multiple instead of single server requests

We experience 2 strange problems using Microsoft Analysis Services Cubes with Flexmonster Accelerator .
 
Problem 1:
We have one cube with several dimensions (Agents, Products, Sales, etc.).
The Sales dimension have few columns Sale Date, Sale Number, AgentUID, etc.
The cube have one Sales fact (with some calculations we need and 2 columns SaleUID, ProductUID).
When building a report based on this OLAP cube and everything works as expected in the beginning. The problem start after we add the SaleNumber column from the OLAP cube as another dimension in the FlexMonster report config. The report is not erroring but it starts requesting data for each line of the report as separate call to the flexmonster accelerator service.
And when you have 8-9 thousand  lines this is taking a lot of time ( 10 or more minutes) it is working after finishes making the requests but it is very strange that this is the only column that causes this behaviour.
 
Problem 2:
We have multiple reports running from the above mentioned MS OLAP Cube, also we have added multiple filters in the Flexmonster config “dataSource”  \ “subquery” tag and all works as expected and quickly.
Problem starts when the added filters for Date/Time Ranges have values for the future or the past (just any time frame that does not exist in the cube data), the report starts requesting line per line all data from the Flexmonster Accelerator service and it after that finishes we got the empty report screen we expect. That only happens with date/time filters, if we try to filter by none existing agent for example we get empty report very quickly with no additional line by line request to the server
 
Any Ideas what can be making the flexmonster to do the separate data requests to the axelerator service ?

9 answers

Public
Dmytro Zvazhii Flexmonster November 1, 2017

Hello Nikolay,
Thank you for writing. Please try a fresh version of the accelerator from here – https://s3.amazonaws.com/flexmonster/ecrsystems/FLEXMONSTER-2017-ECR-NOV1.ZIP.
Let us know if the issue is still reproducible.
Regards,
Dmytro.

Public
Nikolay Petrov November 1, 2017

first impressions, looks good 
problem 2 is solved 
problem 1 is less of an issues now, we see it loading packets of 10K lines, loads fast as expected.
only think as problem may be the initial loading display, it says something like “10 0000 of 2 500 889 left” and it is looking like it will take forever to load but then you jump from 10-20% loaded to seeing the report.
will let you know if we have any more problems with it.

Public
Tanya Gryshko Flexmonster November 1, 2017

Hello, Nikolay,
Thanks for the update.
It is very nice to hear that your issues are solved.
Concerning the problem with the initial loading display, it is possible to change the messages you see here via localization. Please take a look at the following file: https://github.com/flexmonster/pivot-localizations/blob/master/en.json. Here you can find section "messages" which defines the texts for the pop-up windows. For example, here is how to add custom "analyzingProgress" message:

report : {
localization: { "analyzingProgress": "your custom message" }
// ...
}

Please let me know if you have questions.
Regards,
Tanya

Public
Nikolay Petrov November 8, 2017

we found a problem with https://s3.amazonaws.com/flexmonster/ecrsystems/FLEXMONSTER-2017-ECR-NOV1.ZIP 
the “flexmonster-proxy-ssas.exe” provided is ignoring the “CACHE_MEMORY_LIMIT” setting in the config and it is eating all the server memory after some time.
also, when can we get this as part of the normal Accelerator Service, running this as command line EXE is not the best long term solution ?
best regards,
Nikolay 

Public
Tanya Gryshko Flexmonster November 8, 2017

Hello, Nikolay,
Thank you for your feedback.
This new Accelerator is already available as Flexmonster Accelerator.msi. You are welcome to download it.
Regarding the issue with CACHE_MEMORY_LIMIT, please provide us with additional details.
Which values do you set for CACHE_MEMORY_LIMIT and how much memory is eaten as a result? The value for CACHE_MEMORY_LIMIT cannot be precise. If Accelerator, in general, is using more memory than CACHE_MEMORY_LIMIT it is trying to release some memory by cleaning the cache (if possible).
We are waiting for the updates from you.
Regards,
Tanya

Public
Nikolay Petrov November 9, 2017

we did download latest 2 days ago, unfortunately the AcceleratorService.exe version is not changed (shows 1.0.0.0 as before), but the file size is bigger than what we had before. That version of the AcceleratorService server and client are making the requests the same way as before, line by line and on future dates filters is not loading until all records are downloaded (same as my original description above).
As for the memory problem with flexmonster-proxy-ssas.exe you give us, we currently have CACHE_ENABLED=true and CACHE_MEMORY_LIMIT=1000 in flexmonster.config and memory is not getting under 1500 MB with no activity on the process for more than a hour 
I’m assuming CACHE_MEMORY_LIMIT is in megabytes or it is something else ?

Public
Nikolay Petrov November 10, 2017

update on the memory issue.
Left overnight the “flexmonster-proxy-ssas.exe” returned back to 1G memory limit set in the config.

Public
Tanya Gryshko Flexmonster November 10, 2017

Hello, Nikolay,
Thanks for the updates.
Our Flexmonster Accelerator.msi was updated, but the main changes were in the client-side component. Please make sure you are using version 2.407. You can check the version by clicking on the grid and pressing Ctrl+Alt+i.
After you update both Accelerator and the component, everything should work correctly.
Concerning the cache issue, we recommend trying to set different values for CACHE_MEMORY_LIMIT. For example, 600, 700, etc. CACHE_MEMORY_LIMIT is in megabytes and by trying out different values you can figure which values work best for you to not exceed 1GB.
I would be grateful for your feedback.
Regards,
Tanya

Public
Nikolay Petrov November 15, 2017

thanks, all works now. we where on the wrong version

Please login or Register to Submit Answer