Question: Integration testing competing consumers howto about distributedlock

[madelson]

Hi @stephenpatten thanks for posting. I'm not sure I completely follow what you are trying to achieve and where you are getting stuck. Is the code snippet the code being tested or the code in the test itself? If you can refine down to a small complete example that would help me a lot. Some other thoughts:

Wondering how you might be able to use say, VS 2017 with xUnit to create a multiple consumers who will all try and grab a sqlDistributed lock?

Connection-string based locks are not re-entrant, so if your goal is to test by simulating a situation with multiple contenders for the lock it should be really simple. One option would be to launch multiple threads with Task.Run() or Parallel.ForEach and have them all try to use it. If you want something even more direct, you can use a single thread:

var @lock = new SqlDistributedLock(...);
using (var handle = @lock.Acquire())
{
    // code that runs in here that tries to take the lock will block
}

I'm not opposed to running up LocalDB/SqlExpress...

I do all my tests against SqlExpress, which is very easy. Really any database should work, although if you use a shared database then you can hit issues if multiple people are running the same test at the same time (this can be worked around by randomizing the lock name used by the test).

/*var handle = sqlLock.TryAcquire(timeout); // Does not enforce timeout */

I was surprised to see this. The intended behavior is that Acquire(timeout) throws after the timeout expires and TryAcquire(timeout) returns null after the timeout expires. Thus Acquire() with no timeout is like having an infinite timeout while TryAcquire() with no timeout is like having a timeout of zero. What behavior were you seeing?

[stephenpatten]

Yeah, I owe you a better question, closing this for the time being.