Locking Periods using the Second Hierarchy in the Time Dimension

Question

Hi,
They are not usiing work status for this...
I am at a client, and they used the second hierarchy in the time dimension to lock periods.  If they left the second hierarchy blank, no data can be written to it.  To open the open the period, they code the Second Heirarchy with CONS_OPEN.  Just wondering if anyone has information on this? 
It was working fine, then we needed to run a transport to move BPC to a new machine, and it stopped working.  Not sure if anyone has done it this way before?
Thanks,
Chandra

Answer 1

Dear Kacper,
the way I understood, there are two separate questions in this post, namely:
I'm using a shared resource in multiple actors, and the code accessing the resource can take quite some time to execute. How can can I ensure this does not mess up the timing of one or all accessors?
I'm using a shared resource in multiple actors, and the code accessing the resource can generate errors if I try to access it from multiple places. How can I ensure synchronization between all accessors?
As for the first question, the optimal solution to separate the resource prone to timing issues in a different thread/actor. Put anything that is not strictly timed (file access, network comunication) in a separate loop. For each actor, you will get queue references to communicate with, so instead of logging the measurements in the actors where they are made, just queue them up for a different loop to process.
For the second part, the same logic still applies. Ideally, every singular resource should be handled by its own thread and nowhere else, so if anything else needs data from/to said resource, it can send a request to the dedicated thread or actor.
If, for some reason, this is not sufficient, you have to handle synchronization in some other way. There are a lot of techniques here, for example:
Create a named semaphore or lock as a part of the class. Have class functions use the semaphore before accessing the resource.
Use actor messages. Have a "resource in use" or "resource released" messages sent to all users whenever obtaining or releasing said resource.
Have a separate actor handle all resources, awarding them to threads needing them. This method also allows setting priorities between requests.
These are just a few examples, there are many other options.
Please let me know if this was helpful. 
Kind regards:
Andrew Valko
National Instruments Hungary