Lock Release Timing

GregTomkins

Active Member
v10.1c, WebSpeed, no BIW, APW maybe/maybe not, WS/DB are on separate machines.

Process A and B both contain a single line of code, 'FIND FIRST some_table EXCLUSIVE-LOCK'.


  1. Process A gets the lock.
  2. Process B runs before A ends, thus ,it blocks.
  3. Process A ends normally (and thus releases its lock).

Here is the question: once A gives up the lock, how long should it take for Process B to unblock?

Assume the machines are fast, and there are no other processes in the WS pipeline (eg. connect procedures).

My experience has normally been "instantly", but today, I am seeing "exactly 2,000 ms", and I am wondering what configuration / versions etc. could explain this.
 
You're welcome - the exact 2 second wait triggered a memory somewhere, so I knew it had to be there... ;-)
 
v10.1c, WebSpeed, no BIW, APW maybe/maybe not, WS/DB are on separate machines.

Forgive my question being completely off-topic, but I'm just curious. Given your APW comment I assume you are running Enterprise RDBMS. Why no BIW, and "maybe" no APWs? Is this database read-only?
 
Forgive my question being completely off-topic, but I'm just curious. Given your APW comment I assume you are running Enterprise RDBMS. Why no BIW, and "maybe" no APWs? Is this database read-only?

I second that question.

Especially as Mr. Tomkins is an otherwise generally sensible person with a track-record of rational behavior. We might need to engage a therapist or something...
 
There is no doubt I could use some therapy. But not over APW's and BIW's.

In production databases, we use them like they are going out of style. I was testing this behaviour on a development system that had neither, and a demo system that has APW's only.

Say, I am flabbergasted by this whole 2-second thing; partly because it happens at all, and partly because we have been running remote AppServers for many years now and NOBODY here knew this. But Stefan did!!
 
Back
Top