Web Problem Detection not working

[abnerlusung]

Hi all,

I want to explore the https://www.dynatrace.com/support/help/observe-and-explore/davis-ai/problem-and-root-cause but it's not working.

I'm working with JQuery, Spring Boot.

I have tried the following scenarios:

1) Get Request for 5 minutes

2) Get Request with Http Code 500

3) Get Request for Image 404

4) Javascript Error

5) Every 500 ms for the #3

6) Every 1000 ms for the #2

Next is the Settings > Anomaly Detection

Detect increases in JavaScript errors is Turned ON

Using fixed thresholds

1% Alert if this custom error rate threshold is exceeded during any 5 minute period

10 minimum number of action per minute

High Sensitivity

1 minute Amount of minutes the observed traffic has to stay in abnormal state before alert

Next is for Request Errors:

HTTP 500-599 Capture/Impact Apdex/Consider for Davis

HTTP 400-499 Capture/Impact Apdex

Errors with failed image requests Capture/Consider for Davis

Why am I not getting Problems? 

Regards,

Abner

[Romanenkov_Al3x · Accepted solution]

Static thresholds should be working, if there will be traffic (10 minimum number of action\requests per minute)

Automatic baseline thresholds will work after two hours for service.

If your service has no load for 80% time - there will no alerts. Your service must have traffic more than 20% time on the week.

Regards,

Romanenkov Alex

[abnerlusung]

Hi Romanenkov Alex,

Thank you for the reply.

So what should I do?

[Romanenkov_Al3x]

Abner, as I understand you want to try automatic baselining. 

If it is your test environment (imagine this) you need to have some synthetic traffic, for example using selenium with web driver. Make some scenario - for example - susscessfull login without errors.

Make it work for 24 hours. 

Check that your traffic exceeds "10 minimum number of actions per minute" or set this value lower.
On the next day you can access your application and have a scenario with error.

If your synthetic has no error at all, you have "0%" baseline error rate.

If your impact will more than both the absolute and relative thresholds - Dynatrace will rise an error.

In this example - for my app expected error rate 1.5% If I will get 10% - Dynatrace will rise error for me. 

If you can't\won't wait time - you can use fixed thresholds. For my app I know  the expected error rate. For example errors by entering CAPTCHA. I always miss this 😉

For me it is 1.5. For my SLA for this web application is 5% errors (other errors not related with entering CAPTCHA) and I want to get notified if there are others errors in my example. I set this threshold - 6.5 to achieve this. This is only example for me.

Regards,

Alex

[Romanenkov_Al3x]

For example for services - you need 2 hours to get first baseline value and automatic thresholds working. 

[abnerlusung]

HI @Romanenkov_Al3x ,

Do you mean for 2 hours, I need to open my web, run my XHR, then end session?

What do you mean by "service"?

Regards,

Abner

[Romanenkov_Al3x]

I recommend to have 24 hours for it.

Some events - for example - unexpected low traffic will rise only after 7 days.  

Service it is my favorite object in Dynatrace. It is frontend\backend object of you app. For example you deploy Apache HTTPD or NGINX with "Hello this is NGINX default page". When you call HTTP request to the web server, Dynatrace will create "Service" for you.  All requests related to this endpoint (for example - localhost:80) will in this object. 

Please, read more documentation about it. Dynatrace Doc Team do their work best.

Regards,

Alex