CRYPTINIT error while opening an encrypted database should not leave ipcs (ftok semaphore)
Final Release Note
NO RELEASE NOTE NEEDED. Tagged in [#927 (closed)]
Description
See YDBTest!946 (comment 333219963) for background on the issue. In that comment, I had mentioned that distinguishing the 1 user vs 32Ki users case is not straightforward. But thinking more about this, I realize we could distinguish them if we know we created the ftok semaphore just now. In that case, it has to be the 1 user case and so we should be able to delete the ftok semaphore. Therefore, this should be reasonably straightforward to implement.
Draft Release Note
Edited by K.S. Bhaskar