How do I get dynatrace to notify me of an incident beforehand?

[SebastianLozano]

Normally dynantrace notifies me of the incident when it is already present on a server, however, how do I get it to notify me before the incident happens?

[AntonPineiro]

Hi,

What do you mean? Do you have some example?

Best regards

[SebastianLozano]

Hola Anton.

Un ejemplo es cuando tienes tu servidor y dentro de ese servidor tienes una base de datos y esa base de datos empieza a tener degradación en tiempo de respuesta o por alguna razón hacer que los servicios se alenten ¿hay manera de que dynatrace me ayude  notificarme ese problema antes de que los usuarios los reporten?

[Google Translate:

Hello Anton.

An example is when you have your server and within that server you have a database and that database begins to have degradation in response time or for some reason cause the services to slow down, is there a way for Dynatrace to help me notify me of that? problem before users report them?]

[AntonPineiro]

Hola,

Si ya ves el problema en la sección de problemas, deberías de estar seguro que tu problema es parte de un perfil de alertas.

Luego tienes que crear una integración del perfil de alertas con una notificación, ya sea email, ServiceNow, etc. https://docs.dynatrace.com/docs/shortlink/problem-notifications

Un saludo

[AntonioSousa]

@SebastianLozano,

It's nice to ask for something that hasn't happened yet, but not all is possible. Dynatrace does have some functionality in that domain, and some I remember immediately include:

• If a disk is getting full, Dynatrace will notify you before it gets full. There is discussion on how to improve on the available functionality (based on 97% full) and I believe it's near.
• Most of the resource alerts tend to alert before the real problems begin.
• Alerts based on Garbage Collection also tend to appear before big problems occur.
• Synthetic tests might alert you when no other users have experienced problems.

Now, there are cases where Dynatrace will not alert you before the problem. Nor any other solution. I'm quickly thinking of cases that have happened to me in recent years:

• Server crashes out of the blue or has a hardware issue. Most of these are completely unpredictable.
• Someone cuts a cable or disconnects physically a machine.
• Some ISP, mandated by a court, cuts your Internet connectivity.

One of the good things about the Dynatrace Community is that you can share your ideas with Dynatrace, as Product Ideas. I have been lucky to have multiple Ideas that I posted implemented

[SebastianLozano]

buen dia antonio Sousa, entiendo la parte de los servidores ¿ pasa lo mismo a nivel aplicativo? ¿hay manera de que dynatrace me indique que aplicacion va a fallar con previo aviso?

[Google Translate:
Good morning Antonio Sousa, I understand the part about the servers. Does the same happen at the application level? Is there a way for Dynatrace to tell me which application is going to fail with prior notice?]

[AntonioSousa]

@SebastianLozano,

At the application level, it is more difficult to predict when an application will fail. Garbage Collection is in my opinion one of the most valuable indication that a .Net/Java application will start failing. And Dynatrace correctly alerts on that.

[SebastianLozano]

¿como se hace la recoleccion de basura ? ¿me podrias explicar?

[Google Translate:

How is garbage collection done? could you explain me?]

[AntonioSousa]

@SebastianLozano,

Dynatrace has a great ebook, where GC is also explained:
https://www.dynatrace.com/resources/ebooks/javabook/how-garbage-collection-works/

Other useful links might include:

https://docs.dynatrace.com/docs/setup-and-configuration/technology-support/application-software/java...

https://docs.dynatrace.com/docs/setup-and-configuration/technology-support/application-software/java...

[PierreGutierrez · Accepted solution]

Hi @SebastianLozano 
I share with you 2 alternative solutions that I consider useful, for scenarios like the one you mention.
I think if it is possible 🎯

Alternative 1
Needing to know about a problem, before it becomes a Problem in dynatrace, means that you are needing a more sensitive alert. In that case, I suggest that you evaluate the alert parameters globally and/or specifically what needs to be reported.

The main objective of dynatrace is to perform a Backend analysis of the Davis parameters and self-learning, before generating a PROBLEM.
In this way, over-alerts to anomalies that do not end up being a Problem are avoided.

Alternative 2
 If you have already evaluated that your Problem generation parameters are correct, and you still need a prior alert (previous alert means at the time of detecting an anomaly or strange behavior)

I suggest the following to create a "pre-alert":

1.- Identify the entity to observe
2.- Create a Metric Event with optimal alerting parameters, configured in custom info mode.
3.- Create a Workflow, implementing rules that identify the Metric event created (Considering in the logic that it must be an active and created event)
4.- Implement a step in the integration workflow towards email, Teams, Slack (depending on what you need)

At this point there are suggestions and points to take into account that must be shared with the user, such as: The custom info alert will be made according to the parameters of the Metric Event and in relation to this the number of times it will notify the channel integrated into the workflow.

I can tell you that it does work, and there are several ways to make it dynamic 💡
I hope it's helpful 💪

[PacoPorro]

I want to highlight "forecasting".
Please take a look on this  https://www.dynatrace.com/news/blog/automate-predictive-capacity-management-with-davis-ai-for-workfl...