TM1 - disappearing data

[joannetheodore]

My client insists that they could see their Feb payroll figures last week, but they have now disappeared... I've re-run the process that my predecessor used, but it is not uploading the figures to some of the companies in the group. Has anyone seen this before? Thanks in advance!

[Steve Vincent]

Hi Joanne, you'll have to be more specific on structures and the process used to upload data to be able to get any help on this. If the processes have not changed then something else must have, either a view the process uses or a structure it relies on. Are any error messages seen when the data is loaded?

[joannetheodore]

Hi Steve,

No error messages - it just says "process completed successfully"...

The format of the report being uploaded has definately not changed.

[Michel Zijlema]

Hi,

Could it be that you have a destroyed hierarchy in one (or more) of your dimensions - because of which values will load OK, but will not consolidate to the top level element(s)? Did you check if values are there on N-level (or by Export as ASCII data...)?

Michel

[stephen waters]

Joanne,
What is the data source you are using for the process? Can you check that the figures for the "missing" companies are in the source file? If so, and there is no error message from th TI process then are those n leve figures loaded in TM1?

[joannetheodore]

If I had destroyed the hierarchy, would that not have impacted the figures that are in there for Jan?

[joannetheodore]

Hi Steve,

It's a csv file - I've checked that the details are actually in the csv file, and they should all be going in on the n level...

[mattgoff]

These sorts of specific issues are very hard to solve in a forum as they're very environment-specific, but I would start by trying to find one specific row in your source data that's not making it into TM1 and using it as exemplar data to trace through your model and TI process. Usually that will lead me to the source of the problem. I would probably ignore the specifics of the trouble reports you've had so far-- I find that they are often a red herring and bias your bug hunt. You can also add some ASCIIOutputs to see what's going on (oh how I wish TI had a debugger, or better, was written in a "real" language ).

Matt

[Steve Rowe]

Another thing you can do is export the target area from the cube using right click and Export as ASCII. Leave the definition of the export as general as possible and make sure all the tick boxes are ticked. Hopefully that will be small enough to import into Excel, then a couple of filters and you should be able to find the data.

From what you say I'd guess there is a structural issue in the cube.

1. The data is in the source file. Tick.
2. The TI process executes without error. Tick.
3. The TI is processing the file you think it is.??? Try and run the load with the source file open, if it does not error then you are processing a different file.
4. Since the TI is executing without error and you are processing the correct file the data must be making it into TM1.
5. Therefore the error has to be in the way the data is being viewed, either the structure of a dimension means the data loaded at the N level is not showing at the C level, a rule is preventing the display of C level data or the users are just looking in the wrong place.

HTH
Cheers

[Martin Ryan]

As another diagnostic tool, make sure you have logging turned on before running the TI process then check the transaction log afterwards. I've seen a few situations where people have written to the wrong scenario or year, so the data hasn't shown up where they expected.

Martin

[Jeroen Eynikel]

Some more things to do / check:

- Do you get a single record for a cell in TM1 in your source, or are there several that need to be added up?

If it is the second case and you are not accumulating while loading only the value of the last record for that cell is going to be displayed.

- Do a savedataall in the epilogue of your process, certainly if you have logging disabled.

I do agree that the easiest way is to investigate a single specific case that seems wrong first.

Jeroen

[Klopse]

Joanne,
please check another kind of reason. If there is an active rule on the target now and the process loads data without warnings (TI formula like IF(CellIsUpdateable(sCube,D1, D2...), CellPutN, Do nothing) you will get no effect on target.
Sometimes, some "specialists" activate rules by an attribute or a cube value. If that happens all data on affected cells will be deleted. If You deactivate the rule by deleting the attribute value, the target is clear...and the data "disappear".

Regards


P.S. Be sure to activate an "save data all" chore to prevent data losses by unexpected server crashes.

[jorelb]

I have seen this before and it actually happened to me last night - was the erring element recently added? I use a dimension worksheet to update our account dimension and use a process that also passes a parameter to a sql script that extract data from Oracle. In both instances this happened, it only affected the newly added elements and their attributes were not completely defined (the attributes did not have a bearing to the process). I cannot identify 100% what fixed the issue but the data showed up after I defined the missing attributes and opened the views and subsets I used in the process views.

[jorelb]

It happened again in our Development environment and as soon as I defined all attributes and re-run the process the null cells got populated.