RSPLSE Locks_Concurrency Lock

[manohar_pappireddy]

Hi,

We are on SAP NW BPC 10,BW7.3,EPM-SP14.

One user's clear DMP failed with Error Obtaining Concurrency lock(Foreign_lock)message.

I have checked RSPLSE locks.At that time no locks were there.

Checked DMP logs.At that particular time there was a currency conversion job running.

But the organizations in Clear and Currency Conversion DMPs are different.

As per my understanding,locks will not happen even at that time some users trying to read the data.

Issue2:

And today I have observed RSPLSE lock entry today.

One job is currency conversion and other job is tax calculation.Here also organizations are not same.

But still there is an entry in RSPLSE.

So what are the overlapping criteria for locking mechanism?

Thanks,

Manohar

[former_member182709]

Manohar,

The issue should be solved by setting parameter RECLEVEL_NR to 10/50. See blog http://scn.sap.com/community/epm/planning-and-consolidation-for-netweaver/blog/2014/07/23/have-you-e...

Charlie

[manohar_pappireddy]

Hi Charlie,

I will test these parameters in our development system.

Thanks,

Manohar

[former_member182709]

Hi Manohar,

I believe this issue can be solved by changing this parameter. Looking forward to your feedback.

Charlie

[former_member200327]

Hi Manohar,

Issue 1: did you see concurrency lock that has time later than error in your DMP? If yes, then your error was overridden by next error. RSPLSE keeps only last error. Additionally, do you have 1 App server in that box or more than 1?

Issue 2: BPC locks data sometimes on each individual member and sometimes on intervals. So, if one process locked A - C and another is trying to lock B - you'll get an error. So, please check in those 2 lock if at least one of them is on interval? Also, please check if you see 2 user IDs in Locking Conflict screen or second ID field is empty?

Regards,

Gersh

[manohar_pappireddy]

Hi Gersh,

1.Yes,we have 10 app servers.Till now I was under the impression that RSPLSE locks are not app server speicific.Today I have checked and different locks in different servers.

Little bit confused here.If centralized locking mechanism not there in place,how correct concurrency lock happens?If one user connects to Server 01 and writing the data for a set of selection and other user connects to different app server and trying to modify the same set of data,concurrency lock happens or not?

2.One user was running currency conversion on "Operating Companies" Node and other user was running Clear job on "Support Units"

So there is no chance of overlapping.We have only one hierarchy for this dimension.

Thanks,

Manohar

[former_member200327]

Hi Manohar,

1. You are correct - locking mechanism is central. I.e. it checks against common set of locks no matter from which App server request comes in. Reporting is App server specific - that info is stored on App server from where locking request came from. That's why you can see different info on multiple App servers. But this is a good thing (if you know about it:), because you can see not only the last lock if it came from a different server.

2. As you probably know lock happens at the base members hence having them under different Nodes doesn't guarantee that lock will never happen. This depends on how those base members were locked: individually or as an interval. Can you please check in that Locking Conflict for that Dimension if members in last 2 columns same or different? This will show if lock was individual or on intervals.

Regards,

gersh

[manohar_pappireddy]

Hi Gersh,

1.Thanks a lot for the response.Now the first issue is cleared.

2.Lock was showing as interval for Organization dimension.

Can we control this behavior?

I feel it is not correct to have locks even selections are different.

In our system there are 6 teams (based on Organization dimension).

When one team member performing calculations,other team member should not have any issue to modify his Organizations' data (with copy,clear,input,calculations).

How can I achieve that?

Thanks,

Manohar

[former_member200327]

Hi Manohar,

Yes, you can control that.

Perhaps one of the Packages was running on multiple Org members. BPC decided that it's too many to lock them individually and locked as an interval. Org member(s) for the other Package probably fell inside that interval.

Let's say 1st Package ran on Orgs ABC, ...,  KLM and the 2nd Package ran or Org DEF. 1st Package will create a lock on ABC - KLM and 2nd will try locking DEF. This will create a locking conflict.

So, solution would be to make BPC think that 1st lock doesn't have enough members to lock them as interval. For this you have to find out how many members of Org Dimension can be in one Package and make RECLEVEL_NR and INTERVAL_NR at least that big. Then BPC will lock them individually and if 2nd Package runs on different Orgs they won't lock each other.

Just don't make those parameters too big because it will slow the locking down.

Regards,

Gersh

[gregor_dieckmann]

Hi Manohar,

please read note 928044 (sizing of the BW enqueue server) and the BW standard-specific usage of the BW lock server.

Regards,

Gregor

[manohar_pappireddy]

Hi Gregor,

Thanks for the reply.

There are no charecteristics "Lock charecteristics" tab in RSPLSE.

Does it means it takes all the charecteristics for locking?

Thanks,

Manohar

[gregor_dieckmann]

Hi Manohar,

- ensure that you use the right BPC generated InfoCube name (something /CPMB/...)

- if nothing was explicitly maintained in RSPLSE all characteristics are lock relevant (RSPLSE, second tab, table on the right hand side; don't forget to press ENTER after input of the technical InfoCube name

- if there is no lock relevant characteristic each lock request would try to lock the whole InfoCube

- the most relevant RSPLSE settings in the BPC usage are explained in the above sizing note, so make 'reasonable' settings for

-- Lock Attempts

-- Wait Time in Seconds

(read the F1 help...) in RSPLSE, in expert mode. And read the BPC Administration Guide, Write Back Parameters.

Regards,

Gregor