RE: [hobbit] curves went to zero but didn't have a report?

["Hubbard, Greg L" <greg.hubbard (at) eds.com>]

Thanks, Steve.  Thinking further, another guess is that the node got rebooted (due to a crash or some other action) and the Hobbit client did not restart.  There should have been purple alarms.


________________________________

	From: sholmes42 (at) gmail.com [mailto:sholmes42 (at) gmail.com] On Behalf Of Steve Holmes
	Sent: Tuesday, June 24, 2008 10:04 AM
	To: hobbit (at) hswn.dk
	Subject: Re: [hobbit] curves went to zero but didn't have a report?
	
	
	[bottom]
	
	
	2008/6/24 Andrew Chen <Andrew.Chen (at) ehealth-china.com>:
	

		Hi Hubbard,

		        Thanks for your reply.  I have checked the conn, it  only stopped 10m(13:45~13:55).  I  want to know if the hobbit can check the server's CPU and Memory when they Zero. If I can recevie some email from hobbit. So I first know it. Can you give me some advice for hobbit setting? I am learning hobbit:-)

		
________________________________


		From: Hubbard, Greg L [mailto:greg.hubbard (at) eds.com] 
		Sent: 2008年6月24日 21:26
		To: hobbit (at) hswn.dk
		Subject: RE: [hobbit] curves went to zero but didn't have a report?

		 

		Looking at this graph, there is always some memory utilization and some CPU utilization (seen via load average).  I think the real problem is that something went wrong with the agent on this machine, or it could no longer communicate with the Hobbit server during this time period. You should review the alert history for this node and then make sure that you configure alerts appropriately.  I suspect that you will find that all of the agent-based tests went "purple" at about 1:30.  The other thing to check is the "conn" column -- if the system was not reachable, then the purple alarms for the agent-based tests would have been suppressed in favor of the reachability alarm.

			 

			
________________________________


			


	I think the answer to your question is: not without some programming. I.e. you are looking for a parameter in hobbit-client.cfg similar to the PROCS config with which you can specify both a minimum and a maximum allowable number of processes, but in the memory case you want to be able to specify a minimum as well as a maximum on memory usage (ditto for cpu).  

	To do this you would have to write your own script. Note, this says nothing about whether the usage *really* is dropping to zero or not. I think Greg is probably right about the agent. 

	Steve