Forum Discussion

mnagel's avatar
mnagel
Icon for Professor rankProfessor
8 years ago

device model templates

AD is nice, but sometimes AD leads to missing warnings for elements not present but that probably should be.  Having a minimum required device template of some sort could help avoid this, but I am not sure (yet) of a good approach here that doesn't end up with a lot of error-prone manual effort.  I do know the fact that this missing power supply in a stack is not reported as a problem is wrong, though.

FAN 1 is OK 
FAN 2 is OK 
FAN PS-1 is OK 
FAN PS-2 is OK 
SYSTEM TEMPERATURE is OK 
System Temperature Value: 28 Degree Celsius 
System Temperature State: GREEN 
Yellow Threshold : 46 Degree Celsius 
Red Threshold    : 60 Degree Celsius 
 SW  PID                 Serial#     Status           Sys Pwr  PoE Pwr  Watts 
---  ------------------  ----------  ---------------  -------  -------  ----- 
1A  C3KX-PWR-715WAC     LIT15300JR3 OK              Good     Good     715/0 
1B  C3KX-PWR-715WAC     LIT15300JQ8 OK              Good     Good     715/0 
2A  C3KX-PWR-715WAC     LIT15300JML OK              Good     Good     715/0 
2B  UNKNOWN             unknown     No Input Power  Bad      N/A      0/0 
3A  C3KX-PWR-715WAC     LIT15300JQY OK              Good     Good     715/0 
3B  C3KX-PWR-715WAC     LIT15300JSY OK              Good     Good     715/0 

 

Thanks,

Mark

 

 

2 Replies

Replies have been turned off for this discussion
  • Hi Mark -

    If I understand your situation correctly, this use-case could be addressed with a PropertySource targeted at the specific device class coupled with a singleton (non-AD) DataSource designed to collect data from each of the expected FRU modules.

    In this case, you'd write a PropertySource that would specifically identify the device in question and and assign a targeted system.category (e.g. "CiscoCatalyst3750"). Then you could write/modify a single-instance DataSource -- which apply only to "CiscoCatalyst3750" --  that would specify as datapoints from each of the known FRUs in a 3750. By having the datapoints alert on a NoData condition, any C3750 with a FRU that wasn't responding would trigger an alert as expected.

     

  • Yes, something like this could work.  I was trying to think of something that did not include datasource applies-to-partitioning since I find that to be a source of serious headache -- difficult to maintain, difficult to know what applies to what without deep diving, and due to lack of inheritance it is too easy to create semantic divergence where there should not be any.  You are probably right that this is all I will be able to do at this time, was just trying to think of a method that didn't further encourage that methodology.

    Thanks,

    Mark