We have 44 process tests set up for this one remote machine. What we need to have is an alert that will send out one page/email if any one of those processes goes bad.
I know that we can set up an alert for each one of those tests but if all 44 of them went down at the same time we would get 44 pages/emails. And we don't want that.
All these processes are loosely part of the same app. So if one goes down we just need a page and then one of our admins can log on to the box and see what the problem is.
Please help me know if Host Monitor has this type of functionality.
Thanks
Multiple tests - have one alert
It's easy to activate only one alert when several tests fail, all you have to do is to make tests dependent of other tests as Alex said. For this to work correctly, each "slave" test must have a slightly higher polling interval than the "master" test, and then you must configure the alerts to execute only if you encounter "No answer" status in the "slave" test two or more times in a row. This is important, because even if the tests are run almost simultaneously, a "slave" (or dependent) test can run just slightly before the "master" test, and then you will get an alert even if you have configured dependencies (if the connection problem appears right after the "master" test, but before the "slave" test).
The problem remains, however, regarding "Good" alert messages, if you have configured tests to send a "Good" alert message once the connection has been restored. If one of the "slave" tests notices the connection problem before the "master" test does it will still change it's status into "No answer", until the "master" test also reacts to the problem and changes it's status to "No answer". Only after that the "slave" test will change it's status to "Waiting for master" (the second time it runs the test). If the "slave" test changes it's status to "No answer" even for a second, it will send a "Good" alert message after the status changes to "Good".
The workaround for this would be a setting where you could force the status of a "slave" test remain as "Host is alive" unless the test result is "No answer" for two or more times and the "master" test status is "Host is alive". Then the status of the "slave" test would not change into "No answer" even for a second before the "master" test has been run at least once after the "slave" test has noticed the problem. In other words, the "master" test would run twice before the "slave" test does, and the "slave" test status would change directly to "Waiting for master", as it should.
I haven't found this kind of setting in HostMonitor, but I hope there will be one in future releases. That would settle this issue for good. If I am mistaken, please do correct me - I would really like to get this thing working!
I'm sorry for this long and confusing explanation, but this is a bit complicated issue.
Regards,
Mikko
The problem remains, however, regarding "Good" alert messages, if you have configured tests to send a "Good" alert message once the connection has been restored. If one of the "slave" tests notices the connection problem before the "master" test does it will still change it's status into "No answer", until the "master" test also reacts to the problem and changes it's status to "No answer". Only after that the "slave" test will change it's status to "Waiting for master" (the second time it runs the test). If the "slave" test changes it's status to "No answer" even for a second, it will send a "Good" alert message after the status changes to "Good".
The workaround for this would be a setting where you could force the status of a "slave" test remain as "Host is alive" unless the test result is "No answer" for two or more times and the "master" test status is "Host is alive". Then the status of the "slave" test would not change into "No answer" even for a second before the "master" test has been run at least once after the "slave" test has noticed the problem. In other words, the "master" test would run twice before the "slave" test does, and the "slave" test status would change directly to "Waiting for master", as it should.
I haven't found this kind of setting in HostMonitor, but I hope there will be one in future releases. That would settle this issue for good. If I am mistaken, please do correct me - I would really like to get this thing working!
I'm sorry for this long and confusing explanation, but this is a bit complicated issue.
Regards,
Mikko
Some "recheck master test before set Bad/Unknown status for dependant test" option?The workaround for this would be a setting where you could force the status of a "slave" test remain as "Host is alive" unless the test result is "No answer" for two or more times and the "master" test status is "Host is alive". Then the status of the "slave" test would not change into "No answer" even for a second before the "master" test has been run at least once after the "slave" test has noticed the problem. In other words, the "master" test would run twice before the "slave" test does, and the "slave" test status would change directly to "Waiting for master", as it should.
Regards
Alex
Yes, exactly - I'm glad you could figure out what I meant even though I wrote such a long and confusing message...
An option for each test to recheck the status of the master test, or maybe a possibility to define a certain number of "Bad" results before switching the status to "No answer" - that would do it.
Regards,
Mikko
An option for each test to recheck the status of the master test, or maybe a possibility to define a certain number of "Bad" results before switching the status to "No answer" - that would do it.
Regards,
Mikko
We have this task in "to do "listAn option for each test to recheck the status of the master test
What about "Repeat test" action?or maybe a possibility to define a certain number of "Bad" results before switching the status to "No answer" - that would do it.
http://www.ks-soft.net/hostmon.eng/mfra ... #actRepeat
Regards
Alex
Improvement
How about a test method where I check the status of some tests?
I.e. If Ping_Test1 =BAD and Ping_Test2 =GOOD than Message Mike
or if ping_Test1 = BAD and Ping_Test2 = BAD than Message Steven
This would be far more flexible than dependency "kludge" (IMHO)
thanks!
Mike
I.e. If Ping_Test1 =BAD and Ping_Test2 =GOOD than Message Mike
or if ping_Test1 = BAD and Ping_Test2 = BAD than Message Steven
This would be far more flexible than dependency "kludge" (IMHO)
thanks!
Mike
You can implement such behavior using "advanced mode" actions
http://www.ks-soft.net/hostmon.eng/mfra ... ncedaction
Regards
Alex
http://www.ks-soft.net/hostmon.eng/mfra ... ncedaction
Regards
Alex