New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Incident muting (for downtimes, flapping, acknowledgements) #190
Comments
Yesterday, @bobapple, @lippserd, @julianbrost and me discussed this and agreed on the following: Extended State EventsSources will be able to include additional details while transmitting These details are as follows: mute: A boolean that indicates whether notification transmission for the associated object should be suppressed. Repeated TransmissionTwo successive Handling During No Active IncidentNone. Yet. Handling During An Active IncidentOnce notification suppression is switched on, the incident history should indicate this with an appropriate entry. This should also be the case if the incident has just been opened. (i.e. as the second entry) Escalations are triggered as usual. The transmission state of resulting notifications should remain in a new state called In case notification suppression is disabled and the incident closes, previously suppressed notifications are not transmitted. On the other hand, if the incident is still open, all notifications that were previously suppressed must now be transmitted at once reflecting the current severity. |
The object has to be marked as muted so that if an incident is created, it is handled correctly.
I'd just call it
No, that should simply notify about the state from the unmuting event (similar like if it was a severity change). Otherwise, it might send multiple notifications and also to no longer active contacts in a schedule. |
I've meant exactly that! |
As both @nilmerg and @yhabteab have expressed doubts about this suggestion now, let me explain why I came up with it. To me, |
I can't follow you, honestly. Please elaborate this, if you like, in person. |
If more causes of a notification being suppressed are added, how would you represent them in the history? Would they all get Other reasons could be something like "incident was muted manually in the web interface". And we already have another situation where you could argue such a history entry should be added as well: "incident has a manager and regular recipients weren't notified". I'd differentiate these, the question is whether to do this within the Yes, |
FINE! I'll move on to |
To me, this should be obvious based on the preceding events. There's is no reason why a specific notification has been suppressed, just one since when any of them are.
This is one of the reasons why an object can be muted. Basically very similar to a native way to mute an object, or incident, explicitly. Though, in this case issued by the source itself. Whether we'll keep the functionality that such can declare a manager, hasn't been discussed... |
I don't really understand what you're saying here on a language level. Do you want to say that there multiple reasons for suppressing a notification may be active at the same time so there's not necessarily a single reason why the notification wasn't sent? |
No. That's what you're implying by suggesting to define multiple suppression types. I'm referring to this:
There is only a single reason at a time. If suppression is enabled, all and any notifications are suppressed. The reason is visible in the history of the incident. That is why I don't see your argument against |
Oh, I don't want to split the reason why the source muted the object up into multiple reasons. All of the reasons specific to Icinga 2 should still be considered one reason here. However, when other suppression reasons are added within Notifications, it should be possible to distinguish them. For different reasons, you'd have to look for different previous events for more details ("Icinga 2 muted the object: in downtime", "John Doe muted the incident: [comment written by the user]", ...). There's no plan to add such other reasons as part of this issue, the question is simply whether some other naming would be a better starting point if we want to do this in the future. As of #190 (comment), i.e. before I started a discussion about this, how would you have presented a notification event with |
e.g. (top to bottom)
|
Reviewing the code of #146 and testing it showed some issues that where the proper solution requires some kind of incident muting mechanism.
The text was updated successfully, but these errors were encountered: