After change to http.sys, errors on production servers after a few days of operation

[magosk]

Hi Alexandre and thanks for the feedback! I will look deeper into the things you mention. My suspicions regarding our main web application has strengthened. Besides the memory usage pattern I saw, by checking all times this service was restarted (e.g. due to a patch being applied) and putting it on the same timeline as occurrence of errors and server restarts, I can see that when we have had longer periods of operation without errors, this coincides with having at least one restart of the main web application service in this time interval. So it might be that some resources are returned to the system when the process is terminated (but also that it might be too late if errors have started to occur in the system). Actually, I have already started memory and resource leak analysis (using CodeWatch from the Nexus Quality Suite). Although I see some that looks to be purely IW related, I am focusing on leaks in our own code to begin with. I have found and fixed a few cases so far, but all of those should have been present also before the http.sys upgrade (and thus likely not the cause of the problems we had).

I will post an update here when I have more information.