cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

Chart exception count filtered by message

mkulov
Guide

Hi,

 

Can we chart exception count filtered by exception message?

Any workaround if it is not possible out of the box?

 

Thank you!

Martin

Digital Performance Optimizer
10 REPLIES 10

JamesKitson
Dynatrace Leader
Dynatrace Leader

You con't do that out of the box because to do that we would need to be storing long term a metric for every exception which would be way too any timeseries metrics and use up a lot of storage.

What you can do though is create a calculated service metric with the source as request count and filtered down to the specific exception class or message that you are interested in. Then a new metric will be stored for that exception and you can chart it or use it to create alert events as needed.

julien_duhamel1
Participant

Hello Martin,

 

Yes, it's possible. For this you must create a "Calculated service metrics"

In this exmple i would like track the locktable with Dynatrace.

 

Use "exception class" and select what typologie of exception you would receive (In my exemple it's for SQL request).

julien_duhamel1_0-1616057922146.png

 

Dont forget to split with Dimension : Exception class and message 

julien_duhamel1_1-1616058148424.png

last step : 

Create a custom Dashboard with your metric inside like standart metric.

 

If you would like to navigate to your exception, you must create a direct link to "Multidimensionnal analysis"  like this  : 

julien_duhamel1_2-1616058599080.png

 

 

Hope can help you

Regards

 

Julien D. 

 

AlanZ
Organizer

Hi all,

Once the metric is collected, is there any way to drill down or find the purepath that have the exact exception message and thus related to an exception message metric ?

 

There is a filter for exception class but the exception message filter does not seem to exist.

alain_z_cote_0-1618259340288.png

 

 

Hey Alain,

The exception class is bug. You cannot retrieve your exact message with it.

Thats why i play with differents scenario : 

 

 

1rst : Select the service and filter with exception details, (working well but cannot transform into a dashboard unftornatly).

2nd way : Set requests attributes and filter with for display all transactions you are looking for. You can retrieve this information with Purepath or Diagnostic tool, does not matter.

Thanks Julien,

There is no "exception message" filter available in the service MDA analysis view, only "exception class". Since the "exception class" in our case is general and applies to many transactions and the "exception message" is the only thing specific to the purepath we're looking for, seems we're out of luck.

julien_duhamel1_0-1618439735475.png

 

Usually, i'm looking generical message, who can be relevant like that. I'm dril down on my critical service.. and show messsage "exception message". It's little bit boring, but it's much easier to drill down in ton of log files :=)

Malaik
Guide

Hi @julien_duhamel1 

Thanks for your time and your post, very interesting (Merci beaucoup).

 

I have 2 questions if possible:

 

Same thing for me I want to split each Generic Exception By message, or filter by message... (one generic exception have a lot of exceptions: the difference is the MESSAGE).

 

Second thing, the result is not correct for me

I use this calculated metric

alatarsa1_1-1624264920136.png

alatarsa1_2-1624264935390.png

and that is the result:

alatarsa1_3-1624264983720.png

 

in the diagnostic tool:

alatarsa1_5-1624265013322.png

 

 

Question: why the system count 4*2 (8Exceptions) knowing that I have only one Purepath that have 2 exceptions with the same message

 

Thanks for your help.

Good question, to my understanding : You are using executing 4X the same lib time your webrequest.

 To confirm, is you made a purepath/purepath., you should have same information. (Exception)

 

This exactly why, i group "message".  Try to add "Exeception class" in dimension field. Should be help.

 

It's complex to create "familly of message" in level of Webservice, this is why i prefer looking side SQL execution, you find some very valuable informations.

 

To my perception, read message in real time, it's incredible powerfull for RUN teams., you can catch "Token issues", "cerficate expired", "every applicative errors" . 

 

Hop can be help

 

Julien 

Hello @julien_duhamel1 

Totally agree, but this is the requirement of our customer 😞 .

Hi @julien_duhamel1 , another thing

 

I want to make the same thing with the response time, impossible for me to split by message.

 

I want to make a dashboard to link Requests Count and Response Time for each Exception Message.

 

Thanks