Recently migrated west wind app from Windows 2012 to Windows 2019. No problems reported during test phase, but once migration went live, getting daily consistent complaints that site somewhat at random displays a white page instead of proper content largely when trying to edit an order, though issue happens for other pages. I'm told the user can try again and the problem often clears up on a subsequent try. It's not clear if they hit backbutton and click link to make it work 2nd time, or simply F5-refresh on the white page. The issue does not seem limited to any particular user. From some quick math the issue occurs about 7-8% of the time each day.

Wondering if anyone has run into this situation before?

I updated IIS logging to capture response bytes (sc_bytes), and can easily see which requests are failing as the response is 145 bytes, instead of the correct ~66K bytes. The most obvious clue is "garbage" appended to the query string in cases where the response fails.

HOWEVER, you can see that the WWWC app receives the request w/o any garbage in the query string. Further more, I've manually pasted the garbage captured in the IIS Log onto the URL in the browser, and the WWWC app handled it just fine, full ~66K response / no white page.

At a loss to explain what is causing the problem and no idea how to resolve issue. Customer is understandably very upset.

Any advice, ideas greatly appreciated.

PS - Running older version of WWWC so the feature to log response data from the app does not exist. May hack the older wwwc code to add some rudimentary response logging, but haven't been given clearance to do so by customer, since it's not as simple as just enabling the feature if this had been written using latest WWWC version.

Thank you -Steve

That looks like serious memory corruption. How are you running your Web Connection server instances?

One thing that I know can cause this is if servers are not registered correctly as Out of Process components. Make sure the COM Servers are set for Single Use Out of Process and not to Multi-Use or worse running as InProc servers.

Most likely this is the ISAPI extension? What version of Web Connection? I highly recommend not using the ISAPI extension and switching to the .NET module. The ISAPI module has been deprecated for nearly 10 years now and so doesn't get tested anymore. I doubt the issue is something specific to the new server OS - more likely some difference in configuration that might be causing this corruption. Switching to the .NET Module should be of minimal effort - it's just a matter of switching the script maps.

The .NET Module can't have that kind of memory corruption as the runtime integrity checked rather than the old C++ code which had the potential for this sort of thing.

+++ Rick ---

Thank you Rick for your helpful responses.

I haven't replied in awhile because the problem just "magically" disappeared shortly after your posting. The "IT" guys won't comment on having made any changes to the server, but other issues with a desktop application also seem to have suddenly started working, again by "magic".

Of course, it is possible that it's some random issue like failing hardware or something and perhaps IT really didn't do anything. Who knows.

The logs are now clean and users no longer report any issues.

If it comes back, I will try out your suggestion to use the .NET Module.

Appreciate your feedback -Steve