Wish: UP alerts

[jooray]

Hello,

 I would like to have better alerting, in particular:

- UP alerts (when something is fixed or gets fixed, "UP" alert is fired). This is standard feature of nearly any monitoring software (I have used Nagios and Mon before).

- better alerts configuration: how many failures are required for alert to be sent (after three different failures, alert is sent to this address, after 5 failures to that, after one failure, it's fed up to this script, etc.).

    Juraj.

[Sancho Lerena]

Hello,

 I would like to have better alerting, in particular:

- UP alerts (when something is fixed or gets fixed, "UP" alert is fired). This is standard feature of nearly any monitoring software (I have used Nagios and Mon before).

Sure!, some people have asked for it, but at this moment is not implemented, we have been more focused on new network engine, graphic reporting and Windows agents, but for 1.3 and 2.0 versions will be a massive improvement on alerting systems, so dont worry that this feature WILL BE on 1.3.

- better alerts configuration: how many failures are required for alert to be sent (after three different failures, alert is sent to this address, after 5 failures to that, after one failure, it's fed up to this script, etc.).

Could you explain with an example this feature ?, I dont understand exact   meaning of this....

Thanks.

[jooray]

Configuring this:

- when some monitor fails once, do nothing
- when it fails twice, send e-mail
- when it fails four times, send sms, send e-mail to additional four e-mail addresses
- when it fails five times, do a call using voip

For some services, it is common, that they fail sometimes (network traffic, ...). So I'm only interested, if the test fails five or more times.

F.e. I don't care, if load is temporarily 20. I do care, when it's more than certain threshold for 20 minutes.

Most monitoring systems (f.e. nagios, mon) can be configured to do this.

[Sancho Lerena]

Configuring this:

- when some monitor fails once, do nothing
- when it fails twice, send e-mail
- when it fails four times, send sms, send e-mail to additional four e-mail addresses
- when it fails five times, do a call using voip

For some services, it is common, that they fail sometimes (network traffic, ...). So I'm only interested, if the test fails five or more times.

F.e. I don't care, if load is temporarily 20. I do care, when it's more than certain threshold for 20 minutes.

Most monitoring systems (f.e. nagios, mon) can be configured to do this.

Now you can do all of this, it's not easy to implement, but you can do. Yo need to setup (for following this example), three alerts associated to a module. You setup a minimum of 2 and a max of 2, and associated response to send an email.

You setup another alert, with a minimum of 3 and a max of 3, with a response (a new one defined by you) that execute an script that sends an SMS and four emails to different address.

You setup the last alert, with a minimum of 4 and a max of 4, with a response (a new one defined by you) that execute an script to make a voIP call.

Is not simple, but it could be done. Really I don't know how to make it more easy to implement. Ideas be welcome :-)

[jooray]

Hello,

  well, min alerts and max alerts are quite problematic. I set max alerts to 1 and when one machine went down, I got an alert every 5 minutes, so I suppose it does not work that well (1.2 beta3).
 
  Min alerts is not documented anywhere (there's docs for max alerts only), so I don't understand exactly how it works.

  But one more thing. I don't think, that setting 5 alerts per service by creating new alert is a viable option. For example, I am monitoring about 40 servers with Pandora, each has about 8-10 monitors, so it's 400 services. If I had to create alert for each one manually, that would be 400x5=2000 filled in forms. So this is not a right way to do it. I created some alerts by inserting them to database, but it's quite difficult to manage.

  Maybe looking at Nagios would help. You create an alert group and when defining a service, you assign alert group with exact settings. When creating service, you just assign alert group and all alerts are assigned automatically. That would be a way how I would do it.


   Juraj.