed to avoid duplicate notification with the Temperature parameter threshold.
•The Debug mode was modified to ensure full compatibility with BPPM 9.x.
•EMC Storage Systems:
▪Controllers and storage processors are now reported as Controllers under the Other Devices category.
▪The ID of the storage processors and controllers is no longer displayed twice in the label of the corresponding instance.
▪The speed of each port is now monitored with the LinkSpeed parameter of the MS_HW_NETWORK class.
•For all fiber-connector systems (storage systems, FC ports, HBAs and SAN switches), the connection type of the port is no longer detailed in the instance name. Only the main acronym is being used: "FC Port N", "FC Port E", etc.
•NetApp Filers - NVRam Battery Monitoring: The NVRam Battery Status is now supported by the connector.
•Hardware Sentry MIB-2 connector now supports 10 G port speed.
•On Windows systems, the discovery of Physical Disks has been improved to reduce the number of WMI queries.
•On Windows systems, the number displayed in the icon label to identify the physical disk now matches with Windows' administration tools.
Version 1.8.01 of Hardware Sentry fixes issues relating to the monitoring of:
•Sun Sparc Servers: The connector now takes into consideration the Exponent of a Temperature sensor.
•IBM DS Storage Systems I345 Battery: The battery status of Battery maintenance charging was not recognized as an OK status.
•NetApp Filers: The identifier of the Physical Disk ID/ Disk Shelf WWN was changed to prevent false missing alerts to be generated for Physical Disks, Temperature, Fans and Power Supplies. The original identifier can now be found in the Infobox under Identifying Information.
•IBM Director 6.2.1 Monitoring: The detection criteria query "SELECT Name FROM __NAMESPACE" has been removed to prevent the connector to fail unnecessarily.
•IBM Director 6.3 or later: Since temperature and voltage thresholds are stored in a different property depending on the version of the IBM Director Agent used, temperature and voltage thresholds were missing. This version of the KM takes into consideration both pre 6.3 and 6.3 threshold properties.
•Sun Servers running Solaris: When a Physical Disk failed or was removed and reported a size of zero bytes, the monitoring of all the Physical Disks failed.
•Cisco Telnet Switches:
▪The connector will no longer add a unique identifier to Chassis Fans if they already have one.
▪The connector adds the hostname or the hostname information to the Connected to Physical Address property of the FC Port.
•Data Domain Storage Systems: Devices with blank IDs were being created causing false missing alerts.
•IBM DS3000, DS4000, DS5000 Storage Systems:
▪The Physical Disk status of Replaced was not interpreted correctly.
▪Power Supply Fan failures were reported as unknown failures and did not generate alarms. The connector will now alarm on Power Supply Fan failures and warn on any unknown failures.
▪'$' characters were not protected when running OS commands in a Telnet or SSH session, preventing the monitoring of VIO servers from a UNIX server.
▪The connector has been modified to test for an unrestricted shell and if found run all commands with their full path. If a restricted shell is found, commands will continue to be run with no path specified. AIX Connectors not designed for VIO Servers will also now be deactivated.
•Servers equipped with HBA Cards: Hardware Sentry has been modified to prevent the creation of duplicate instances of the same HBA card when the KM is in automatic detection mode.
•Recent Oracle (SUN) x86 Servers Using IPMI:
▪The KM now recognizes the manufacturer identifier of Oracle and processes SUN/Oracle specific sensors properly.
▪Oracle/Sun specific power supply sensors (PSx/VINOK and PSx/PWROK) are now correctly interpreted.
▪The debugging information has been improved.
•Disks on VMware ESX servers: The KM now discovers and monitors Physical Disks even if when there is no presence sensor for the Physical Disks.
•Sun Solaris - Environment (prtdiag, lom): Power Supply and Memory Module failures were erroneously reported as Unknown. The connector has been modified to report these specific failures as Alarm.
•HP StorageWorks MSA 2000 & P2000: Discovery stopped when the string "#" was found in the xml command output; this string was identified as the prompt, which caused the capture of the full output of the command to stop.
•EMC Storage Systems: FC and Ethernet ports that were not configured or not in use were inaccurately reported as Failed (Stopped) with a status in Alarm. Such ports are now reported with an OK status.
•HP EVA Disk Arrays: The detection criteria for the connector have been modified to recognize the output of the P6000 version of SSSU. Also, passwords with exclamation marks (!) no longer cause the batch file to fail.
•Ethernet ports on Cisco Nexus switches or Cisco MDS switches embedded in a Blade Chassis: The timeout issue occurring with SSH/Telnet connections has been fixed.
•Brocade, McData and Fibre Alliance Switches monitored using SNMP: Depending on the model, firmware versions and manufacturer of SAN switches, different values have been used to convert from Kb/s to Mb/s (100, 125 or 106.25). The KM now detects which value has been used and uses it to calculate the link speed of the port and its bandwidth utilization.
•Sun ILOM SSH: Physical Disks were not monitored if State sensor was not available.
•HP Proliant Systems using HP's NIC Teaming: The monitoring of the physical cards in a NIC teaming configuration is now done through the operating system instrumentation. The HP Insight Management Agent - Network Connector now monitors the status of the logical network cards only.
•The KM now properly handles white spaces in the %PATROL_HOME% environment variable when executing nawk.exe on PATROL Agent version 3.9.00 and higher.
•Some parameters could be activated by the discovery even though no data was available, notably the Power Consumption parameter.
•On Windows systems, remote LUNs mounted by an HBA were incorrectly discovered as internal Physical Disks.
•Since the introduction of ANYINST/ALL_INSTANCES thresholds in v1.8.00, devices with an identifier containing over 100 characters could have their parameters deactivated by the Event Management KM. This problem is fixed with version 1.8.01.
•Some versions of the Sun Blade Chassis firmware have incomplete identifiers for Power Supplies that prevented these components to be properly discovered and monitored.
•On older HP servers that are not on VMWare's ESX HCL, environmental sensors were not found by the HP agent for ESX, but could be found using ESX's inbuilt hardware monitoring system. From now on, if sensors are not found by the HP connector the PM will revert to using the VMWare connector.
•Alert on missing parameters are now properly closed and the corresponding events are triggered accordingly.
•The PATROL Agent must be executed on a server where the regional settings are set to 'English (United-States)' in order to avoid collected values to be rounded up and therefore lead to inaccurate calculation or untimely alerts.