TM1 Locks

[Suharsh]

Hello All,

I have a Query regarding TM1 locking. We have a chore that runs for almost 40-50 mins and users are locked out for entire duration.
The chore has a few data load processes but most of the processes are not data load. They are either sub processes or to set some variables and like that..

Some of the team members argue that it is expected that users will be locked out till the chore does not completes. But I believe it should not be so, and the users should be locked out only for the data load processes.

Can breaking up the Chore and running the individual process one by one, help this issue?

Any help would be greatly appreciated.

Thanks

[qml]

First of all, try switching Parallel Interaction on in your TM1 server. If your processes do not make any metadata changes then there should be no locking with PI on. In case of any dynamic subsets being recalculated, subsets or views being created or changes to dimensions, there would be locking. Some of it you can avoid by e.g. reusing existing subsets and views instead of creating them on the fly etc.

[George Regateiro]

You should find this link helpful on PI http://www-304.ibm.com/support/docview. ... wg21460595.

You will have to look at your processes and verify that you are not doing any meta data updates and the like, but you should see minimal locking in TM1 with PI turned on.

Also I dont know if it is in the cards for you but TM1 10 has introduced the ability to release locks in between chore processes. If your chore must create a lock this feature might be helpful to you in minimizing the lock. Something to think about in upgrade considerations.

[PlanningDev]

Untill TM1 10.1 any chore that contains meta data updates in any process within the chore will cause locking for the duration of the chore. You can get around this using TM1RunTI.exe to execute your updates rather than a chore or using ExecuteProcess in serial in a master TI. However, it creates more code, therefore more potential maintenance and you still have to be careful as it's not quite as simply as simply serializing calls to TM1RunTI.exe. In simple cases its that easy, but if you're TI's read data from objects updated by previous TI's then it gets a lot more difficult to control.

The easy way to make the locking go away is to put all the metadata updates in a chore in serial and instead of the actual process that calls the data load, simply create a process to call TM1RunTI.exe to call the data updates. This way you can set the ExecuteCommand wait setting to 0 so the chore will complete and release the locks from metadata updates and the Data update will continue on.

[Suharsh]

Thanks Everyone for your inputs.

I will try implementing the current process through RUNTI command line utility and since we have meta data updates as well in some processes, it is causing lockout for the entire Chore, so will see if it can be segregated and run.

Thanks everyone I will keep posting my observations if I have any.

-Suharsh