SNMP OIDs w/ non-accessible indexes, Cisco Environmental MIB

[jjason]

Hello,

My original goal was to simply monitor the air intake temperatures of my Cisco routers to make sure the IDF rooms are not overheating and cooking the equipment. (We had one building shutdown because of this.) Little did I know I was going to touch upon several outstanding feature requests for Cacti.

Here are the items that I have run across:

• 1. SNMP OIDs with non-accessable indexes.
  2. Cisco router OIDs may change on reboot - how do you detect the reboot and redo the indexes?
  3. Spikes in the graphs due to reboots.

SNMP OIDs with non-accessable indexes:

This is the one that I cannot solve. The in both the entity and environmental MIBs, the indexes are not accessible. This subject has been covered in the following thread, but it doesn’t appear to be applicable to the latest versions of Cacti.

• http://www.raxnet.net/board/viewtopic.p ... mperatures

Here is the link to the official Cisco doc on the subject of environmental monitoring:

• http://www.cisco.com/en/US/tech/tk648/t ... 4b9e.shtml

And the MIB entries that are driving me to write this posting – note the “ACCESS not-accessible” on both in RED:

From "CISCO-ENTITY-SENSOR-MIB.my"

• entSensorValueEntry OBJECT-TYPE
  SYNTAX EntSensorValueEntry
  ACCESS not-accessible
  STATUS mandatory
  DESCRIPTION
  "An entSensorValueTable entry describes the
  present reading of a sensor, the measurement units
  and scale, and sensor operational status."
  INDEX { entPhysicalIndex }
  ::= { entSensorValueTable 1 }
  
  entPhysicalIndex OBJECT-TYPE
  SYNTAX PhysicalIndex
  -- Rsyntax INTEGER(1..2147483647)
  ACCESS not-accessible
  STATUS mandatory
  DESCRIPTION
  "The index for this entry."
  ::= { entPhysicalEntry 1 }

From "CISCO-ENVMON-MIB.my"

• CiscoEnvMonTemperatureStatusEntry ::=
  SEQUENCE {
  ciscoEnvMonTemperatureStatusIndex Integer32 (0..2147483647),
  ciscoEnvMonTemperatureStatusDescr DisplayString,
  ciscoEnvMonTemperatureStatusValue Gauge32,
  ciscoEnvMonTemperatureThreshold Integer32,
  ciscoEnvMonTemperatureLastShutdown Integer32,
  ciscoEnvMonTemperatureState CiscoEnvMonState
  }
  
  ciscoEnvMonTemperatureStatusIndex OBJECT-TYPE
  SYNTAX Integer32 (0..2147483647)
  MAX-ACCESS not-accessible
  STATUS current
  DESCRIPTION
  "Unique index for the testpoint being instrumented.
  This index is for SNMP purposes only, and has no
  intrinsic meaning."
  ::= { ciscoEnvMonTemperatureStatusEntry 1 }

Cisco router OIDs may change on reboot

Another very annoying Cisco characteristic is that Cisco MIBs may change on reboots. This is most common if you add/subtract blades in a Cisco router. This has been covered well in the following link:

• http://www.raxnet.net/board/viewtopic.p ... +backwards

raX identified the two solutions (quoted below) and hinted that we would see the solutions as early as v0.8.6 in a note dated 6/23/2004.

• "1) When the device's uptime goes backwards
  2) When the number of interfaces in a device changes"

Spikes in the graphs due to reboots

While this is a little annoying, I can live with it. This fix works for now: "The only way to avoid graph spikes is setting correct maximum value in data source settings (which is in most of cases the maximum speed of the link in bytes/sec). Rrdtool will just disregard any value above it and you will not have spikes in case of device reset. " (http://www.raxnet.net/board/viewtopic.p ... ght=spikes and item #4 on the Cacti FAQ in the Graphs section.)

Summary and Request for Help

I would greatly appreciate some help here since my programming skills are not up to the task. If bribery of any sort helps, please let me know.

And in case I forget to mention it anywhere else, Cacti rocks!

Thank you!

Yours,
Jason

[Guest]

Jason,

raX is working on this for 0.8.6.

TheWitness