Using System Insight Manager (SIM) through integrated Lights Out (iLO) technology only

[chuckk281]

Nick had a question on management software:

*************************

Does anyone know of a (preferably customer viewable) document that lists which features of SIM and associated plug-ins (such as PMP) will work when the SIM server can only see the server through its iLO port.

I have found a document that tells me I can get asset/inventory information, and fault monitoring via SNMP forwarding, but the document only says “not all management features are available”. We documentation to say which, if any, other SIM and plug-in features will work.

I’ve a further 2 questions on SIM / Insight Control that I’m struggling to find any, or any up to date, documentation for;

SIM vs Microsoft System Center Operations Manager

-          Do we have a document anywhere that compares the features available when monitoring HP servers with either SIM of SCCOM with the HP management packs loaded?

How to manage non-HP x86 Windows servers with HP SIM

-          Do we have a more up to date version of this whitepaper, the only one I can find is dated 2006. We are specially looking to find out what we can manage/monitor on Dell servers and blades in SIM. The customer is moving to HP, but has a lot of existing DELL servers that will be in production for a while yet, and wants to monitor them from a single platform, including, if possible, extended monitoring such as performance.

******************************

David has some words of wisdom:

********************************

The first thing to understand is what the customer is trying to achieve.  If it’s concerns around SNMP and security, those usually can be addressed to a customer’s satisfaction or they can move to the ProLiant WBEM (WMI) providers.  If it’s concerns around software maintenance, the agent components and SMH are a small component of the ProLiant Support Pack, and important drivers (hp-health, etc) and utilities (ACU, IML viewer) are things that every system really needs, so no gains there.  Also, if there’s any management product in use other than HP SIM (e.g. Microsoft System Center, HP Operations Manager), they expect to be able to talk to the host itself and mine the data instrumented there and aren’t intelligent enough to take a vector off a management processor and infer information from that source and associate back with the host.

Unfortunately, there’s not a definitive document on this, and what there has been is not up-to-date, so I won’t copy it here.  Let me net it out for you if I can…

1.        It depends on the version of iLO

• Lights-Out 100 has no abilities in this area.
• iLO only acts as a proxy for SNMP traffic to/from the agents in the OS and can forward SNMP traps that originate from the OS agents.
• iLO2 added a basic subsystem status page and got the ability to originate some SNMP traps on its own related to infrastructure (e.g. fan failure, thermal events)
• iLO3 is adding more SNMP traps, although it may not be reflected in the currently available version of the firmware.  In most cases, however, the information available in the traps and on the iLO3 web interface will be limited to telling you that something is wrong but may not have the detail about WHAT is wrong (e.g. “Memory Error” as opposed to what the agents/providers might report as “Total Correctable Memory Errors = 34” which in the latter case may not require a service incident).  Some additional detail might be available by examining the IML from iLO.  Information that originates with an OS counter (e.g. NIC link down) isn’t available and no events can be sent.

2.       Without agents on the host, HP SIM will only show up/down (Normal/Critical) status in the HW column based on the response of the system to a ping.   The status conditions reflected by the ProLiant Status Array OID for Major or Minor won’t apply.  Since SNMP trap delivery isn’t guaranteed, this is an important backup status method and a system could respond to a ping and show green but be operated in a degraded condition that needs attention (e.g. drive failed in a mirror and now the lone remaining drive is a SPOF) but no one would notice.  Since even iLO3 doesn’t have total visibility to devices connected to a SmartArray controller, this is a very real concern.  To digress for a second, iLO3 can reflect individual drive status in the web interface, and that status is based on what the LED on the drive is showing; iLO3 is able to gather this information because while it can’t see into a SmartArray controller, the drive backplane gets the LED condition from the drive and a smart cookie engineer figured out that he could see that from iLO3…they really do make me proud sometimes…

3.       Usage of Insight Remote Support and Insight Remote Support – Advanced isn’t possible by just talking to iLO and requires the presence of the agents or providers.

4.       The only feature of Insight Control that can function by talking only to iLO is power management; all other features require the ability to communicate with agents or providers on the host.

5.       Detailed hardware configuration information is not available from iLO (e.g. drive model, serial number and firmware).

This table from a presentation delivered last Spring at the WW Ambassador Summit details some of the capabilities and caveats:

Function	Description	Available
Embedded device status, alerts, & configuration	Alerts, devices status and device configuration information for all major embedded subsystems: embedded Smart Array & drives, memory, CPU, temp, power supply, fans.	Yes
Core health driver functions	Graceful shutdown, Auto Server Recovery	Yes
Pre-failure alerts (SNMP and WBEM)	Alert on drive, memory, CPU issues prior to failure	Yes
Stand-up device status, alerts, & configuration	Alerts, devices status and device configuration information for HP branded PCI options: array controller and drives, NICs, Fiber channel HBA’s, etc.	No
Core OS data	Server name, OS name and version, IP address, etc.	Host name, IP Address
Firmware and driver info.	Name and version number of HP firmware and drivers installed on each system	Limited firmware
OS Event Log integration	Hardware events written to OS event log to facilitate scraping by management applications	IPMI data only
In-band and out-of-band alerting	Ability to send alerts either from the iLO IP address or the production NIC IP address	No (alerts from iLO only)
Basic performance data	Disk, CPU, and memory utilization statistics over last 15 polling cycles (typically 30 minute interval) and ability to set utilization thresholds.	No
Diagnostic data	Disk read and write errors, network packet collisions, etc.	No
Browser-based user interface	Browser-based user interface for viewing systems management data.	Yes, accessed via iLO IP address (not server IP)

****************************

I hope this helps. ANy other input and comments?