Generating a ticket for site down rather than device

Hi Dele,

if you have Dynamic Model license: depending on your needs, you can create a site class, with a relationship with devices (you will have to "build" this topology link between a device and a site : at discovery time, through inventory integration, ...). Then you can automatically count the number of devices down (or unresponsive) in the site. Or propagate a Site::Down problem to device Unresponsive.

HTH,

--Fred

== Monitor your Smarts environment using APG ReportPack for Smarts health ==
Frederic Meunier
Solutions Watch4Net Inc
APG & Smarts InCharge integration
http://www.watch4net.com

fred,
thanks you but how can i use dynamic modelling license to create a site with links to device and how does it develop the link between site and device also the discovery through inventory propogation
can i have like a sample as i have use dynamic modelling once how do i define the site and then also the devices and then form a relationship

dele

Hi dele,

(1) I don't recommend to refine existing problem or propagate new problem to existing symptoms except if you really know what you are doing, so lets assume that you want a brand new event which will raise at the same time as the UnitaryComputerSystem::Down,

(2) disclaimer : this may not be the best solution for you as it may depend on other needs / constraints. I'm not responsible for any issues with the model here. This is an example. You may consider aggregate instead of a new event.

(3) the model:

refine interface UnitaryComputerSystem {
// translate the IsUnresponsive to an int so we can sum them easily
computed attribute unsigned short IsUnresponsiveToInteger = ( IsUnresponsive ? 1 : 0 );
}

interface Site : ICIM_Collection {
// automatically count the number of devices in the site
computed attribute unsigned short UCS_Total = | UnitaryComputerSystem(ConsistsOf) |;
// automatically count the number of devices not responding in the site
propagate attribute unsigned short sum UCS_Unresponsive
<= UnitaryComputerSystem, ConsistsOf, IsUnresponsiveToInteger;

// new event when all devices are down (you may use a threshold)
event AllUCSInSiteDown =
( UCS_Unresponsive == UCS_Total );
export
AllUCSInSiteDown;
}

a really simple one, everything is done automatically using existing attributes and relationships.

(4) this model doesn't need the "problem" license so the simple dynamic model license should be enough to compile and use.

(5) the only thing missing is the creation of the topology. They are multiple solutions to do so, depending of your env. If the name of the device itself gives the site information, you can create the site in the discovery post process. If you have this information in an external repository, you can build an interface with it. ...

(5') for example : assign devices (which inherit from UCS) to Site using the ConsistsOf relationship, ex:

dmctl insert Site::Site1::ConsistsOf Switch::S1

and let the MODEL do the job!

HTH,

--Fred

== Monitor your Smarts environment using APG ReportPack for Smarts health ==
Frederic Meunier
Solutions Watch4Net Inc
APG & Smarts InCharge integration
http://www.watch4net.com

Message was edited by:
FredMeunier

fred,

Thanks this is useful and would try this out with my colleague
but wondering what syntax goes into the { when creating an event for e.g UnitaryComputerSystem and _Collection {
have you any more document on dynamic modelling or MODEL
are you based in the US as the time slot to my reply seems to be from the US or canada

Dele

Fred,

Another question is that the scenario has changed slightly
we need to create a different scenarion on this site modelling

IT_NoOfComponents would need to be a count of all the switches & routers etc in each Site.

- IT_NoOfFaultComponents would need to be a count of all the devices which are marked as down (Using Is_Unresponsive?? From each device maybe?))

- We then generate two different events

o IT_SiteDown ¿ Where all the devices in the site are marked as down (based on the above 2 attributes).

o IT_SiteAtRisk ¿ Where one or more of the devices are marked as down

(based on the above 2 attributes).

Thanks for your quick response
dele

Hi dele,

not sure to understand your request "what syntax ..." ... can you elaborate ?
To answer the other questions:
- no I don't have more document on dynamic model/MODEL than the (old) official ones. I know DynModel/MODEL because I did several integration using them. I'm an old fan of Smarts MODEL even if I did have some issues with Dynamic MODEL since SDK's EOL,
- I'm EDT.

Regarding the "different scenario" I don't understand your question. I think you have everything in the MODEL I gave. Just change the UCS_Total by IT_NoOfComponents and UCS_Unresponsive by IT_NoOfFaultComponents and you have it. If the question is how to generate 2 different events, it will depend if you want them at the same time (AtRisk and Down) or not (i.e. Down without AtRisk). You can use a boolean for this, at the event declaration.

--Fred

== Monitor your Smarts environment using APG ReportPack for Smarts health ==
Frederic Meunier
Solutions Watch4Net Inc
APG & Smarts InCharge integration
http://www.watch4net.com

Fred,

i am also seeking for information on the event handling whereby we have two events At Risk and Down or not ( Down without AtRisk) You said i should use Boolean i am not too sure of this .
Also this file has to be created in a directory smartsDIR /repos and then compiled and loaded unto the domain manager.
Thanks

Dele

Dele,

I gave you (equivalent of Down) :

event AllUCSInSiteDown =
( UCS_Unresponsive == UCS_Total );

if you need to have Down without AtRisk, simply define a logical operation which will make those 2 events exclusive, something like (using a boolean or directly in the event definition):

event AtRisk =
( UCS_Unresponsive >= 1 ) && ( UCS_Unresponsive != UCS_Total );

As for the model, you have to compile it and put the .ldm in the /model directory. You can try to load the model at runtime although I don't recommend it (I did experience DM issues after that - core or other non desirable side effects, multiple times).

My 4 cents,

--Fred

== Monitor your Smarts environment using APG ReportPack for Smarts health ==
Frederic Meunier
Solutions Watch4Net Inc
APG & Smarts InCharge integration
http://www.watch4net.com

Fred,

you are a star in a million
thanks i have learnt a lot
but i also wanted so also say that is it possible to create selective groups in SAM with various pattern matching(Siteid) and priority
and also add the devices to this groups
WILL This work as another option.
there is also polling group in IP AM/PM i dont know if this would seem like an alternative

dele

Hi Dele,

I use the built in SystemRedundancyGroup in AM to do this for our "High Availability Network" locations. I use the Perl API to insert a group and link the two or more devices to it. The analysis is then out of the box. Here's a snippit of the code to do it (the indents are goofed up using this web form...):


$deviceid = This is one of your devices that you want to be part of the group...
$rgroup = This is the name of the group you want to create...

$rgclass = "SystemRedundancyGroup";

$rgroupsmarts = "$rgclass\:\:$rgroup";
$objref=$session_am->object($deviceid);

$found = 0;
foreach $ul ( $objref->get(PartOf) ) {
print " Existing: $ul\n";
if ( $rgroupsmarts =~ /^$ul$/ ) {
$found = 1;
}
}

if ( $found == 1 ) {
print " Found existing link to: $rgroupsmarts\n";
} else {
print " No link found to $rgroupsmarts, checking for existance of $rgroupsmarts\n";
$parentexists= eval{$session_am->object($rgroupsmarts)->isNull();};
if ( $parentexists == 1 ) {
print "$rgroupsmarts does not exist, creating...\n";
$session_am->create($rgroupsmarts);
$objrg = $session_am->object($rgroupsmarts);
}
print " Linking: $rgroupsmarts\n";
$objref->insertElement( "PartOf", "$rgroupsmarts" );
}


Karl

Dele,

yes you can create selective groups in SAM, but by default, you won't have any event attached specifically to this groups. So you won't be able to open tickets directly on a "site". You will only be able to see groups with the group view, and see the highest severity of the group's notifications.
Polling groups in IP DM is another story. You can group devices/components to apply polling/threshold policies but, again, you won't have any event attached to these groups.
Aggregate notifications may be an option, but it won't be easy to aggregate the right notifications for your requirements.
SystemRedundancyGroup (as Karl mentioned) is another option, although the concept here is not SystemRedundancy but "group of devices". You may inherits a Site from SystemRedundancyGroup, and it will work. Using a "real" Site class can be useful in the topology browser.

HTH,

--Fred

== Monitor your Smarts environment using APG ReportPack for Smarts health ==
Frederic Meunier
Solutions Watch4Net Inc
APG & Smarts InCharge integration
http://www.watch4net.com