TM1 Forum

Has anyone seen this problem? It is the first time I've ever seen it.

We have a TI that reads a text file and updates a cube. Just updating a calendar dimension with American and TM1 formatted dates with an integer as the primary element name. Text file has the key, the American date and TM1 date in three columns.

Broke this up into two groups, one for 1/1/1920 to 12/31/1959. Created the file with a spreadsheet, saved it as a csv file. It loaded just fine.

Then, using the spreadsheet to create the second group, 1/1/1960 to 5/31/2014. Saved that to the same file, just overwriting it. Ran the TI, nothing updated. Added asciioutput and Lo! the same 1920-1959 records are what is being read! Open the file with a text editor, the second group of records is what is in there. The Preview on the data tab screen shows the expected 1960 group of records, but that is not what it is processing. Data Source name and Data Source Name on Server entries match.

Have double & tripled checked the datasource tab. Have stopped and restarted the server. Have searched the sources folder and no files contain the 1920 set of data. They do contain the 1960 set of data which, again, is what pops up on the preview on the data source tab.

I have a user who has reported this a number of times over the last couple of years, but I never saw it until now. Just thought it was user error.

It seems to be holding the old source file in memory. I've never seen anything like this. Has anyone seen this behavior and how is it prevented?

Your profile shows both 10.2.2.7 and PA2, but you haven't indicated which one this is in. Unless it's PA2 and the server team has done something funky relating to this in PA2 (which I doubt), I don't think that what you suspect is in fact happening.

You need to disregard the preview window. It will be pulling from whatever the "Data Source Name" is, and that will be relative to your CLIENT machine.

That's because you may not have access to the server's own paths, and vice versa. You need to be able to see a file to edit the TI process, so TI provides the two different paths; one for the client and one for the server. If you have a data source which is shown as C:\Data\Myfile.cma, it will be C:\Data on YOUR hard drive. If the Data Source Name For Server is pointing to the "same" location, it will be looking at its OWN hard disk drive. The files may have the same name on your machine and on the server, but they can be completely different files.

The name of the file that the server processes will be in the DatasourceNameForServer variable. I recommend that you export that with your ASCII Output commands, then RDP to the server if you can and look to see what is in that location from the server's point of view. I'd bet a few bucks that you'll find that it's a file that contains the earlier dates.

Okay, never mind. Problem was this was a one off and running this in PA2.0. I started the process off in PAW, selected the datasource there originally.
The Data Source name for Server was model_upload\filename.csv instead of the path and file name in the first data source entry block.

Current taking eggs one at a time and smashing them on my forehead.

Alan Kirk wrote: ↑Fri Aug 17, 2018 9:00 pm Your profile shows both 10.2.2.7 and PA2, but you haven't indicated which one this is in. Unless it's PA2 and the server team has done something funky relating to this in PA2 (which I doubt), I don't think that what you suspect is in fact happening.

You need to disregard the preview window. It will be pulling from whatever the "Data Source Name" is, and that will be relative to your CLIENT machine.

That's because you may not have access to the server's own paths, and vice versa. You need to be able to see a file to edit the TI process, so TI provides the two different paths; one for the client and one for the server. If you have a data source which is shown as C:\Data\Myfile.cma, it will be C:\Data on YOUR hard drive. If the Data Source Name For Server is pointing to the "same" location, it will be looking at its OWN hard disk drive. The files may have the same name on your machine and on the server, but they can be completely different files.

The name of the file that the server processes will be in the DatasourceNameForServer variable. I recommend that you export that with your ASCII Output commands, then RDP to the server if you can and look to see what is in that location from the server's point of view. I'd bet a few bucks that you'll find that it's a file that contains the earlier dates.

Sorry, Alan. I was typing my response addition while your answer came through and did not read it first.

It does appear that PA2.0 kept a copy of the file as I had dropped it into the TI while working in PAW. Lesson learned about it.

FWIW, this was a one off to do a one time load of a text file. I always set dataSources in the Prolog tab otherwise.