"Fixed a bug where apache child processes had the same random seed."
This may be related to an issue that Chris Ahrendt discussed last September (2015) in a post about "Issues with gateway quizzes generating with the same seeds".
We are using version 2.12, however, and seem to be having the same problems with a gateway quiz. For example, one student took the quiz six times. Here are the seeds for the four problems in each quiz
v1: 103 1153 1020 846
v2: 3409 921 4516 1888
v3: 2210 3531 1336 4344
v4: 2210 3531 1336 4344
v5: 4247 2260 3225 4579
v6: 2210 3531 1336 4344
As you can see, versions 2, 3, and 6 all used the exact same four seeds and thus the problems were identical in the three versions (although in different order). Versions 2 and 3 were taken about 27 minutes apart. Version 6 was taken about 40 minutes after version 4.
The suggestion last September was to add the line
PerlChildInitHandler "sub { srand }"
to the apache configuration file. Is this what the fix was for version 2.12? I looked at our apache2.conf file and do not see anything about PerlChildInitHandler in that file. Is there a difference between whether we are running apache or apache2? The WeBWorK configuration we are running is the one obtained from the MAA site by one of our network techs.
I have not yet added the suggested line to our apache2.conf file since the instructor of the course didn't want to make any changes before the due date of the quiz (so all students would take it under the same conditions, even if that condition caused some of them to get duplicate copies.) But should the fix in 2.12 have prevented this problem from occurring?
Thanks for any insight anyone might be able to offer.
Larry Riddle
Agnes Scott College