TM1 Forum

TM1 10.2.2

I'm trying to track down a setting or something that would effect the amount of time it takes to open a subset editor (SE) for any given dimension. We have several dimensions that are quite large and the amount of time it takes to open the SE has become problematic. We had assumed it was related to the number of elements in the dimension...but then I did a little experiment and found that not to be the case.

I took two dimensions that had a large disparity in the amount of time it took to open an SE. Neither dimension has dimension level security, neither has ruled attributes, and neither has an MDX subsets.

Dimension 1 = ~100 elements and takes 2-3 seconds to open
Dimension 2 = 2 elements and takes 15-18 seconds to open

Also, I'm just opening the subset editor from the Server Explorer by double clicking on it from within Perspectives. The issue is further exacerbated when opening the SE/drop downs from the web.

My Question...what would cause TM1 to take 15-18 seconds to open a subset editor on a dimension with only 2 elements in it and no element attributes to speak of...yet a dimension with 50 times as many elements and several element attributes takes only 2-3 seconds.

Thoughts?

Terry

Hi

That is a really long time to open a subset editor. 100 elements is still pretty small. I would check that the server is not running out of physical memory and that it is not running out of disk space.

Regards

Paul Simon

We think we figured it out...we're restarting the server now after deleting some 14,000 control object subsets which are being created (one at a time) each time a specific process is run.

We originally coded the process not to delete the subset when the process finished but to instead re-use the subset next time...thereby limiting the locking/blocking that resulted from adding a subset to a dimension via TI process. But something has gone haywire and it just kept creating more and more of them.

I originally stated that subsets weren't the issue because I couldn't see the control object subsets. They wouldn't display in Perspectives because there were too many of them. I had to go into the folder structure on the back end to see all the .SUB files.

I'll let you know if that turns out to be the issue.

So, we've confirmed it is the subsets that are being created by processes. Unfortunately we're not able to eliminate all of them so this is going to be an ongoing issue.

Does anyone know if there's a way to get TM1Web to ignore control subsets? That would solve a lot of our problems.

terry720 wrote:So, we've confirmed it is the subsets that are being created by processes. Unfortunately we're not able to eliminate all of them so this is going to be an ongoing issue.

Does anyone know if there's a way to get TM1Web to ignore control subsets? That would solve a lot of our problems.

What do you mean you're not able to eliminate all the subsets? Why on earth not?
If all the subsets (and presumably views as well) are created with TI then there must be a naming convention in which case it is straightforward to do use the WildcardFileSearch function on a loop through the data directory sub dirs and delete first all TI processing views then all TI subsets. There are bedrock processes that do exactly this with no need to recode anything.