A question on operation periods...

Mogge · Post by **Mogge** » Tue Jul 20, 2004 5:09 am

Hi!

I have a installation of 11 servers that all are monitored by HM and like anyone else in this forum I think it´s (HM) doing a great job!! Long to see the next version, Alex

Now, I have my monitoring split up in three periods during the day.

1 = Primary time (from 7:30 AM to 04:00 PM mon-thu and 07:30 AM to 00:30 PM on friday)
2 = Secondary time (from 06:00 AM to 07:30 AM + 04:00 PM to 06:00 PM mon-thu and 06:00 AM to 07:30 AM + 00:30 PM to 06:00 PM friday)
3 = Other time (any other time)

I have scheduled my alerts in 4 categories all depending of severity - e.i. I want a alert on SMS + mail on every service disrupture in the primary period but I only want a mail in the sekundary period. In all other periodes I don't want any alert on service disruption.

In the same time I have made the alerts depending on other alerts and the sending of SMS "Good" alerts depending on the succesfull sending of a SMS "Bad" alert.

All in all it works quite nice.

But!

Problem 1:
If I get an alert of some kind lets say 10 min. before the primary period starts and it is not configured to send an SMS at that time but want's it to send a SMS at the start of the primary period... What then... I have tried working with the possibility to reset or restart all the tests at the change of the periods but haven't found the right kind of "reset" yet. I figure that I have to execute an "action" thru the "options/scheduler" containing some sort of code/script...

Problem 2:
The "options/scheduler" can only contain 5 schedules - I think it to few to do what I'm trying to do. Can we get more than 5, please...

Kind regards
Mogens

Marcus · Post by **Marcus** » Tue Jul 20, 2004 9:16 am

For problem 1: Create two tests. One for each time period. Just make sure that only one can be active.

KS-Soft · Post by **KS-Soft** » Tue Jul 20, 2004 11:37 am

I have tried working with the possibility to reset or restart all the tests at the change of the periods but haven't found the right kind of "reset" yet.

Right kind of reset?
- enable "Reset Recurrences when test gets disabled" option (Behavior page in the Options dialog)
- use HM script to disable and enable tests
-----------------
DisableAllTests
Enable AllTests
-----------------
But I am not sure you will like it. HostMonitor will start alerts for each test ("bad" alerts for failed tests, "good" alerts for alive items).

Probably some new option "deferred action" could help. But... What if test changed status several times in "other" period? You want to receive all messages?

Regards
Alex

Mogge · Post by **Mogge** » Wed Jul 21, 2004 12:43 am

Hi Alex.

Your right

I don't like the side effect of using "Reset Recurrences when test....". I is actually the effect I'm looking for but only the "Bad" alerts.

Let's look aside from the tecnical part for a moment. I'm sure that we are able to get some kind of solution so let's try looking at the "usage" part for just one second.

The meening of monitoring some equipment is to get alertet if something is not functioning. But it all depens on when and where the equipment is failling. Let's take an example. If the spooler service on a print server is failling friday night I don't care and thereby don't wan't to know until monday morning when the "seconday period" starts. It's 1½ hour before users normaly log on (an when the primary period starts) so I have plenty of time to get the service startet. If the server was scheduled to restart during the weekend then the service was startet automaticaly and then I can't use the alert anyway. It's startet allready. No action is required from my part. Sure I wan't a log where I get the information on the service breakdown but it's not required until I read my mail monday morning. It could be the monitoring of a spooler service or another service like a Exchange IS service. The IS service can be stopped on purpose if I like to make a offline defragmentation during a service windows the night between saturday and sunday. It is not the ideal situation that it's alerting me on doing it's job....

Maybe it's not the right word(s) but it's like "posponing" the alert until a period starts that contains a configuration that is suppose to alert by SMS (or another action).

Another example: If a service failes at 05:55 AM (and the secondary period starts at 06:00 AM witch would trick a SMS in this situation) no alert is send. If the same service failes at 06:05 AM I get the alert.
The ideal situation would be to get the alert at 06:00 AM where the existence of a running service is required. It's like getting a total status on the equipment at a specific time.

Now back to the tecnical part.

I could create multiple tests but come on.... That would require a lot on maintainance of the same information (and configuration) on several tests. It's old programming - this is the new sentury

The surgestion of using "Reset Recurrences..." is actually a surgestion in the right direction. I have locked into the "DisableAllTests / EnableAllTests" and it's also in the right direction. But the "command" I need is to be able to "set" the/all tests to status "Good" at a specific time. Then I would get the alerts if the test fails subsequently.

Well - it's about all I can explain. It's a little difficult to explain if you don't think on the "usage" instead of the "tecnical" part of HM.

Still - HM is great! I have worked with several monitoring programs and HM beats many of the "big" programs witch costs a mill. bobs. in functionality! Great!

Regards
Mogens

Marcus · Post by **Marcus** » Wed Jul 21, 2004 2:11 am

I could create multiple tests but come on.... That would require a lot on maintainance of the same information (and configuration) on several tests. It's old programming - this is the new sentury

Indeed, but it would solve your problem right away. And it sounds more work than it realy is:

-Change default
-Export settings
-Search and replace schedule (and folder locations if wanted)
-Delete second set
-Import settings

And most of it (if not all) can be scripted.....

To solve your problem, it must be possible to assign multiple schedules to a test and every test - schedule combination should keep track of a list of actions for that test in the defined schedule. It sounds simple

KS-Soft · Post by **KS-Soft** » Wed Jul 21, 2004 11:20 am

Mogge,

I hope I understand your needs but...

But the "command" I need is to be able to "set" the/all tests to status "Good" at a specific time. Then I would get the alerts if the test fails subsequently.

It could have various side effects. E.g. some dependant tests (that should not be executed) will be executed.

If we implement some "StartAllBadActionsForFailedTest" command, it could have side effects either. E.g. some actions not just send a message, some actions may reboot service or server....

Probably "deferred action" option that could be enabled to some specified actions (and disabled for another) is most inoffensive solution...

The ideal situation would be to get the alert at 06:00 AM where the existence of a running service is required. It's like getting a total status on the equipment at a specific time.

You can easily implement this using standard HTML report

Regards
Alex

Mogge · Post by **Mogge** » Thu Jul 22, 2004 1:59 am

Hi again

It could have various side effects

Year... but no... If you have your dependensies configurated correctly it schouldn't be a problem.

I just tried the "Reset 'Recurren....'" in options... as you stated it runs every "God" and "Bad" action....

A "StartAllBadActionsForFailedTest" command could be quite nice

Sure I can see the side effects concerning rebooting servers and so on

I don't know what else I can say - maybe a suboption to the "reset 'recurren...'" witch allow you to send only "Bad" actions????

Regards
Mogens

Mogge · Post by **Mogge** » Sat Jul 24, 2004 3:52 pm

Hi again

maybe a suboption to the "reset 'recurren...'" witch allow you to send only "Bad" actions????

Could this be a possibility, Alex?

Kind Regards
Mogens

KS-Soft · Post by **KS-Soft** » Sat Jul 24, 2004 4:05 pm

No, I don't think so.

StartAllBadActionsForFailedTest command looks much more realistic.
Probably it can be with some parameter - "depth" of the actions. E.g. StartAllBadActionsForFailedTest 1 will start actions with "Start when 1 bad result occurs" option.

BTW: Why you don't like simple HTML report with list of Bad test items?

Regards
Alex

Mogge · Post by **Mogge** » Sat Jul 24, 2004 4:12 pm

Hi Alex

Thanks for your answer!

A StartAllBadActionsForFailedTest will be very usefull!

About til HTML report: I was talking 'bout a "reporting of status" "thru" a SMS gateway. That means "Report al Bad status" thru SMS.
An HTML report will be good on a Intranet but Í'm not allways connected so an SMS "Report" is what I'm looking for

Kind regards
Mogens

KS-Soft · Post by **KS-Soft** » Tue Jul 27, 2004 7:27 pm

Ok, I have added this task into "to do" list

Regards
Alex

Mogge · Post by **Mogge** » Thu Oct 21, 2004 7:31 am

Hi Alex

Any plans on when this is going to be implemented?

/Mogge

KS-Soft · Post by **KS-Soft** » Thu Oct 21, 2004 5:24 pm

H'm...
We always have dilemma - continue endless small improvements or make complettely new version.
Probably we will release 1-2 new 4.xx versions and start version 5 development. Unfortunatelly I cannot guarantee this option will be implemented in version 4

There are hundreds of items in "to do" list.

Regards
Alex