Dependent Test Not dependent on one Bad action

Nick Danger · Post by **Nick Danger** » Thu Oct 30, 2003 9:05 pm

I started playing with a downloaded copy of Version 4.1 I think it does not act the same as version 3.12, which behaves "correctly."

I have two tests A and B.
A is the Master.
When A fails B is run.
B's Alert profile contains a Bad (X) and a Good action (Y).
"Good Status Action" Y is NOT "dependent on one Bad action."

When Test A fails, triggering Test B, Good Action Y is not initiated as soon as Test A fails. (Which is what I want it to do.) (Which it does in ver 3)
It seems to only be launched after B has gone from Bad to Good. But "Good Status Action" Y is NOT "dependent on one Bad action."

Synchronize counters is selected, Synchronize alerts is not.

Please let me know if there is something I may have done wrong or if this is an intentional change.

Keep up the great work.

KS-Soft · Post by **KS-Soft** » Fri Oct 31, 2003 4:35 pm

"Good Status Action" Y is NOT "dependent on one Bad action."
When Test A fails, triggering Test B, Good Action Y is not initiated as soon as Test A fails. (Which is what I want it to do.)
It seems to only be launched after B has gone from Bad to Good.

Looks like you need to change "Repeat 1 time" option to "Repeat until status changes" (property of the good action Y).
Version 3.12 was a Beta version, all versions after 3.18 perform actions similar to version 4.

Regards
Alex

Nick Danger · Post by **Nick Danger** » Sat Nov 01, 2003 4:26 pm

But what if I want the Good action to be performed only once and not repeatedly until Status changes?

(Sorry, I thought this was an easy Bug report, I didn't mean to have a tech discussion here...)

KS-Soft · Post by **KS-Soft** » Sun Nov 02, 2003 7:46 pm

When Test A fails, triggering Test B, Good Action Y is not initiated as soon as Test A fails

But what if I want the Good action to be performed only once and not repeatedly until Status changes?

Looks like you do not need "good" action for Test B, you need "bad" action for Test A. I mean If you need to start some action when Test A fails, add this action to profile that belongs to Test A.

Regards
Alex

Nick Danger · Post by **Nick Danger** » Sun Nov 02, 2003 8:39 pm

Maybe I can be clearer if I am more concrete. Actually there is an A, B and C. C is dependent on B which is dependent on A. Normally only A is Good. B and C are Waiting for Master.

This is a remote server and remote router. If the remote server is Good I don't need to test the remote router. If the remote server is Bad, I want to test the remote server for latency (same test, different TTL) (not terribly graceful, I know, but effective). If it fails both of those, then I want to test the remote router. The reason I don't want Bad actions on all of these is, as you can guess, if the WAN is down, I do not want to get three Bad messages. By only having Good actions, I am able to find out exactly which is the last functioning link in the chain: the server, the server with high latency, or the router. Of course, the last test in the series has a Bad action...WAN down.

Because these are Dependent, the last Good test in the chain will go from Waiting for Master to Good, with no Bad status. Because it is sending an email, I don't want it to repeat until status changes. I want it to only send one message.

Hope that makes sense. Again, if the Good action not firing when the dependent test changes from Waiting to Good is intentional, I'll just try to figure something else out.

gdvl · Post by **gdvl** » Mon Nov 03, 2003 1:38 am

Why make it so difficult ?
Make 2 tests: one tests the router (test A), and one tests the remote server (test B). Test B is dependent on test A.

When test A fails, test B becomes 'waiting for master'. There is one alert, and you know exactly where's the problem.
When test B fails, you know test A is ok, and you know there is a problem with your server. Again, there is just one alert.

Regards,
Gert

KS-Soft · Post by **KS-Soft** » Mon Nov 03, 2003 4:22 pm

I agree with Gert. Router does not depend on server, server depends on router.

Regards
Alex

Nick Danger · Post by **Nick Danger** » Mon Nov 03, 2003 10:05 pm

Thanks for all your suggestions, but because a latent WAN circuit requires a different response than a down circuit, it really would be best if I had three dependent tests. (Because these are all remote to my location, the server depends on the router. If the remote server is up, the remote router MUST be up.)

Here is my question: Can HostMonitor handle a chain of 3 Master/Dependent tests?

Master-->Dependent/Master-->Dependent
Test A-->Test B---------------->Test C

KS-Soft · Post by **KS-Soft** » Mon Nov 03, 2003 11:06 pm

Here is my question: Can HostMonitor handle a chain of 3 Master/Dependent tests?

Yes, of course.

Regards
Alex