10.7 - Procedures just 'stop' populating cube without any change in cubes/entities/procedure... he

Has anyone else out there had issues in 10.7 with Procedures that were working just suddenly stopping working and populating cubes with 0's......AND known how to get around it?

It can be thins as simple as standard/straight dataflow that 2 minutes ago populated the target cube, add a new action to the end of the procedure for brand new cubes, then run the procedure again and now for some unexplained/BOARD reason the first action in the procedure which was no touched is push a pile of 0s to the target cube which clears all the data.

We're trying to build a PoC for a presentation next week and it's more than slightly concerning that we have absolutely 0 consistency in the behaviour/performance of BOARD Procedures.

We have even tried restarting the service AND server ... and that poses it's own issues in that it can then take half a dozen or more 'logins' before it stops booting you out and finally allows you to stay in. I can have a capsule open, so am clearly logged in, then when I click to go to DataModel it says i've been logged out and have to sign in again.

I've also been unable to make a structure change to a cube AND have it save with any certainty. The cube screen will show one structure, the same cube in the dataflow layout another, I go back and re-open the cube in the cube screen and the changes I saved are no longer there.

Perhaps the above are all linked, but they don't seems to be machine specific as my colleagues are experiencing the same.

If anyone can give some ideas on how to get around this it would be much appreciated...as the logs don't say anything.

------------------------------
Brendan
bbroughton@bi5.com.au
Australia
------------------------------

Find more posts tagged with

Accepted answers

All comments

Alessandro Como

Good Morning,
have you tried to debug the procedure to be sure that there is no active selection before the dataflow and see the layout of this?
try to create a new cube and create a new procedure with only the dataflow to test the correct working.
For the problem with datamodel i think that it could be corrupted... have you got some backups?
Regards

------------------------------
Alessandro Como
IT Manager
COMINI SRL
Italy
------------------------------
-------------------------------------------

Brendan Broughton

Thanks Alessandro, we spent an entire day trying to debug and see if we could explain the behaviour, we also extracted cleared and reloaded everything, and restarted everything, and still couldn't work out what was causing it.

As I mentioned the dataflow is fine, it works perfectly... the issue is that it's only SOMETIMES..it doesn't always work..yet there is no change in anything in the procedure OR in the source cube. I can run it, see the values in the Dataview, run it again, then the values in the target cube disappear, yet they are still in the source cube...and there is absolutely no change in the actions or selections in the procedure, nor the structures of the cubes or entities.

I can only think there is something funny happening in the DB too - if only there was a way to find out what that is and how it went bad. We managed to find one cube that BOARD could no longer display in a dataview for some reason so we removed that, but not confident that's the end of the problems.

------------------------------
Brendan
bbroughton@bi5.com.au
Australia
------------------------------
-------------------------------------------

Björn Reuber

Hi Brendan,

what is the structure of the Cube? Has it a sparsity? Are some (except Time entity) entities set to dense?

Regards
Björn

------------------------------
Björn Reuber
Support Team Manager
Board Deutschland GmbH
Germany
------------------------------
-------------------------------------------

Brendan Broughton

Hi Bjorn, None of them were sparse...all dense It happened for a few of us across a few different cubes, mine had no time entities on them.

Some of mine were using [R] entities and/or their non-[R] versions.

Some times moving from a cube that had 2dims, both [R], to a cube with the same structure using the non-[R] entities wouldn'y work.

We were wondering if the [r] had broken the db somehow

------------------------------
Brendan
bbroughton@bi5.com.au
Australia
------------------------------
-------------------------------------------

Björn Reuber

Hi Brendan,

not related to the R, more to the all Dense cubes. Had a similar issue, that on all Dense cubes changes are not displayed directly. In my case a Save db to disk step solved it. Can you try to add those steps after a df to dense cubes?

regards
Björn

------------------------------
Björn Reuber
Support Team Manager
Board Deutschland GmbH
Germany
------------------------------
-------------------------------------------

Brendan Broughton

HI Bjorn, Thanks for the tip!!! We'll put that in and see if it makes a difference. Any idea why it happens, or what conditions in the dataflow make it more likely to happen?

------------------------------
Brendan
bbroughton@bi5.com.au
Australia
------------------------------
-------------------------------------------

Brendan Broughton

Hi Bjorn, that's not the issue, as i've encountered it again already and it doesn't make any difference.

What is even more concerning in this one is that the first DF in the procedure won't put ANYTHING into the cube, yet the second DF (which was a COPY of the first with an algorithm thrown in) will populate the cube. I've even gone all the way back to doing nothing in the DF other than putting a '1' into the cube... and it doesn't work...I get an empty cube as shown below.

That is the result of this extremely basic procedure with all the other actions disables.. and the save to disk applied..

but if I enable 7 and 8..

This is the working dataflow..

The only difference between that DF and the original not working DF was the original didn't need the d block, it just used the c block (to apply to hours, as opposed to day)

I now get data in the cube as is expected from the calculation.

I've cleared the cube multiple times and it keeps getting populated with the second dataflow only.

How on earth can that even be possible in BOARD?! If it can't flow a 1 to the cube, then how/why can a more complex multi-block algorithm possible populate it? I have tried everything, adding a new member to that entity to see if that's the issue, copying the one working dataflow and applying with the other members selected, absolutely nothing worked.

But in trying all that now the 1 dataflow that WAS working doesn't work and now my cube stays empty - and all that I changed in the structure of anything was adding a new member to the Allowance Base entity.

I can only think the DB is buggered...but the fact they get buggered so easily and frequently is very frustrating.

------------------------------
Brendan
bbroughton@bi5.com.au
Australia
------------------------------
-------------------------------------------

Björn Reuber

Hi Brendan,

which version are you using? Can you please send some screenshots for DF number 5?
Or (if its a small db) can you open a TK with this case, then I can try to analyze it directly

Regards
Björn

------------------------------
Björn Reuber
Support Team Manager
Board Deutschland GmbH
Germany
------------------------------
-------------------------------------------

unknown

Hi Brendan,

Encountered yesterday an issue as well in a prod db. We were testing multiple scenarios, but a datareader which had to result a quantity=1, gave a quantity=2 for some cell combinations.

After wasting a full day in the web client trying to find workarounds, we opened the db in the windows version where we got the instantenous pop-up that a maximum item number went over the limit. After fixing this, the system behaved as normal again. Unfortunately that pop-up doesn't show up in the web client, so maybe it is something you can check.

Probably you will have done that already, but it could be something to take into account for anyone else reading this thread having an unexplicable reason why BOARD was providing inconsistent results and expected in the back of their head a similar pop-up in the datamodel in the web client.

Cheers,

Jonathan

------------------------------
Jonathan Baetens
Team Leader
TOOLSGROUP ITALY SRL
Italy
------------------------------
-------------------------------------------