"Non-Alerting" neutral check mode?

[frilby]

As I use HostMonitor for more complex checking of systems I am wanting to setup checks that rely on the status of other checks (in the usual dependancy fashion - either as direct dependancies or as expression based dependancies).

The problem here is that in a lot of cases with these more complex checks, I do not want the underlying "sub-checks" to affect the status of HostMonitor in any way (i.e. I need them to be transparent and be there simply to allow the status fo these checks to form the basis for a more complicated check).

It is the top level check that I wish to derive alerts, etc from, and not the underlying checks used to determine the overall status.

One or more of these sub-checks could be in error most of the time, but the status of these checks is not important in isolation (and should not affect HostMonitor).

I hope this makes sense?

Ideally a check box would possibly be the best option. This would effectively disconnect the check from the normal "bad" or "unknown" status conditions inherent with all HostMonitor checks (similar to un-checking the "treat unknown status as bad" checkbox - e.g. a "do not alert" or "basic check only" checkbox).

Is this possible?
Dows anyone else see a way around this with the current features of HostMonitor?

Cheers,
Frilby

[KS-Soft]

I think you may move such tests into designated folder and mark 2 options in Folder Properties dialog
- Use folder's own report settings
- Test status should not affect tray icon color
In this case tests will not appear in reports and HostMonitor will not change icon to red when test fails.

Regards
Alex

[frilby]

Alex,

Thanks for the quick reply. This solution does work, but only for the tray icon and not for the visual tree view in the console, etc.

We use the console as our monitor screen, and even with the options you mentioned set, the folders, etc still go red on a bad (or unknown) result.

I am hoping there may be the possibility of a check box or feature (even a folder based setting similar to the "Test status should not affect tray icon colour" that you mentioned, but one that prevents any colour change to the folders, etc (I wouldn't mind if the test colour changed if that's easier, but from a support perspective it's not good to have folders "in the red" all the time as you tend to get used to it and not realise that other things have gone into error.

I hope this clarifies it a little better, and hope there may be an easy solution to this in a future release...

Thanks again for your help Alex...

Cheers,
Frilby

[KS-Soft]

I think we can mplement such option. I have added your request into "to do" list. Medium priority.
The most difficult problem - how this option should be named?

Regards
Alex

[frilby]

Many thanks Alex... Looking forward to the feature being released.

And yes, I agree that naming it may be a problem as I could not think of an appropriate term either.

The best I can think of is a "Turn off visual alerting" checkbox ???

[genasea]

I posted a request on Nov. 17th (see 'Change Status feature') requesting something that is similar to this and could be used for this request as well. I am also requesting that a feature be added to allow the operator to decided when a test goes into alarm. Right now, a test goes into alarm (bad, unknown, etc.) upon the first occurance of passing its alarm threshold. I asked if this could be changed to say allow 5 CPU tests that are above 90% (say 5 minutes in a row) prior to going into alarm (setting test to bad). That way many of the performance based alarms that occur are not triggered the second a test fails.

This same request could be used for your request as well. The feature would have an area that would allow the user to enter how many times the test would be beyond the threshold (or not responsive to ping tests as an example) prior to going into alarm (In your case it could be bad forever, and you do not wish to be alarmed so you could set this to '0' to never alarm).

Just a thought.

[KS-Soft]

Your request is much more complicated for implementation

1) we should implement some new status, e.g. "Warning". And option that can be set for each test item separately, e.g. "Use Warning status for 1st N failed probes"

2) We should modify Log Analyzer to process this status. How it should be processed? As Bad status? Or it should be considered as Bad only if its followed by Bad satus? What if test status changed to Warning and then back to Ok? Should we consider Warning as Good? as Bad? as Warning? What we should display in charts?

3) We should redesign action manager. Probably Warning status should be ignored by regular actions but it may be used in advanced actions. Right?

4) Modify Report Manager

5) How HostMonitor should calculate statistics? Alive%, Dead%, Alive Time? Simply add additional fields? But if test changes status to Warning and then to Bad, summary time should be added to Bad counters, right? If test changes status to Warning and then to Ok, should we increment Alive counters or Dead counters? Or Warning counter only?

6) Minor changes in Color profiles, new color item, right?

7) What about Unknown status? Should HostMonitor use Warning for 1st N "Unknown" results? or should not? Or may be this should be additional option?

This is just quick thought, I am sure there are more problems...

Regards
Alex

[Kapz]

Funny - we just had a talk at the office today regarding many of these issues, and from our point of view the answers are:

1) we should implement some new status, e.g. "Warning". And option that can be set for each test item separately, e.g. "Use Warning status for 1st N failed probes"
A: Yes, a new status "Warning" and an option to set N.

2) We should modify Log Analyzer to process this status. How it should be processed?
A: Your own suggestions are fine - perhaps a setting in an .ini file could let the user determine how the final calculations should be done (e.g. should the test(s) with status Warning be converted into Bad or Good's in the final result - or converted at all)

3) We should redesign action manager. Probably Warning status should be ignored by regular actions but it may be used in advanced actions. Right?
A: Exactly - perhaps a switch similar to "Treat unknown as Bad" could come in handy here ?

5) How HostMonitor should calculate statistics?
A: Same as question 2, I guess.

6) Minor changes in Color profiles, new color item, right?
A: Right

7) What about Unknown status? Should HostMonitor use Warning for 1st N "Unknown" results? or should not? Or may be this should be additional option?
A: My first thought is that Warning is an independant option although there may be quite a few places where Unknown and Warning should be treated the same way.

> This is just quick thought, I am sure there are more problems...
... and I'm pretty sure that you are right

Our biggest problem regarding the visual alarms is that 9 out of 10 visual alarms that we see in HM are e.g. CPU's having a spike that very same millisecond that it's beeing tested by an RMA.
Someone then has to find the right customer in the list, find the right server in the list, find the right test, right click on the test and refresh it - and then the CPU is back to normal.
A staus called Warning that can be customized so that it 1) won't change the color of HM's tray icon and 2) can be customized so that it can use the same colors as tests with status OK would surely be very welcome !

Thanks !

Kasper :O)