Hi All,
When I look at the list of transactions, I see a responsetime. Once I click on a transaction I see a list of the thansaction (in this case a aspx page) and also all discovered WCF services.
I notice:
1) that these services are NOT included in the total responsetime that I see in the list page. Also the percentages of the WCF services can be a multitude of the total responsetime of the whole thing
2) The number of calls of the WCF service are set to 0, while I am sure they are being called
(see attachment)
Any ideas, especially on the first one?
Or is that by design?
If so, is there a way to include the service calls?
Owin
Hi Owin,
Regarding query on response time , Can you please provide screenshot for flow map in default view (transaction flow view) and also screenshot similar to one listed below with graphs for response time of BT and art metric of its external calls in view of the graph in metric brower in UI, we see similar behavior in local and the behavior is standard across versions it looks:
Regarding other query:
- We see similar issues in other controlelr environments but user selected time trend (in this case last 3 days as per below screenshot) the data is not reflecting due to some UI cache issue for particular view in updating load as referred in attached screenshots
- We have some known bug on 3.9.x controller UI version around tree view in syncing data as per flow view or load/cpm graphs for selected time, We would like to inform bug around this is already fixed as per internal request on 4.0 UI version which is released already. (internal bug referred CORE-36362 for team reference)
We request you to check the behavior for lower time trends like last 1 hour or other to see if it is UI refresh issue or bug as we stated above.
Let us know if that information helps.
Regards,
Arun
Hi Arun,
For the first issue (services not included in total time):
- You already have the flowmap in default view. It is in the original post. Correct?
- I have attached the metrics browser data in this reply
Is this what you are looking for?
For the second issue:
I tested with 1 and 3 days, but in both cases I see a lot of 0 calls. No difference there
Regards,
Owin
Hi Owin,
We confirm this is expected behavior as the data/ART ms in tree view shown is spent on that tier/backend and the Response time at BT level is aggregated one for the API, Hope that clarified the product behavior.
ref doc: https://docs.appdynamics.com/display/PRO39/Measure+Distributed+Transaction+Performance
Regards,
Arun
Thanx for your reply.
So how am I able to see the total response time of my transactions in a list?
Regards
Hi,
Response time shown in Response time graph Or in "business transactions" screen for particular BT is the aggregated response time over multiple calls and if you want to see per call , we request you to check total response time /ART ms field in transaciton snapshot, Hope that information helps.
Regards,
Arun
Hi Arun,
I have started this whole thread just because the "Business Transaction" screen is NOT showing aggregated response time... So in my original example on the BT screen 5950ms was shown and not the aggregated time of 29977ms
Owin
Hi Arun,
Do you maybe have an answer on my last question (how to get the total affregated response times in a BT-like overview)
Thanx again
Owin
Hi Owin,
Are you looking for something like rest response manipulation case where rollup arg used on rest api request as stated in another post http://community.appdynamics.com/t5/AppDynamics-Discussions/Trying-to-show-slow-calls-very-slow-call...
If this is the case and you are looking for clarification on response time on tree view we request you to raise help ticket for better clarification from support team as i see the data in tree view looks to be standard behavior.
Regards,
Arun
Hi Arun,
I think I am having trouble with the default behaviour, so maybe you can verify this for me:
Suppose I have a screen in my application called "screen.aspx" and for this screen to load, the user has to wait 10 seconds. And say that 10 seconds is 2 seconds .Net and 8 seconds a webservice call.
Will then the response time in the BT overview show 2 seconds and not 10?
If so, I will open a ticket with the helpdesk. But I just feel so strongly that in this example the only thing anybody would be interested in is the value of 10 (and drill down to the specific components later)
Thanx,
Owin
Hi Owin,
There seems to discrepency in Tree view, we request you to see transaction response time distribution using "View old dashboard" "View html dashboard" which is default flow map view for now, Or check the behavior after controller upgraded to 4.x latest version as we have couple of similar bugs around the response time distribution in sub calls in tree view which is fixed already in 4.x version release. as referred in below screenshot see if that helps to clarify response time spent on exit calls and remote services and correlated tiers :
In the below example:
146 ms = 1.2 ms on MovieSearchSite IIS tier + WCF exit call ART ms is 32.8 ms + 4.6 ms on Coreservices tier and ADO.NET/JDBC ART ms is 107.4 ms (approx equals to 146 ms of ART of BT /login.aspx)
Regarding other query, we confirm if the calls are not async type the total response time of BT should be 10 and the value 10 should be shown as sum of response times spend in flow view of BT dashboard on tiers/backends involved, Let us know if that answers your query.
Regards,
Arun
Hi Arun,
I do not have an issue with the tree view. So let's focus on 1 problem: the BT response time list:
You say that in my example 10 seconds should be shown. But I am seeing 2 seconds instead of 10 in the list.
Since this behaviour is not correct, what can I do to solve it? Update the controller? I now have 3.9.5.2 build 11
Owin
Hi Owin,
There seems to disconnect in the screen you are referring Can you point the screenshot with 2 ms shows up in controller UI , please make sure the screenshot have view including component details, as i see in UI we do not show sum of ART ms for calls in given time period (only way is to to use rest api using rollup as i stated earlier) and the data shown in UI aggregated data over multiple calls, Please do send the screenshot as i do not locate the screenshot in other than tree view ex: http://community.appdynamics.com/appdynamics/attachments/appdynamics/appdynamics-discussions/2446/1/...
Regards,
Arun
Arun,
The 2/10 seconds was just an example to simplify the problem.
here are 2 screenshots that show the Business Transactions for 1 certain Business Transaction (BT) and the Transaction flow (TF) when I click in that BT.
Owin
BT:
FT:
Hi Owin,
Thanks for clarification, We confirm this is due to UI discrepency in tree view and in combination with UI update delay when user navigated from Business transactions screen on drilldown of particular BT on navigating to BT dashbaord and delay in updating the total response time the data is bit more recent than in previous screen as ui takes time to load
and also regarding the sub calls of WCF / remote exit calls tree view have bug in old version of ui and we request you to refer BT flow view where right response time distribution shows and BT ART ms is properly shown and distributed over correlated components, please refer below screenshots, Let us know if that addresses your concerns
Regards,
Arun
Hi Arun,
Sorry, It does not:
1- I am missing the solution for the discrepancies. Is it solved by installing controller version 4, or will it be solved in a next release?
2- What do you mean with "old version of UI"? is that something different than the controller? I have no influence in unpating the UI, since I am using the SaaS solution
3- I do not feel that 147.4% in your screenshot is a bug. The actual bug in my view is that the top level is not aggregated that should be 135+263+199+38 ms. If that is not the case than 147.4% is also not a bug... So should it be aggregated, or should it be like it is in your scrceenshot?
4- Please fix the BT overview list. I cannot go into the flow view for every single BT (we have 140 of so). I need to be able to see total response times for all BT's in 1 overview
In general it is hard for me to understand your english, so I do not know if I am interpreting everything correctly. I am still not sure if things (especially point 4) are by design (which seems very onlogical to me), or if it is a bug somewhere and if I or you can do seomething about it. I hope that with a clear answer on above 4 questions it will get a bit clearer
Thanx,
Owin
Hi Owin,
Please find our response to your queries inline:
>>>> 1- I am missing the solution for the discrepancies. Is it solved by installing controller version 4, or will it be solved in a next release?
Answer: We do not see tree view discrepency in 4.0 UI atleast as per screenshot below from 4.0.0.1 controller ui version:
>>>>> 2- What do you mean with "old version of UI"? is that something different than the controller? I have no influence in unpating the UI, since I am using the SaaS solution
Answer: we understood your concern and we request to use Flow view for now instead tree view which looks to have discrepency in updating distributed response times in older / 3.9.x controller ui version
>>>>> 3- I do not feel that 147.4% in your screenshot is a bug. The actual bug in my view is that the top level is not aggregated that should be 135+263+199+38 ms. If that is not the case than 147.4% is also not a bug... So should it be aggregated, or should it be like it is in your scrceenshot?
Answer: screenshot provided by me in previous update is to show same discrepency in old version, refer expected output in screenshot provided above in this update, in brief, the tier at parent level should only represent the time spent on that component on tree view which is <= 135 ms but not 135 if there is some time spent on sub tiers/exit calls like we see in previous post screenshot
>>>> 4- Please fix the BT overview list. I cannot go into the flow view for every single BT (we have 140 of so). I need to be able to see total response times for all BT's in 1 overview
Answer: We confirm the "'Response time (ms)" column represents the aggregated response time over the total number of calls under "Calls" column in Business transactions screen
Say: you have made 4 calls to /logout.aspx in earlier screesnhots provided and if each request taken as below response times: 10 ms, 100 ms, 20 ms, 30 ms than the Calls column represents value as 4, and "Response time(ms)" column as 40 (10+100+20+30)/4 = 40 ms We will not show sum of response times, we will show average over the multiple calls of BTs in BT list screen.
I apologize that if details i provided is not clear enough, I would suggest you to raise help ticket at help.appdynamics.com or email help@appdynamics.com if above explanation does not answer still so that team from support will try to answer your queries in better approach, i tried to clarify your queries to the best of my information so far.
Regards,
Arun
Thanx,
So then I at least will update the controller to get rid of the discrepancies. Thanx for that.
About my point 4: you say "'Response time (ms)" column represents the aggregated response time over the total number of calls under "Calls" column in Business transactions screen.".
With aggregated response time you mean that it should include calls to other tiers, like my WCF services, correct?
If so, then we can conclude that in my setup this is not the case. Will that issue also be solved when I upgrade the controller?
Regards,
Owin
Hi Owin,
If the remote service calls/exit / JDBC other backends calls invoked by particular BT in Business trasanctions screen synchronously, Yes you are right the "Response time (ms)" includes the response time of backends,
But regardles of synchronous Or asynchronous the Calls made to backends / WCF by BT are not part of Calls column, let me explain with an example
You are monitoring Class (class1.method1) as BT /servlet1 and have made 10 calls to class1.method and each taken 20 ms for each of the 10 calls then Respnse time is 20 ms and Calls column value is 10 , but say each call to Class1.method1 is invoking WCF backend/ JDBC backend 1000 times than the Calls column in Business trasnsaction screen will not represent the internal calls by class1.method1 as we are monitoring only class1.method1 calls in BT grid, however the response time taken by sub calls by class1.method1 are part of "Response time (ms)" , Hope it is clear
However you can see calls/min, response of backends under Analyze -> Metric browser -> backends section for each backend, Hope that information helps.
Regards,
Arun
Hi Arun,
It is clear to me.
Is it clear to you that, as I have shown in my screenshots, the sub calls are NOT included in the BT grid?
So, you state: "however the response time taken by sub calls by class1.method1 are part of "Response time (ms)". I do NOT see this behaviour in the BT grid.
Question remains if this is solved when upgrading the controllen from 3.9 -> 4.0?
Owin
Hi Owin,
It depens are the calls to WCF/backends by BT are synchronous or asyncchronously made (in separate thread) if it is async calls that it is expected behavior, Hope that clarifies.
Regards,
Arun