On Mar 15, 2004, at 3:42 PM, Aparajita Fishman wrote:
My instincts tell me that it has something to do either with triggers or with some other stored procedure on the server. I have a client in Boston who was having hangs of the server and it turned out to be a locked record problem that caused a deadlock in a trigger.
I looked through my triggers and couldn't find anything that would cause this. I did find a couple of triggers where I didn't formally declare $0, but since this is an implicit parameter, I doubt that caused it. (Could it?) I tweaked a few memory settings as recommended by Steven Willis, but again the nature of the problem does not point to either of these.
Remember, there were some crashes when NO ONE was logged it and using the system (web or 4D Client). Some of these crashes were within a couple of minutes of each other. Then others were a day or two later. Bart Alcorn reported the same symptoms, though it's entirely possible that we've both made similar mistakes in a trigger. I agree that triggers can be tricky, but other than the $0 parameter, I haven't made many changes to these triggers. They worked fine in 6.7.
Steve and I suspect that it's a housekeeping task of some sort. When we discovered 4D Client reusing an Active4D process, this further validated our guess. Perhaps now that the Active4D housekeeping task is on the server and the actual web process are on the client, this provided the level of separation we needed. (???)
What about the crash logs from the server? You never sent them to me.
Yes, the crash log we sent you WAS the server log. It hasn't crashed since we separated the two, so there's no need to send you two logs. If you never got it, your mail must have been down or something. Steve sent it to you first, then I did later. The one I sent you was on March 11, at 10:58 AM CST (to your personal email address).
Thanks again, Doug
