Thomas Heute
February 13, 2015 at 4:22 AM

On 02/12/2015 11:25 PM, Catherine Robson wrote:
Hi -

We see that alerting on response time and downtime are part of what we hope to provide in the first iteration of Hawkular.  We'd like to get started on the designs related to alert definition/configuration.  I'm hoping that you can all share some of the requirements around alert definitions that you think we need to have for Hawkular at this point.  I don't want to overdo it by looking at JON - I'd like to start simple. 
+1
Here's the requirements for the web console as I currently am thinking of them, but would like the team to comment on them and add/remove requirements as you see necessary.

Overall Alerts
As an administrator of a website, I would like to have all alerts sent to me through e-mail.
Ad an administrator of a website, I would like to have all alerts sent to me via text message.

That may be implicit but:
    - As an administrator of a website, I would like to have all alerts listed only  to the console.
Ha! Right :) 

We may not implement SMS right from the beginning but having 2 ways may be good to initiate the design. (Later we should embed Aerogear UPS and have a small app for push on phones, note to potential readers, a good student subject or contribution).

1 important thing:
    - we need some alert "profiles", if I look at my 1000s resources I may want them all to follow a same profile and when I want to change who receive an email I should do that in a single place (and not go through the 1000s resources). There would be several profiles.
Great - this makes sense.  Do we need to hand enter user information (e-mails/phones) or can some of this information be gathered through KeyCloak potentially? To clarify exactly what we think a "profile" contains - please verify below.

An alert profile is a place where users can set up alerting contact information and rules for many resources.  An alert profile contains:
Another alternative is that the resources are not mentioned in the profile, and you just assign the profile when you're working in the resources.  This feels much more like an "Alert contact group" than a profile to me in that case, so it is just a terminology change I think to make it clearer for what to expect from this capability.

3 potential improvements that we may want to think about in the design right now (or not):
    - Different addressee: Support for sending email/SMS to someone else but the owner
Let me see if I can expand on this use case to make sure we're all on the same page.

Precondition:  An alert fired.  It was sent to person A.
Step 1: User sees the alert, and wants to "share" this alert with person B.
Step 2: User manually enters Person B's e-mail or SMS information.
Step 2 alternate:  User selects from a dropdown list of existing known users to find Person B, and Person B's preferred contact method is used for sending the alert.
End Goal:  Alert is sent to Person B based on the method chosen above.
    - Escalamation: if resource is down for 5 min, send me (or someone else) an email, if still down after 30min send me a SMS
Could this use those alert profiles too?
    - Multiple alerts for 1 particular event: if resource is down for 5 min, send me an email, send my boss an email and send the IT guy a SMS



Downtime
As an administrator of a website, I would like to configure Hawkular so an alert is sent to me every time the system goes down.
As an administrator of a website, I would like to configure Hawkular so an alert is only sent to me after the system is down for a certain length of time, so I'm not alerted if there is a very minor downtime event.
+1

Response time
As an administrator of a website, I would like to configure Hawkular to alert me when my website's response time is slower than a threshold I have set so I know there may be performance problems.
It would have to be for some configurable period of time
Ok - so you would never want to alert if we go over *at all* for this metric, you would only ever want to alert based on a time interval it was above the threshold for.

Are there any other "settings" to the alerts that we should be considering at this point?

At some point in the future we may want to have a warning state, but I don't want to surcharge this thread :)

Thomas
Thanks,
Catherine


_______________________________________________
hawkular-dev mailing list
hawkular-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/hawkular-dev

Catherine Robson
February 12, 2015 at 5:25 PM
Hi -

We see that alerting on response time and downtime are part of what we hope to provide in the first iteration of Hawkular.  We'd like to get started on the designs related to alert definition/configuration.  I'm hoping that you can all share some of the requirements around alert definitions that you think we need to have for Hawkular at this point.  I don't want to overdo it by looking at JON - I'd like to start simple.  Here's the requirements for the web console as I currently am thinking of them, but would like the team to comment on them and add/remove requirements as you see necessary.

Overall Alerts
As an administrator of a website, I would like to have all alerts sent to me through e-mail.
Ad an administrator of a website, I would like to have all alerts sent to me via text message.

Downtime
As an administrator of a website, I would like to configure Hawkular so an alert is sent to me every time the system goes down.
As an administrator of a website, I would like to configure Hawkular so an alert is only sent to me after the system is down for a certain length of time, so I'm not alerted if there is a very minor downtime event.

Response time
As an administrator of a website, I would like to configure Hawkular to alert me when my website's response time is slower than a threshold I have set so I know there may be performance problems.

Are there any other "settings" to the alerts that we should be considering at this point?

Thanks,
Catherine

--
Catherine Robson
User Experience Design
Red Hat JBoss Middleware
c: 978-944-3825