Locking in EntityFramework - calling empty stored procedure releases lock acquired using sp_getapplock

Question

I have a patient table in the SQL Server database. Clients connect to the database and checkout one of the patients and do some operations on it and save it afterwards.
Since multiple clients might access the patient at the same time. I wanted to implement pessimistic locking mechanism, So that if one client can acquire a lock to a particular patient at a time.
Also if one of the clients locks a particular patient and either crashes or closes the client application the lock should be released. In case of crash the lock should be released when we detect that the connection of client with the database is closed.
I found out that I could use sp_getapplock and sp_releaseapplock to
do achieve this and specify @LockOwner
= 'Session' when calling sp_getapplock,
this ensures that the lock is released once the database session with the client closes.
From the code, I call a stored procedure in database to acquire the lock. This acquires the lock correctly and if i close off the client app or the client app crashes the lock is released. This is behaving as I wanted it to.
The problem happens when I call another stored procedure which is supposed to release the lock. As soon as the stored procedure is entered the lock is released, before we even execute sp_releaseapplock
@patientId, @LockOwner = 'Session' the line which is actually supposed to release the lock.
I tried creating an empty stored procedure (just commented out
DECLARE @result int
EXEC @result = sp_releaseapplock @patientId, @LockOwner = 'Session'
This stored procedure still releases the lock I acquired using sp_getapplock.
I can't seem to understand why this is happening. It should only release when I explicitly call `sp_releaseapplock @patientId, @LockOwner = 'Session'``
CREATE PROCEDURE [dbo].[uspReleaseAppLock]
@patientId nvarchar(max), -- Patient ID
@ReturnValue int Output
AS
BEGIN
DECLARE @result int
EXEC @result = sp_releaseapplock @patientId, @LockOwner = 'Session'
SET @ReturnValue = @@SPID
END
Could someone tell me why this is happening. What am I missing here ? I checked that the session ids of both the stored procedures (for acquiring and releasing the lock) are same, so it dose not seem that a new session is being created by entity frame work
for each call to the stored procedure. Something else is going on which I am not able to comprehend. Any ideas ?

Answer 1

Hi oalvi,
The Stored Procedure(SP) sp_getApplock initialed with "@LockOwner='Session'" is
session scoped. Your procedure
[dbo].[uspReleaseAppLock]
did the same work as you explicitly call "sp_releaseapplock @Resource = @patientId, @LockOwner='Session'", after the SP
uspReleaseAppLock
called or the current session ends, the lock released.
It doesn't make sense to call the uspReleaseAppLock to release the lock initialized in another session. If you try, you get the error like below, as I mentioned, the lock is session scoped.
"Cannot release the application lock (Database Principal: 'public', Resource: 'patient1') because it is not currently
held"
Regarding the same session ID you get from calling 2 SPs separately in Entity Framework(EF). As per my limitation knowledge on EF, If the maxsize of the project connection
pool equals 1, you may get the same session always.
Some cautions for sp_getApplock
Locks can be explicitly released with sp_releaseapplock. When an application calls sp_getapplock multiple times for the same lock resource, sp_releaseapplock must be called the same number of times to release the lock.
Due to the connection pool, the sessions may last long, that's what you need to take into account.
By the way, is there any reason that you have to use the sp_getApplock? What is the procedure doing that you cannot rely on SQL Servers build-in concurrency control mechanisms?
If you have any question, feel free to let me know.
Eric Zhang
TechNet Community Support