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

Service is running fine but transactions are equal to the faillures!

christopher_teb
Organizer

Hi,

How can i explain this to the customer. Dynatrace shows that the failed transactions are equal to the normal transactions of a particular service yet no complaints from the users for that service!

Any idea on that??

12 REPLIES 12

dennis_mispelba
Inactive

Hi Christopher,

a screenshot would be helpful, though I try to give a hint:

Check WHAT actually causes the transaction to be marked as failed (via Errors Dashlet and then potentially by drilling down to the PurePaths). It might be that what you find is actually not "harmful" to anybody out there. Then you could finetune the configuration of when AppMon reports a transaction to be failed (within the "Error detection" setting of the respective system profile).

Since I dont know your application and what you are actually monitoring, it is a bit difficult to say your exact issue. But with the described approach, you might get a lot closer to a solution.

Best regards
Dennis

AppMon does not tell that the service is failing (when looking at snap2.jpg).

I think the Business Transaction might need a better config. Could you share the "Edit" section of the Splitting measure from snap1 ?

christopher_teb
Organizer

And happens on only one service on that application not all other services

Br,

Chris

christopher_teb
Organizer

Hi Denis,

Find the below screenshots,

measure.jpg

splitmeasure.jpg

bt.jpg

Br,

Chris

Hi Christopher,

your screenshots confirm my assumption. Here is the explanation:
A
Business Transaction is bound to the PurePaths it is filtering on. Now,
with your filter measure, you filter out all the PurePaths, which
contain the specific service call "js_contact_us". Remember, it is
filtering on the PurePaths - not directly on the service call. With your
split measure, you are then splitting on the response code of the
PurePath (!) [the entry point], not on the response code of the
respective service call. Thus, you end up seeing response codes 200 (as
the response was giving back 200 OK, although it contained an error) in
the splittings.

You should get better results if you use your filter measure as your split measure. However, another issue is your filter: it has a threshold of 1 (which normally would be fine in order to capture all), however, it is a response code measure! Thus, 1 is not a valid threshold here, as response codes are 2xx, 3xx, 4xx, 5xx and so on. So you would rather need a filter, which is going on count or response time (and then you can configure the "1"). At the moment, you are basically filtering on all response code for the service call. So review the configuration and it should work.

Hope it helps.

Best regards
Dennis

christopher_teb
Organizer

Hi Denis,

thanks very much for that understanding. Please help me with screenshots on the above knowledge so that i get the right results on the failures and successful service requests.PLEASE!!

Br,

Chris

christopher_teb
Organizer

Yes i put the filter measure in the split but i still get 404(failed HTTP response code) for all!

snap1.jpg

snap2.jpg

Br,

Chris

dennis_mispelba
Inactive

Hi Christopher,

excuse me, but first of all I would need to know WHAT you actually want to achieve. From the current explanations and the descriptions you gave in the beginning of this post, it is not completely clear to me. What is the underlying question you want to address?

For your row in "snap2.jpg" - it is totally correct to have 100% of 404 errors marked as failed - simply because 404 is "not found" response code and thus is an error in itself.

Best regards
Dennis

christopher_teb
Organizer

Hi Denis,

Ok the thing is like this,

It shows that all the transactions fail. Yet no complaints raised for that service!

My question is, what is a completely failed transaction?. This because Dynatrace shows 100% failed transaction which is not the case!

Br,

Chris

dennis_mispelba
Inactive

Hi Christopher,

a transaction - a PurePath - in AppMon is the visualization of an end-to-end communication from request to response. Along that route (from frontend via middle-tiers all the way to the backend) failures can occur. However, not every failure must lead to a response of the frontend, which classifies the request as a failure. You can have response codes of 200 by the frontend, although failures occurred.

I recommend you check the different status a PurePaths can have (especially green and orange vs. red)

https://community.dynatrace.com/community/pages/vi...

In short: orange means PurePaths contains errors, red however means the PurePaths contains errors and also does not have a response code of 200! Thus, those errors qualifies to mark the PurePath (transaction) failed!

Those transaction marked as failed contribute to the failure rate, which you encounter in your scenario.

I recommend you analyze the PurePaths by right clicking and drilling down to the PurePath from the data you pick up in your business transaction.

I simply think your business transaction is not configured for what your actual goal is.

Best regards
Dennis

christopher_teb
Organizer

Thanks Denis,

In this environment am in, they are mostly interested in the issues that could cause a total failure of a transaction/purepath. What are the types of failures that caused a total failure of a transaction/purepath.

Br,

Chris