Dashboard and Data Explorer - difference between built-in error metrics

[bernhard_s]

Hello,

I'm building a Dynatrace dashboard and I am confused about the built-in metrics related to errors.

According to the below Dynatrace documentation [1] on built-in metrics, Dynatrace offers the following error rates for use on a Dashboard:

1. builtin:service.errors.client.rate: Failure rate (client-side errors)
2. builtin:service.errors.fourxx.rate: Failure rate (HTTP 4xx errors)
3. builtin:service.errors.server.rate: Failure rate (server-side errors)
4. builtin:service.errors.fivexx.rate: Failure rate (HTTP 5xx errors)
5. builtin:service.errors.total.rate: Failure rate (any errors)

[1] https://www.dynatrace.com/support/help/how-to-use-dynatrace/metrics/built-in-metrics

The documentation is not specified about what these errors are about, therefore I am asking here in the Community forum for clarification.

My questions:

• What is the difference between client-side errors and HTTP 4xx errors?
• Likewise, how can I distinguish HTTP 5xx errors from server-side errors?
• Are the HTTP 4xx errors part of the client-side errors, and are HTTP 5xx errors part of server-side errors?
• Is the metric "Failure rate (any errors)" then the error rate when considering all 4 error categories (HTTP 4xx, HTTP 5xx, client and server errors)?

Thank you for your support!

Best regards, Bernhard

[Mizső · Accepted solution]

Hi @bernhard_s,

Could you please check this part of the documentation I think you will find the answers for your questions.

Configure service failure detection | Dynatrace Docs

Best regards,

Mizső

[bernhard_s]

@Mizső   - No, this does not answer my questions. BR Bernhard

[zietho]

The referenced docs page referenced by @Mizső I believe does explain the difference of detected errors and how to configure them if they don't meet your need.
For example reading the explanation tells me that 500-5xx will always be interpreted as a server error. So 5xx metrics.rate is a server error while 400-4xx (yes there is a typo on that page) is a client error.
Then it also says that missing HTTP response code as client-side errors or server-side. So my naiv assumption is that 4xx is everything from 400-4xx while client side or then server-side (500-5xx), respectively can be more than just the actual 4xx or 5xx errors. E.g. when you also take missing response codes into account. 

Answer to 

• What is the difference between client-side errors and HTTP 4xx errors?
• Likewise, how can I distinguish HTTP 5xx errors from server-side errors?

Id therefore say yes to 

• Are the HTTP 4xx errors part of the client-side errors, and are HTTP 5xx errors part of server-side errors?

And yes to: 

• Is the metric "Failure rate (any errors)" then the error rate when considering all 4 error categories (HTTP 4xx, HTTP 5xx, client and server errors)?

But maybe @tom_rothschaedl can cross-check this, pls.