Skip to main content

HP Proliant (Commercial)

ZenPacks.zenoss.HP.Proliant

This ZenPack provides support for monitoring HP Proliant Servers (G6, G7, G8, G9, and G10), including blade servers with the HP Proliant c7000 Enclosure. Discovery and monitoring are performed using SNMP, iLO (XML/HTTP), and/or WBEM protocols depending on environment.

This ZenPack currently supports HP Proliant servers running Windows or Linux (with installed/configured HP SIM SNMP subagent), HP Enclosures (via SNMP) as well as iLO devices (via either iLO or SNMP) and VMWare ESXi (using HP custom ESXi with WBEM/CIM access configured).

It is important to note that not all features of this ZenPack are supported across all protocols, since each of them provides a part, but not all, of the total information used by the ZenPack. This limitation is based on the amount and quality information that is available from each of the supported protocol. What this means is that certain metrics for a given component type (such as a temperature sensor) may or may not be available depending on the type of device that is monitored and/or on the protocol used to collect that data.

The features added by this ZenPack can be summarized as follows. They are each detailed further below.

  • Initial discovery and continual synchronization of relevant components

  • Performance monitoring

  • Event management and monitoring

  • Command line utilities for troubleshooting and diagnostics

  • Optional Service Impact with addition of Virtana Service Dynamics product

Releases

Version 3.3.5

Version 3.2.2

  • Release on 2017/04/03

  • Requires WBEM ZenPack, PythonCollector ZenPack (>=1.2)

  • Compatible with Zenoss Core 4.2.x, Zenoss Core 5.0.x, Zenoss Core 5.1.x, Zenoss Core 5.2.x, Virtana Service Dynamics 4.2.x, Virtana Service Dynamics 5.0.x, Virtana Service Dynamics 5.1.x, Virtana Service Dynamics 5.2.x

External Dependencies

This ZenPack relies on several protocols to achieve coverage of various aspects of the HP Proliant architecture. At the most basic level, SNMP should be configured so as to facilitate access to HP/Compaq MIBs (CPQSINFO-MIB, CPQRACK-MIB, and so on) and the associated OIDs. Support on a bare metal install of Linux or Windows OS, for example, requires installation of HP SIM and its associated software components, including an SNMP subagent and associated configuration for its use with the OS SNMP agent. Configuration of these software components is covered by various documentation available from HP and is beyond the scope of this document.

SNMP is also supported natively by HP OnboardAdministrator (if available) and optionally by the iLO management interface.

As noted above, the iLO management interface can be monitored (as a Virtana device) via iLO protocol (essentially XML via HTTP) or SNMP (which can be enabled in the iLO GUI interface).

An additional important item to note is that the iLO management interface GUI requires that the associated server's FQDN and IP address be set manually. This ZenPack requires the configuration of FQDN (or hostname) and IP to be set in order for Virtana to create associations on its side between monitored Proliant servers and their associated iLO device in Virtana.

Support for HP Proliant hardware running VMware ESXi OS relies on a custom ESXi image available from HP. This custom image provides support for Proliant-specific monitoring and performance data that is not available using the version of ESXi provided by VMware. Additionally, WBEM support must be configured and enabled on the ESXi host.

Discovery

The following components will be automatically discovered using either the SNMP, iLO, or WBEM related connection parameters you provide. The properties and relationships will be continually updated on Virtana' normal remodeling interval which defaults to every 12 hours.

HPEnclosure

  • Attributes: Firmware Version, Model, Part Number, Serial Number, Spare Part Number

  • Relationships: Cooling Fan, Device Bay, Interconnect Bay, Onboard Administrator, Power Supply, Temp Sensor, HPTemperatureSensors, HPThermalSubsystem

HPDeviceBay

  • Attributes: BIOS Revision, Blade Device Link, Firmware Version, Part Number, Product ID, Service Tag, Slot Number, Spare Part Number

  • Relationships: Enclosure

HPEnclosureCoolingFan

  • Attributes: Location, Part Number, Present, Redundant, Spare Part Number

  • Relationships: Enclosure

HPInterconnectBay

  • Attributes: Product Name, Location, Part Number, Present, Serial Number, Spare Part Number, Switch Index

  • Relationships: Enclosure

HPOnboardAdmin

  • Attributes: OAIndex, UUID, Bay Number, Part Number, Present, Redundant, Role, Serial Number, Spare Part Number, Enclosure Serial Number

  • Relationships: Enclosure

HPTemperatureSensor

  • Attributes: Location, Threshold, Threshold Type

  • Relationships: Enclosure

HPChassis

  • Attributes: Total RAM, Product ID, Product, Server Name, Serial Number, Enclosure Manager Serial, Enclosure Manager Name, Enclosure Manager IP, Enclosure Manager URL

  • Relationships: Processor, Cooling Fan, Logical Drive, Management Controller, Memory Slot, Network Interface, NIC, HPOSHypervisor, PCI Device, Physical Drive, Power Supply, System Board, Temp Sensor, RAID Controller, FC Adapter, FC Port

HPFibreChannelAdapter

  • Attributes: Description, Device ID

  • Relationships: Chassis, FC Port

HPFibreChannelPort

  • Attributes: Description, Device ID, WWN, WWPN

  • Relationships: Chassis, FC Adapter

HPCoolingFan

  • Attributes: Name, Type, Hot Plug, Location, Present, Redundant

  • Relationships: Chassis

HPArrayController

  • Attributes: Model, Firmware Revision, Slot, Product Revision, Partner Slot, Serial Number, Daughter Board Type, HW Location, Number of Buses, Rebuild Priority, Expand Priority, Internal Ports, External Ports, Write Cache, Partner Serial Number, ROM Revision, Encryption

  • Relationships: Chassis, Logical Drive, Physical Drive

HPLogicalDrive

  • Attributes: Redundant, Capacity, Stripe Size

  • Relationships: Chassis, RAID Controller

HPManagementController

  • Attributes: Firmware Version, Firmware Date, ipv4 Address, Active License Type, Model, Serial Number, Security Override Switch, url

  • Relationships: Chassis

HPMemory

  • Attributes: Module Number, Cpu Number, Frequency, Size, Type, Technology

  • Relationships: Chassis

HPNetworkInterface

  • Attributes: AutoSense, Capabilities, Description, Full Duplex, Link Technology, MTU, Max Speed, Name, Port Number, Port Type, Speed

  • Relationships: Chassis

HPNetworkInterfaceController

  • Attributes: MAC Address, ID, Type, IP Address, Description

  • Relationships: Chassis

HPPCIDevice

  • Attributes: Physical Slot, Board Name, Slot Type, Slot Width

  • Relationships: Chassis

HPPhysicalDrive

  • Attributes: Capacity, Firmware, Model, Speed, Location, Manufacturer, productId, Serial Number

  • Relationships: Chassis, RAID Controller

HPPowerSubsystem

  • Attributes: Part Number, Serial Number, Max Output, Present, sparePartNumber

  • Relationships: Enclosure

HPServerPowerSupply

  • Attributes: Part Number, Serial Number, Max Output, Redundant, Present,

HPProcessor

  • Attributes: L1, L2, L3, Coprocessor Count, Coprocessor Speed, Coprocessor Type, Cores, HyperThread, Model, Socket, Speed, Stepping, Thread Count

  • Relationships: Chassis

HPSystemBoard

  • Attributes: Product Name, ROM Version, Serial Number, TPM Module

  • Relationships: Chassis

HPTemperature

  • Attributes: Location, Threshold, Threshold Type

  • Relationships: Chassis

Monitoring

The following metrics will be collected every 5 minutes by default. The average statistic is collected, and the graphed value is per second for anything that resembles a rate.

HPFibreChannelAdapter

Protocol

Metrics

WBEM

—

HPFibreChannelPort

Protocol

Metrics

SNMP

cpqFcaHostCntlrCondition, cpqFcaHostCntlrStatus

WBEM

FramesTooLong, EncodingDisparityErrors, RXBroadcastFrames, ErrorFrames, MulticastFramesTransmitted, LossOfSyncCounter, LinkFailures, LinkResetsReceived, BytesReceived, DumpedFrames, FramesTooShort, CRCErrors, PacketsReceived, PacketsTransmitted, LossOfSignalCounter, TXBroadcastFrames, DelimiterErrors, LinkResetsTransmitted, PrimitiveSeqProtocolErrCount, BytesTransmitted, AddressErrors, MulticastFramesReceived

HPChassis

Protocol

Metrics

SNMP

status

ILO

status

WBEM

status

HPCoolingFan

Protocol

Metrics

SNMP

status, speed

ILO

status, speed

WBEM

status, speed

HPDeviceBay

Protocol

Metrics

SNMP

status, cpqRackServerBladeWattageAvg, cpqRackServerBladePOSTStatus, cpqRackServerBladePowered, condition

HPEnclosure

Protocol

Metrics

SNMP

status

HPEnclosureCoolingFan

Protocol

Metrics

SNMP

status

HPArrayController

Protocol

Metrics

SNMP

cpqDaAccelBattery, cpqDaCntlrPerfCpuPercentBusy, cpqDaAccelWriteErrs, cpqDaCntlrBoardStatus, cpqDaAccelBadData, cpqDaCntlrBoardCondition, cpqDaAccelStatus, cpqDaAccelErrCode, cpqDaCntlrPerfAvgLatency, cpqDaAccelReadErrs, cpqDaCntlrCondition, cpqDaAccelCondition, cpqDaCntlrPerfCommandCount, cpqDaCntlrPartnerBoardStatus

ILO

cpqDaAccelBattery, cpqDaCntlrCondition, cpqDaCntlrBoardStatus

WBEM

cpqDaAccelBattery, CacheReadPercent, cpqDaAccelStatus, CacheParityReadErrors, CacheModuleTemperature, ControllerTemperature, CacheParityWriteErrors, cpqDaCntlrCondition, CacheWritePercent

HPLogicalDrive

Protocol

Metrics

SNMP

status, sectorsRead, condition, driveWrite, driveRead, sectorsWrite

ILO

condition

WBEM

status

HPManagementController

Protocol

Metrics

SNMP

status, cpqSm2CntlrBatteryStatus, interfaceStatus, cpqSm2CntlrPendingAlerts, cpqSm2NicCondition, cpqSm2CntlrVideoStatus, cpqSm2CntlrAlertStatus, cpqSm2CntlrKeyboardCableStatus, cpqSm2CntlrRemoteSessionStatus

ILO

status

WBEM

status

HPMemory

Protocol

Metrics

SNMP

status, condition

ILO

status

WBEM

status

HPNetworkInterface

Protocol

Metrics

WBEM

status, BytesReceived, PacketsReceived, PacketsTransmitted, InternalMACReceiveErrors, InternalMACTransmitErrors, BytesTransmitted

HPNetworkInterfaceController

Protocol

Metrics

ILO

status

HPOnboardAdmin

Protocol

Metrics

SNMP

status

HPPhysicalDrive

Protocol

Metrics

SNMP

cpqDaPhyDrvHardWriteErrs, cpqDaPhyDrvDrqTimeouts, cpqDaPhyDrvHardwareErrs, cpqDaPhyDrvWrites, cpqDaPhyDrvOtherTimeouts, cpqDaPhyDrvFormatErrs, cpqDaPhyDrvBadTargetErrs, cpqDaPhyDrvCondition, cpqDaPhyDrvMediaErrs, cpqDaPhyDrvReads, status, cpqDaPhyDrvHardReadErrs, cpqDaPhyDrvNotReadyErrs

ILO

status

WBEM

MediaFailures, ReadOperations, HardwareErrors, SectorsWritten, SectorsRead, OtherTimeOuts, BadTargetErrors, WriteOperations, status, NotReadyErrors, HardReadErrors, DRQTimeOuts, FormatErrors, HardWriteErrors

HPPowerSubsystem

Protocol

Metrics

SNMP

status, currentPowerSupply, condition

HPServerPowerSupply

Protocol

Metrics

SNMP

status, currentPowerSupply, condition

ILO

status

HPProcessor

Protocol

Metrics

SNMP

status

ILO

status

WBEM

status

HPSystemBoard

Protocol

Metrics

SNMP

hoursInService

HPTemperature

Protocol

Metrics

SNMP

status, currentReading

ILO

status, currentReading

WBEM

status, currentReading

HPTemperatureSensor

Protocol

Metrics

SNMP

status, currentReading

Event Management

Event management is handled both actively and passively. Passively, this takes the form of SNMP traps, translations and correlations for which 516 notifications defined in the various HP/Compaq MIBs are included in the form of Event class instances with transforms.

Active Event management revolves around the tracking of the numeric state indicators defined variously according to SNMP, ILO, or WBEM and mapped to corresponding Virtana event severities.

Enclosure

Event

Protocol

Condition

Clear Event

snmp

status is Other or OK

Clear Event

esx

status is Unknown or OK

Info Event

snmp

status is Disabled

Warning Event

snmp

status is Degraded

Warning Event

esx

status is Degraded or Minor Failure or Error

Critical Event

snmp

status is Failed

Critical Event

esx

status is Major Failure or Critical Failure or Non-recoverable Error

Device Bay

Event

Protocol

Condition

Clear Event

snmp

status is Other or OK

Clear Event

snmp

cpqRackServerBladePOSTStatus is Other or Started or Completed

Clear Event

snmp

condition is Other or True

Warning Event

snmp

status is Degraded

Warning Event

snmp

cpqRackServerBladePOSTStatus is Failed

Warning Event

snmp

condition is False or Power Staged Off or Reboot

Critical Event

snmp

status is Failed

Cooling Fan (Enclosure)

Event

Protocol

Condition

Clear Event

snmp

status is Other or OK

Clear Event

esx

status is Unknown or OK

Info Event

snmp

status is Disabled

Warning Event

snmp

status is Degraded

Warning Event

esx

status is Degraded or Minor Failure or Error

Critical Event

snmp

status is Failed

Critical Event

esx

status is Major Failure or Critical Failure or Non-recoverable Error

Temp Sensor (Enclosure)

Event

Protocol

Condition

Clear Event

snmp

status is Other or OK

Warning Event

snmp

status is Degraded

Critical Event

snmp

status is Failed

Interconnect Bay

Event

Protocol

Condition

Clear Event

snmp

status is Other or OK

Clear Event

esx

status is Unknown or OK

Info Event

snmp

status is Disabled

Warning Event

snmp

status is Degraded

Warning Event

esx

status is Degraded or Minor Failure or Error

Critical Event

snmp

status is Failed

Critical Event

esx

status is Major Failure or Critical Failure or Non-recoverable Error

Chassis

Event

Protocol

Condition

Clear Event

snmp

status is Other or OK

Clear Event

esx

status is Unknown or OK

Info Event

snmp

status is Disabled

Warning Event

snmp

status is Degraded

Warning Event

esx

status is Degraded or Minor Failure or Error

Critical Event

snmp

status is Failed

Critical Event

esx

status is Major Failure or Critical Failure or Non-recoverable Error

Cooling Fan

Event

Protocol

Condition

Clear Event

snmp

status is Other or OK

Clear Event

esx

status is OK

Clear Event

snmp

speed is Other or Normal

Clear Event

esx

speed is OK

Warning Event

snmp

status is Degraded

Warning Event

esx

status is Degraded

Warning Event

snmp

speed is High

Warning Event

esx

speed is Degraded

Critical Event

snmp

status is Failed

Critical Event

esx

status is Major Failure

Critical Event

esx

speed is Major Failure

FC Adapter

Event

Protocol

Condition

Clear Event

snmp

status is Other or OK

Clear Event

esx

status is Unknown or OK

Info Event

snmp

status is Disabled

Warning Event

snmp

status is Degraded

Warning Event

esx

status is Degraded or Minor Failure or Error

Critical Event

snmp

status is Failed

Critical Event

esx

status is Major Failure or Critical Failure or Non-recoverable Error

FC Port

Event

Protocol

Condition

Clear Event

snmp

cpqFcaHostCntlrCondition is Other or OK

Clear Event

snmp

cpqFcaHostCntlrStatus is Other or OK

Debug Event

snmp

cpqFcaHostCntlrCondition is Not Connected

Debug Event

snmp

cpqFcaHostCntlrStatus is Not Connected

Warning Event

snmp

cpqFcaHostCntlrCondition is Shut Down or Loop Degraded

Warning Event

snmp

cpqFcaHostCntlrStatus is Degraded or Loop Degraded

Error Event

snmp

cpqFcaHostCntlrCondition is Loop Failed

Error Event

snmp

cpqFcaHostCntlrStatus is Loop Failed

Critical Event

snmp

cpqFcaHostCntlrCondition is Failed

Critical Event

snmp

cpqFcaHostCntlrStatus is Failed

RAID Controller

Event

Protocol

Condition

Clear Event

snmp

cpqDaAccelBattery is Other or OK

Clear Event

snmp

cpqDaCntlrBoardStatus is Other or OK or Disabled

Clear Event

snmp

cpqDaAccelBadData is Other or None

Clear Event

snmp

cpqDaCntlrBoardCondition is Other or OK

Clear Event

snmp

cpqDaAccelStatus is Other or Invalid or Enabled

Clear Event

ilo

cpqDaCntlrCondition is Other or OK

Clear Event

snmp

cpqDaCntlrCondition is Other or OK

Clear Event

snmp

cpqDaAccelCondition is Other or OK

Clear Event

snmp

cpqDaCntlrPartnerBoardStatus is Other or OK or Disabled

Debug Event

snmp

cpqDaAccelBattery is Missing

Debug Event

snmp

cpqDaAccelStatus is Not Attached

Debug Event

ilo

cpqDaCntlrCondition is Not Installed

Info Event

snmp

cpqDaAccelBattery is Recharging

Warning Event

snmp

cpqDaAccelBattery is Degraded

Warning Event

snmp

cpqDaCntlrBoardStatus is Cable Fault or Failed

Warning Event

snmp

cpqDaAccelBadData is Possible

Warning Event

snmp

cpqDaCntlrBoardCondition is Degraded

Warning Event

snmp

cpqDaAccelStatus is Temporarily Disabled or Degraded or Read Cache Unmapped

Warning Event

ilo

cpqDaCntlrCondition is Degraded or Degraded

Warning Event

snmp

cpqDaCntlrCondition is Degraded

Warning Event

snmp

cpqDaAccelCondition is Degraded

Error Event

snmp

cpqDaAccelBattery is Failed or Capacitor Failed

Error Event

snmp

cpqDaAccelStatus is Permanently Disabled

Error Event

ilo

cpqDaCntlrCondition is Major Warning

Critical Event

snmp

cpqDaCntlrBoardStatus is Failed

Critical Event

snmp

cpqDaCntlrBoardCondition is Failed

Critical Event

snmp

cpqDaAccelStatus is Critical Failure

Critical Event

ilo

cpqDaCntlrCondition is Failed

Critical Event

snmp

cpqDaCntlrCondition is Failed

Critical Event

snmp

cpqDaAccelCondition is Failed

Critical Event

snmp

cpqDaCntlrPartnerBoardStatus is Failed

Logical Drive

Event

Protocol

Condition

Clear Event

ilo

status is Other or OK

Clear Event

snmp

status is Other or OK

Clear Event

esx

status is Unknown or OK

Clear Event

ilo

condition is OK

Clear Event

snmp

condition is Other or OK

Warning Event

ilo

status is Unconfigured or Recovering or Ready For Rebuild or Rebuilding or Wrong Drive or Bad Connect or Over heating or Shutdown or Expanding or Not Available or Queued For Expansion or Multipath Access Degraded or Erasing or Predictive Spare Rebuild Ready or Rapid Parity Init In Progress or Rapid Parity Init Pending

Warning Event

snmp

status is Unconfigured or Recovering or Ready For Rebuild or Rebuilding or Wrong Drive or Bad Connect or Over heating or Shutdown or Expanding or Not Available or Queued For Expansion or Multipath Access Degraded or Erasing or Predictive Spare Rebuild Ready or Rapid Parity Init In Progress or Rapid Parity Init Pending

Warning Event

esx

status is Degraded or Minor Failure

Warning Event

snmp

condition is Degraded

Critical Event

ilo

status is Failed

Critical Event

snmp

status is Failed

Critical Event

esx

status is Major Failure or Critical Failure or Non-recoverable Error

Critical Event

ilo

condition is Failed

Critical Event

snmp

condition is Failed

Management Controller

Event

Protocol

Condition

Clear Event

snmp

status is Other or Online

Clear Event

esx

status is Unknown or OK

Clear Event

snmp

cpqSm2CntlrBatteryStatus is Other or OK

Clear Event

snmp

interfaceStatus is Other or OK

Clear Event

snmp

cpqSm2CntlrPendingAlerts is Other or None

Clear Event

snmp

cpqSm2CntlrVideoStatus is Other or Enabled

Clear Event

snmp

cpqSm2CntlrAlertStatus is Other or Enabled

Clear Event

snmp

cpqSm2CntlrRemoteSessionStatus is Other or Inactive

Clear Event

snmp

cpqSm2CntlrKeyboardCableStatus is Other or Connected or Disconnected

Debug Event

snmp

cpqSm2CntlrVideoStatus is Disabled

Debug Event

snmp

cpqSm2CntlrAlertStatus is Disabled

Debug Event

snmp

cpqSm2CntlrRemoteSessionStatus is Active

Warning Event

snmp

status is No Data or Offline

Warning Event

esx

status is Degraded

Warning Event

snmp

cpqSm2CntlrBatteryStatus is Disconnected

Warning Event

snmp

interfaceStatus is Not Responding

Warning Event

snmp

cpqSm2CntlrPendingAlerts is Pending or Remaining

Critical Event

esx

status is Major failure

Critical Event

snmp

cpqSm2CntlrBatteryStatus is Failed

Memory Slot

Event

Protocol

Condition

Clear Event

ilo

status is Other or OK or Not Present or Good

Clear Event

snmp

status is Other or Not Present or In Use or Spare

Clear Event

esx

status is Unknown or OK or Degraded

Clear Event

snmp

condition is Other or OK

Info Event

snmp

status is Present or Added or Upgraded or Unmatched or Partial Use

Warning Event

ilo

status is Degraded or Degraded Module Index Unknown

Warning Event

snmp

status is Missing or Not Supported or Bad Configuration

Warning Event

snmp

condition is Degraded or Degraded Module Index Unknown

Error Event

snmp

status is Degraded

Network Interface

Event

Protocol

Condition

Clear Event

esx

status is OK or Degraded or MajorFailure

NIC

Event

Protocol

Condition

Clear Event

ilo

status is Other or OK

Clear Event

snmp

status is Other or OK

Clear Event

esx

status is Unknown or OK

Debug Event

ilo

status is Disabled

Debug Event

snmp

status is Unknown

Warning Event

ilo

status is Link Down

Warning Event

snmp

status is Link Down

Warning Event

esx

status is Degraded or Minor Failure

Error Event

esx

status is Major Failure or Non-recoverable Error

Critical Event

esx

status is Critical Failure

Onboard Administrator

Event

Protocol

Condition

Clear Event

snmp

status is Other or OK

Clear Event

esx

status is Unknown or OK

Info Event

snmp

status is Disabled

Warning Event

snmp

status is Degraded

Warning Event

esx

status is Degraded or Minor Failure or Error

Critical Event

snmp

status is Failed

Critical Event

esx

status is Major Failure or Critical Failure or Non-recoverable Error

PCI Device

Event

Protocol

Condition

Clear Event

snmp

status is Other or OK

Clear Event

esx

status is Unknown or OK

Info Event

snmp

status is Disabled

Warning Event

snmp

status is Degraded

Warning Event

esx

status is Degraded or Minor Failure or Error

Critical Event

snmp

status is Failed

Critical Event

esx

status is Major Failure or Critical Failure or Non-recoverable Error

Physical Drive

Event

Protocol

Condition

Clear Event

ilo

status is Other or OK

Clear Event

snmp

status is Other or OK

Clear Event

esx

status is Unknown or OK

Clear Event

snmp

condition is Other or OK

Warning Event

ilo

status is Predictive Failure or Erasing or Erase Done or Erase Queued or SSD Wear Out or Degraded (Not Authenticated)

Warning Event

snmp

status is Predictive Failure or Erasing or Erase Done or Erase Queued or SSD Wear Out or Degraded (Not Authenticated)

Warning Event

esx

status is Degraded or Minor Failure

Warning Event

snmp

condition is Degraded

Critical Event

ilo

status is Failed

Critical Event

snmp

status is Failed

Critical Event

esx

status is Major Failure or Critical Failure or Non-recoverable Error

Critical Event

snmp

condition is Failed

Power Supply

Event

Protocol

Condition

Clear Event

snmp

status is Other or OK

Clear Event

snmp

condition is No Error

Warning Event

snmp

status is Degraded

Critical Event

snmp

status is Failed

Critical Event

snmp

condition is General Failure or Bist Failure or Fan Failure or Temp Failure or Inter Lock Open or Eprom Failed or Vref Failed or Dac Failed or Ram Test Failed or Voltage Channel Failed or Orring Diode Failed or Brown Out or Give Up On Startup or Nvram Invalid or Calibration Table Invalid

Power Supply (2)

Event

Protocol

Condition

Clear Event

snmp

status is Other or OK

Clear Event

snmp

condition is No Error

Warning Event

snmp

status is Degraded

Critical Event

snmp

status is Failed

Critical Event

snmp

condition is General Failure or Bist Failure or Fan Failure or Temp Failure or Inter Lock Open or Eprom Failed or Vref Failed or Dac Failed or Ram Test Failed or Voltage Channel Failed or Orring Diode Failed or Brown Out or Give Up On Startup or Nvram Invalid or Calibration Table Invalid

Processor

Event

Protocol

Condition

Clear Event

ilo

status is OK

Clear Event

snmp

status is OK

Clear Event

esx

status is OK

Debug Event

ilo

status is Unknown

Debug Event

snmp

status is Unknown

Debug Event

esx

status is Unknown

Info Event

ilo

status is Disabled

Info Event

snmp

status is Disabled

Warning Event

ilo

status is Degraded

Warning Event

snmp

status is Degraded

Warning Event

esx

status is Stopped

Critical Event

ilo

status is Failed

Critical Event

snmp

status is Failed

Critical Event

esx

status is Predictive Failure or Error

System Board

Event

Protocol

Condition

Clear Event

snmp

status is Other or OK

Clear Event

esx

status is Unknown or OK

Info Event

snmp

status is Disabled

Warning Event

snmp

status is Degraded

Warning Event

esx

status is Degraded or Minor Failure or Error

Critical Event

snmp

status is Failed

Critical Event

esx

status is Major Failure or Critical Failure or Non-recoverable Error

Temp Sensor

Event

Protocol

Condition

Clear Event

snmp

status is Other or OK or Not Installed

Warning Event

snmp

status is Degraded

Critical Event

snmp

status is Failed

Service Impact

When combined with the Virtana Service Dynamics product, this ZenPack adds built-in service impact and root cause analysis capabilities for services running on HP Proliant. The service impact relationships shown in the diagram and described below are automatically added. These will be included in any services that contain one or more of the explicitly mentioned components.

Service Impact Relationships:

  • Device impacts related Enclosure

  • Chassis: Enclosure impacts related Device Bay, Cooling Fan, Interconnect Bay, Power Subsystem, Temperature Sensor

  • Chassis impacts related Processor, Memory, System Board, Temperature Sensor, Cooling Fan, Power Supply, Management Controller, Network Interface, Onboard Administrator, PCI Device, Array Controller, Physical Drive, Logical Drive, Fibre Channel Adapter, Fibre Channel Controller

  • Fibre Channel Adapter impacts related Fibre Channel Port

  • Array Controller impacts related Logical Drive, Physical Drive

  • Device Bay impacts related Chassis

  • Management Controller impacts related Device (iLO)

Data Collection

The HP Proliant ZenPack uses several methods to collect both model and performance data depending on the device types, which fall into 3 categories:

SNMP

Onboard Administrator is HP's solution for hardware management and monitoring. It provides a SNMP interface that Virtana uses to monitor both HP chassis and blade hardware as well as certain OS components with the OA agent installed and configured. SNMP access can also be configured for use with the iLO management interface (via the managment GUI) or with a bare metal install of Linux or Windows by installing HP SIM and related software (including its SNMP subagent)

ILO

Data from ILO devices is provided by a HTTPS interface listening on port 443. The zILOUserName, zILOPassword, zILOUseSSL, and zILOPort zProperties control access to ILO devices. The ILO interface accepts and returns XML-formatted query strings, and this ZenPack features a custom api definining these queries and their output formatting.

WBEM

WBEM support is required for montitoring ESXi on HP Proliant hardware. The zWBEMUsername, zWBEMPassword, zWBEMPort, and zWBEMUseSSL zProperties control access to HP ESXi devices. Data collection relies on the WBEM ZenPack, and also uses a HTTP-based protocol (WBEM). HP provides custom-built ESXi ISOs that provide the HP WBEM software to the base ESXi image.

Endpoint configuration is beyond the scope of this document, but online resources are available at http://h10032.www1.hp.com/ctg/Manual/c04341228

Usage

Adding SNMP Onboard Administrator Device (Enclosure)

Use the following steps to start monitoring the Onboard Administrator using the web interface.

  1. Navigate to the Infrastructure page.

  2. Navigate to organizer Devices/Server/HP/OnboardAdministrator.

  3. Choose Add a Single Device from the add button

  4. Fill out the form:

    • Name can be anything you want.

    • Hostname or IP must be resolvable and accessible from the collector server chosen in the Collector field.

  5. Click ADD.

Adding Linux SNMP or Windows (SNMP) Blade/Rack Device

Use the following steps to start monitoring Linux HP Proliant device using the web interface.

  1. Navigate to the Infrastructure page.

  2. Navigate to organizer Devices/Server/HP/Proliant.

  3. Choose Add a Single Device from the add button

  4. Fill out the form:

    • Name can be anything you want.

    • Hostname or IP must be resolvable and accessible from the collector server chosen in the Collector field.

  5. Click ADD.

Adding ILO Device

Use the following steps to start monitoring a HP ILO device using the web interface.

  1. Navigate to the Infrastructure page.

  2. Navigate to organizer Devices/Server/HP/ILO.

  3. Choose Add a Single Device from the add button

  4. Fill out the form:

    • Name can be anything you want.

    • Hostname or IP must be resolvable and accessible from the collector server chosen in the Collector field.

  5. Click ADD.

Adding ILO Device (SNMP)

Use the following steps to start monitoring a HP ILO device using the web interface.

  1. Navigate to the Infrastructure page.

  2. Navigate to organizer Devices/Server/HP/ILO/SNMP.

  3. Choose Add a Single Device from the add button

  4. Fill out the form:

    • Name can be anything you want.

    • Hostname or IP must be resolvable and accessible from the collector server chosen in the Collector field.

  5. Click ADD.

Adding ESXi Device

Use the following steps to start monitoring a HP ILO device using the web interface.

  1. Navigate to the Infrastructure page.

  2. Navigate to organizer Devices/Server/VMware/ESXi/HP.

  3. Choose Add a Single Device from the add button

  4. Fill out the form:

    • Name can be anything you want.

    • Hostname or IP must be resolvable and accessible from the collector server chosen in the Collector field.

  5. Click ADD.

Combining HP Proliant SNMP monitoring with existing Device Classes

Use the following steps to start monitoring a HP ILO device using the web interface.

  1. Navigate to the Infrastructure page.

  2. Navigate to the desired organizer containing SNMP-monitorable devices, such as Devices/Server/Linux.

  3. Choose Details link above the Devices column

  4. Select the Modeler Plugins menu item

  5. Add the HP.snmp.HPSNMPModeler plugin to the list of Selected plugins

  6. Click Save

  7. Devices in this class should now model and monitor components as they are found

Command Line Tools

Several scripts are included for use with troubleshooting or sample data collection. These scripts reside in the $ZENPACK/ZenPacks/zenoss/HP/Proliant/utilities directory and calling --help will display usage information for each.

  • collect-samples-ilo.sh performs several queries against ILO devices and saves output to individual files

  • collect-samples-wbem.sh performs several queries against WBEM devices and saves output to individual files

  • hp-ilo-samples.py execute given ILO command against remote target and save output

  • hp-wbem-dump.py output list of valid WBEM classnames from ESXi target

  • hp-wbem-query.py execute WBEM query against remote ESXi target

  • hp-wbem-smx-instances perform exhaustive iteration of WBEM classes on remote ESXi target

Installed Items

Installing this ZenPack will add the following items to your Virtana system.

Device Classes

  • /Server/HP/OnboardAdministrator Enclosure monitoring

  • /Server/HP/ILO: iLO device monitoring with iLO

  • /Server/HP/ILO/SNMP: iLO device monitoring with SNMP

  • /Server/VMware/ESXi/HP: ESXi monitoring with WBEM

  • /Server/HP/Proliant: SNMP monitoring with HP MIBs

Configuration Properties

  • zILOPort

  • zILOUseSSL

  • zILOUserName

  • zILOPassword

  • zILOCollectSamples

Modeler Plugins

  • HP.snmp.HPEnclosureModele

  • HP.snmp.HPSNMPModeler

  • HP.ilo.HPILO3Modeler

  • HP.wbem.HPWBEMPlugin

Datasource Types

  • HP ILO Datasource

  • HP WBEM Datasource

  • SNMP Datasource (SNMP)

Monitoring Templates

  • HPArrayControllerILO

  • HPArrayControllerSNMP

  • HPArrayControllerWBEM

  • HPChassisILO

  • HPChassisSNMP

  • HPChassisWBEM

  • HPCoolingFanILO

  • HPCoolingFanSNMP

  • HPCoolingFanWBEM

  • HPDeviceBaySNMP

  • HPDeviceSNMP

  • HPEnclosureCoolingFanSNMP

  • HPEnclosureSNMP

  • HPFibreChannelAdapterWBEM

  • HPFibreChannelPortSNMP

  • HPFibreChannelPortWBEM

  • HPLogicalDriveILO

  • HPLogicalDriveSNMP

  • HPLogicalDriveWBEM

  • HPManagementControllerILO

  • HPManagementControllerSNMP

  • HPManagementControllerWBEM

  • HPMemoryILO

  • HPMemorySNMP

  • HPMemoryWBEM

  • HPNetworkInterfaceControllerILO

  • HPNetworkInterfaceWBEM

  • HPOnboardAdminSNMP

  • HPPhysicalDriveILO

  • HPPhysicalDriveSNMP

  • HPPhysicalDriveWBEM

  • HPPowerSubsystemSNMP

  • HPProcessorILO

  • HPProcessorSNMP

  • HPProcessorWBEM

  • HPServerPowerSupplyILO

  • HPServerPowerSupplySNMP

  • HPSystemBoardSNMP

  • HPTemperatureILO

  • HPTemperatureSNMP

  • HPTemperatureSensorSNMP

  • HPTemperatureWBEM

Component Types

  • Enclosure (HPEnclosure)

  • Chassis (HPChassis)

  • Cooling Fan (HPCoolingFan)

  • Device Bay (HPDeviceBay)

  • Cooling Fan (HPEnclosureCoolingFan)

  • Interconnect Bay (HPInterconnectBay)

  • RAID Controller (HPArrayController)

  • Logical Drive (HPLogicalDrive)

  • Management Controller (HPManagementController)

  • Memory Slot (HPMemory)

  • Network Interface (HPNetworkInterface)

  • NIC (HPNetworkInterfaceController)

  • Onboard Administrator (HPOnboardAdmin)

  • PCI Device (HPPCIDevice)

  • Physical Drive (HPPhysicalDrive)

  • Power Supply (HPPowerSubsystem)

  • Power Supply (HPServerPowerSupply)

  • Processor (HPProcessor)

  • System Board (HPSystemBoard)

  • Temp Sensor (HPTemperature)

  • Temp Sensor (HPTemperatureSensor)

  • FC Adapter (HPFibreChannelAdapter)

  • FC Port (HPFibreChannelPort)

Event Classes

  • /Status/HP

  • /Perf/HP

Required Daemons

Type

Name

Modeler

zenmodeler

Performance Collector

zenpython

 

zenperfsnmp

Upgrade

The HP Proliant Zenpack can be upgraded. To upgrade the ZenPack, install the latest version over the existing one. There is no action for the user to migrate the data. The performance data and events of old ZenPack are retained as per the retain policy settings.

Limitations

Known Issues

Known Issue when monitoring systems with Failover Onboard Administrators

In order to fully support monitoring a system with multiple onboard administrators, the enclosure ip mode option must be enabled and the enclosure must be given a static IP address that is shared between the two onboard administrators. Failure to do so may cause monitoring to stop when the active onboard administrator goes offline.

Known Issue upgrading from 1.0 to 3.0

When upgrading from 1.0 to 3.0, you may see an issue related to catalogs. If you installed the ZenPack from the UI, the UI may report that the ZenPack install failed. As a workaround, install the ZenPack from the command line.

If you observe messages similar to the following:

WARNING zen.catalogservice: Detected catalog inconsistency - to repair, run: 'python /opt/zenoss/ZenPacks/ZenPacks.zenoss.CatalogService-1.0.19.10-py2.7.egg/ZenPacks/zenoss/CatalogService/sync_rids.py'

To finish the installation of this ZenPack, run the script as suggested, then restart the Zenoss instances.

Known issue regarding iLO unknown or N/A values

iLO device modelling reuses the original SNMP components of this ZenPack. Because the iLO does not return exactly the same results as are available via SNMP monitoring, there will be cases where some fields are always set to N/A or Unknown. This is a known issue and is not currently considered a defect.

Known issue regarding iLO monitoring

The servername variable must be set inside the iLO for this ZenPack to work properly with iLO hardware. For iLO 4, go to the iLO GUI under Administration → Access Settings. Under Access Options, update the Server Name field for the iLO server.

Known issue when removing ZenPack

Since this ZenPack depends on the PythonCollector ZenPack for some of its capabilities, the zenpython collector daemon should be stopped manually prior to uninstallation of this ZenPack in order to avoid a traceback error. The zenpython daemon can be restared again once uninstallation is complete.

Changes

3.3.5

  • Make zenpack compatible with twisted 19.10.0 (ZPS-6981)

  • Tested with Virtana Service Dynamics 6.4.1, Virtana Service Observability and Service Impact 5.5.1

3.3.4

  • Add log message to notify when "SERVER NAME" is blank in the iLO (ZPS-6128)

  • Tested with Virtana Service Dynamics 6.4.1, Virtana Service Observability and Service Impact 5.5.1

3.3.3

  • Updated icons with dark-theme compatible versions

  • ILO modeler timeout should follow zCollectorClientTimeout

  • Fix exceptions in impact relationship providers

  • Overhaul impact relationships

  • Hide credentials from logging (ZPS-4338)

  • Add support for ILO Gen 7 physical drives (ZPS-5274)

  • Fix WBEM component metric duplication issue (ZPS-6132)

3.3.2

  • Fix Error modeling Generation 9 iLO hardware when memory/swap info is missing (ZPS-3101)

  • Ensure all NICs are discovered and monitored (ZPS-3645)

  • Correct unrecognized "Good, In Use" status (ZPS-3737)

  • Modeling should succeed when drive enclosure is absent (ZPS-3306)

  • Update Physical Disk possible status values (ZPS-2946)

  • Tested with Zenoss 5.3.3, Zenoss 6.2.0, and Virtana Service Observability

3.3.1

  • Updated unit tests for build platform compatability (ZPS-29150, ZEN-28507)

3.3.0

  • Ported to ZenPackLib 2.0: Extend support to G9 and G10 platforms (ZPS-2086)

  • Add support for Fibre Channel Adapter and Port monitoring (ZPS-2086)

  • Added scripts for debug and data collection of iLO and WBEM clients

  • Revised monitoring template organziation and protocol dependency

  • Added new HP WBEM Datasource and revised all WBEM templates

  • Added multiple new metrics for WBEM monitoring

  • New icons added for device and all component classes

  • WBEM modeling and performance data collection completely revised

  • Additional mappings for SNMP traps (ZPS-679)

  • Fix bidirectional links on Device Status Page (ZPS-745, ZPS-2280)

  • Fix various SNMP modeling tracebacks (ZPS-1621)

  • Fix various issues during upgrade (ZPS-2088, ZPS-2256, ZPS-2257)

  • Fix various ILO modeling issues (ZPS-)

3.2.2

  • Fixed ILO and WBEM modeler plugins to better handle empty results

  • Updated list of recognized XML status strings

3.2.1

  • Updated ILO Protocol handler to address SSL certificate validation issue while preserving HTTP/1.1 compliance

  • Updated XML parser to better handle invalid XML output

  • Fix bad threshold datasource reference in HPArrayController template

  • Corrected type of cycletime parameter

  • Ensure all SNMP templates available to /Devices/Server/HP/ILO/SMNP

3.2.0

  • Added "HP Chassis List" and "HP Device Bay List" reports (ported from HPBladeChassis community ZP)

  • Update Impact documentation (ZEN-25324)

  • Added "zILOCollectSamples" zProperty (raw modeler output storage)

  • Updated zProperties (removed zProperty zILOSSLPort, replace with zILOUsesSSL and zILOPort)

  • Adjust component labels and titles for DynamicView (ZEN-25248)

  • Add ResetClassRelations migrate script (ZEN-25281)

  • Sensible thresholds for Temperature graphs (ZEN-25260)

  • More informative event logging (ZEN-24270)

  • Ensure that ILODatasource, HPStatusThreshold, and GUI use shared reference data

  • Updated graph datapoints and templates (ZEN-24886)

  • Improved datasource/datapoint access methdos for status string retrieval (ZEN-24277)

  • Fix traceback during WBEM modeling (ZEN-24786) updated and new unit testing to handle ILO datasource/modeler testing (ZEN-24810)

  • Revised and simplified all string-to-integer status mapping code

  • Added new ILOProtocolHandler ILO client (ZEN-25277)

  • Added new ILOXMLParser for shared access to ILO output

  • Added cross-reference functions to HPDevice and HPChassis (get_ilo, get_cpus, get_mem)

  • Revised shared component attributes for consistency

  • Remove/migrate code from utils.py and ErrorHandler.py

  • Added ManagedObject base to contine shared component code

  • deprecate ILO2 support (ZEN-22388)

  • Rewrite of all modeler plugins

  • Added HPPluginBase, HPSnmpPlugin to contain shared plugin code

  • Prevent premature data collection termination (ZEN-24277)

  • Add support for RAID controller monitoring via SNMP and ILO (ZEN-24211, ZEN-23388)

  • Fix HPStatusThreshold issue causing data collection to stop (ZEN-23040)

  • Improved handling of components with blank id (ZEN-22872, ZEN-22851, ZEN-21680)

3.1.1

  • Updated zenpacklib.py to address modeler timeouts from attribute-indexed components

3.1.0

  • Numerous enhancements and bugfixes

3.0.2

  • Fixes ZEN-18151. HP.wbem.HPWBEMPlugin modeling traceback on tempStatus()

3.0.1

  • Standardized commonly used code for converting RRD status to formatted string

  • Standardized most component properties to those common between collection methods (some still outstanding)

  • Standardized common datasource/datapoint names throughout templates

  • Simplified customized classes to reduce duplication of function

  • Removed duplicate property definitions with similar names but identical values

  • Removed most component properties that should have been performance or status datasources

  • Standardized unit testing for modeler plugins and class loading

  • Folded deviceMap modeler plugins into component modeler plugins, removed device map plugins

  • Added ZenPack class to remove all RRD templates upon ZP uninstall

  • Changed Fan Speed to display status string (only ILO shows actual speed)

  • SNMP modeler plugins are now appropriate for use wherever SNMP is supported

  • Deprecated HPOSHypervisor and HPThermalSubsystem components

  • Removed COMMAND datasource from HPLogicalDrive (used snmpindex_dct instead)

  • Added ILO datasource queries for HPChassis, HPProcessor, HPManagementController, HPNetworkInterfaceController, and updated ILODatasource to accommodate

  • Added "perfID" to ILO HPTemperature components to correct some missing datapoints

  • Updated migrate methods to remove all templates before re-adding from YAML

  • Removed "uptime" metric from HPSystemBoard since it duplicates the Device sysUptime datapoint

  • Added several status indicators to HPManagementController component (SNMP)

  • Re-based SNMP Device templates on /Server/Linux to ensure consistency

3.0.0

  • Major rewrite of significant sections of the zenpack to use zenpacklib and remove obvious defects.

1.0.0

  • Initial release.