monitoring services, the mom 2005 agent way…

the net effect of how this topic came up was an email i received over the weekend on some monitoring i setup.  the user basically said … hey this is a great alert!  now how do i tell when the service comes back up?

always wanting more.  never satisfied.  anyway, i started looking into this only to find that it wasn’t nearly as intuitive as it should have been.  if i wanted to deconstruct the entire thing and create it myself, that’d have probably worked…but i wanted to understand how the agent did this stuff.

if you look at the agent properties (under global settings), you’ll notice that there’s a “service monitoring” tab.  you can use this tab to specify the check and reporting intervals.  i’m not going to go into detail on those values since really, this post is about much cooler (boring?) stuff.

i’m not sure what it executes under the covers, but i do know that the event id is ALWAYS 21207.  so what is different?  what changes in the events to make the state switch from active to inactive or back to active again.  glad you asked.  here’s a few tables to help you understand the values that’ll come into parameters (which by the way, are not exposed in the events).

 

event & parameter details:

event parameter	value
event id	21207
source	microsoft operations manager
parameter 1	# of seconds from last sample
parameter 2	display name
parameter 3	old state (string)
parameter 4	new state (string)
parameter 5	service name (short name version)
parameter 6	start up type (string)
parameter 7	service context (user id)
parameter 8	old state (numeric)
parameter 9	new state (numeric)
parameter 10	start up type (numeric)

 

service state details:

state	value
stopped	1
start pending	2
stop pending	3
running	4
continue pending	5
pause pending	6
paused	7

 

service startup details:

start type	value
automatic	2
manual	3
disabled	4
unknown	-1

 

based on that, now you can build your own alert conditions by modifying the event rule.  this one, for example, was created by the management pack wizard.  here’s the formula for the state alert:

AttributeValue(Parameter 10) = 2 AND (AttributeValue(Parameter 9)= "1" OR AttributeValue(Parameter 9)= "3")

now, we deconstruct this to see what all of these mean:

• parameter 9 – new state
  • 1 – stopped OR
  • 3 – stopped pending
• parameter 10 – start up type
  • 2 - automatic

so in human language (or so i purport!): if the service is an automatic service and is in the state “stopped” or “stopped pending”, then raise a critical error alert.

don’t forget the rest of your criteria:

• source: Microsoft Operations Manager
• event id: 21207
• parameter 5: short name of the service.