Service based alerts

[Arvind Jayaprakash]

While it is possible to associate an incident with a service, is it possible to have alerts have associated with a service? Associating alerts to just a team solves for associating an alert with people but offers no way to associate an alert with a service. One can always workaround by using specific tags with alerts to capture the notion of service but this is not the same as having them rollup under the "service" concept offered by OpsGenie.

[Pieter Nagel]

I third this feature request. I was looking for a way to inhibit alerts for a service when that service is undergoing maintenance, using something that a human can click in the OpsGenie interface or a machine could toggle via the API. I thought Service and Maintenance was the OpsGenie way to go about this.

Now I discover that every alert will have to be upgraded to an Incident to achieve this. Doable, but it feels like an unnecessary extra layer of indirection.

[Sammy Odenhoven]

I second this feature request. Having moved from PagerDuty to OpsGenie, this is one of the features I miss the most. While I appreciate both tools are different and can't be compared one on one, the ability to associate alerts with a service, but also to define configuration on a service level (e.g. for service A notify through email, for service B notify through Slack channel #devs, for service C notify through Slack channel #company-wide) was a very useful one and I hope to see OpsGenie provide such functionality in the near future too.

[DamCoder]

We migrated from PagerDuty to OpsGenie as well.  I too miss this feature.

[Chris DeGidio]

Hi Arvind, 

I hope your day is going well and this answer reaches you well. Currently, in the incident platform of Opsgenie, you would be able to correlate alerts to a service through the services incident rule. Utilizing the incident rule will ensure that when any specific alert comes in it can be matched to the service and then correlated to an incident as it is created automatically. I have attached a screenshot of where you can set that up. Setting the incident rule up will ensure any specific alert is matched to services and rolled up into an incident.

Please let us know if there are any further questions or concerns and we will be happy to help. 

[Chris DeGidio]

Here is the screenshot that got left off my initial reply. 

[Arvind Jayaprakash]

Thanks @Chris DeGidio . I was already aware of this.

What you showcased is as follows:

1. Ability to associate an incident with a service
2. Ability to autocreate an incident via an alert
3. Use #1 + #2 to have an alert be associated to a service

The shortcoming with this approach not every alert results in an incident. Hence the ability to directly associate an alert to a service under all situations (incident or otherwise) becomes essential.

[Erik Osterman _Cloud Posse_]

It would be helpful to understand why Opsgenie deliberately does not provide the ability to associate an alert with a service (and hence be able to view the relationship between services). 

I agree with @Arvind Jayaprakash that it seems natural that not all alerts are incidents, but all alerts relate to services. 

Since the tool doesn't support it, I am wondering if the way we're architecting for opsgenie is wrong given that a lot of thought has been put into the product.

Clearly tags and extra fields can be used to connote the service names and can be used in incident rules, but that doesn't help communicate the service relationships when looking at alerts, despite services existing as a native feature of the product.

[Alexander Kim]

Just found this as I've been trying to figure this out off and on for a while. It does not make any sense that alerts are not able to be associated unless I am misunderstanding what OpsGenie intends an incident to be.

[Lyle Bush]

I agree - evaluating OpsGenie and this may be a show-stopper due to confusion it may create with existing business processes.  In theory, I guess we could treat a Team as a Service...

[erich2048]

I'm also implementing Opsgenie and very disappointed that I can't associate alerts with a specific service. Only a small fraction of the alerts require an incident response, but reviewing the alerts by service is crucial!

[brett_gardner]

I don't understand this limitation either.

In a multi team, multi service organisation, what is the point of having alerts at all?

Not all alerts would be considered incidents, eg a high load alert. A high load alert might be a one off due to actual high load which is acceptable, if you start seeing them repeatedly, at that point you group them into an incident to start mitigation measures.

And from a reporting point of view, we need the ability to track which of our teams, responsible for various services, are not investigating their alerts.

(edit)

So looking further I can see that you can in fact associate an alert to a team, just not a service. It still seems like an odd limitation though.