Unexpected Checking results
So over the course of the last days we have seen repeatedly unexpected checking results from students which already solved (parts) of the practical exercises.
This issues appears to be very racy and CANNOT be reliably reproduced by us. However, we have identified the potential culprit of the error, which stems from internal django db interactions. We have deployed a fix that hopefully addresses this issue by radically reducing the number of implicit queries done by django in our context processor feature. The fix is readily available in your Gitlab repo and we have already pulled those changes to your VMs (assuming that you did not have any merge conflicts e.g., by having conflicting version on the server vs in the git). We have seen cases of merge conflicts, which prevent this change from being automatically applied, please take care of resolving those conflicts and manually pull the fix on your VM instance.
We have opened a Askbot thread in which we want to track further flaky behavior, so if you notice flaky behavior without you changing anything on the server at all, please ping us in the thread: https://cms.cispa.saarland/askbot/websec2021/question/30/crawler-flakyness/.
Even though the flakyness does not hinder you exam admission, we will still want to help you compete with your peers for the top score in the scoreboard without randomness.