On 28 Sep 2010, at 18:54, Aparajita Fishman wrote:
Then do one more test with no Active4D code at all, just HTML, so we can isolate the problem even further.
I've made a bit of progress with this issue in the last few weeks despite it being just about the most elusive problem I've ever tried to pin down.
If a component is in use, there appears to be a serviceability ceiling of about 1.2 to 1.5 million requests for the Active4D web server running in 4D, at which point any pointers passed to the component cannot be resolved. This applies equally to executable (.a4d) and non- executable (.html) pages (i.e. pure html as you suggested).
I've had a machine dedicated to investigating the problem for about 6 weeks which has now run about 200-300 hours of tests. The phenomenon is still fairly ambiguous but hopefully enough info has become conclusive to assist in determining what the low level issue is. Some things are very conclusive, namely the consistency with which failure occurs and the mode of failure.
What's not so clear or consistent is the number of iterations to fail and the low level nature of the fault - i.e. which subsystem, request handler, plugin, 4D etc. I've tested all independently with negative results. Unfortunately it appears to be the combination which consistently fails.
Full details are in the readme file and my spare machine remains available to run any further tests required which have now become it's daily hobby :)
Best Regards Peter _______________________________________________ Active4D-dev mailing list [email protected] http://mailman.aparajitaworld.com/mailman/listinfo/active4d-dev Archives: http://mailman.aparajitaworld.com/archive/active4d-dev/
