VM /ETC

« Hello My VMworld 2009 Photos

VMworld 2009 Booth Talk – NetApp Rapid Clone Utility »

VM3463 – Monitoring Hardware Health With vCenter 4

September 3rd, 2009 |

Author: Rich Brambley

This VMworld 2009 session took place Thurs at 9:30 am in room 134

Points made by the presenter worth remembering.

Physical failure is unavoidable, and an ounce of prevention is worth a pound of cure
There is a 50% chance that pieces of an ESX Cluster will fail and take down critical services and servers.
You’re not usually staring at a monitoring screen, and you want to be notified as the hardware degrades not afterwards.
You want as much hardware info about a host, from multiple different vendor platforms, and on a single screen
Physical failure is a fact of virtual life
Be proactive about hardware failure and use DRS + hardware monitoring + Alarms

An interesting demo in this session showed the use of the built in vCenter 4 host hardware temperature status alarm generating SNMP traps as well as automatically putting a host in maintenance mode so an administrator can investigate. This action instigated a VMotion evacuation of the VMs on the impacted host and effectively isolated the hardware issue in the environment with minimal or zero impact.

My key take away of this session is that numerous “out of the box” vCenter event based alarms can be leveraged during the warning phase of hardware failures. This includes alerts covering power, fans, cpus, memory, batteries, etc. The ESX host hardware monitoring feature is detected and available automtically in vSphere 4.

My notes:

New Hardware Status Hardware Monitoring service in vCenter 4.0

Trigger alarms on hardware health changes
Expose more sensor information at the VC level
- Storage profile
- Firmware info
Support industry standards
- CIM SMASH
- IBM, HP, DELL, LSI, etc out of the box
Separate from the VIM release cycle
- Tomcat webapp – no load on vpxd
- Not dependent on VIM API changes
HSM – hardware status monitoring service
CIM Server SFCBD /Pegasus
CIM provider

Alerts on hardware health changes

Service divides into various hardware groups
Detect changes to health state of individual sensors
Service maintains group health
Posts event to VC on HW group health change
Triggered alarms execute remediate/alert actions
- SNMP traps
- Enter/exit maintenance mode
- Enter/exit standby mode
- Power off/reboot hos

Demo Video hardware health monitoring system and alerts

2 ESX hosts and vCenter 4 managed
Temperature of a host triggers an alarm so you can take preventative action to prevent downtime
VC is configured to send SNMP traps
Receiving traps on trap receiver
Host hardware temperature status alarm is shipped as a default in vCenter
- Actions send notification trap
- Actions enter maintenance mode when goes fom green to yellow state
- Actions exit maintenance mode when alarm returns to green normal state – this may not be a desirable action as automated. You may want to perform additional steps and exit MM manually

New hardware status tab in vCenter

Get old view if use vSphere client direct to host
New tab / view from vCenter
- Sensors Update every 5 mins
- You can reset sensors
- Print and export page in xml format
- All alerts and warning and details about them are stored on this page
- System event log with drop down filter for all, informational, or just alerts and warnings

More Details exposed

16 groups –bios, cpu, ram, fan, voltage, temp, power, sys board/chassis, network, & more
10 hardware alarms
90 sensor types/classes
160 sensor properties types/classes

CIM Model (Common Information Model)

Db-like modeling language
Supports query modification language
Xml-based transport called cim-xml (over https)
Industry standard supported by many hardware vendors – Dell, HP, IBM, LSI & more

CIM Monitor

Monitors every host in inventory
Calculates / rolls up any changes in status states
Client interaction is secure. User needs to possess the System.view privilege to see

VM /ETC

VM3463 – Monitoring Hardware Health With vCenter 4

Related Posts