none
IE sporadically aborting images being downloaded over SSL RRS feed

  • Question

  • For an out-of-the-box application deployed to our web servers, a few images are sporadically not being downloaded and the broken image (X) is displayed.  The images that fail to download will vary from test to test but are typically restricted to the images in the left navigation menu bar.  There are 5 navigation images and usually 1-3 will be broken for any given request to the page.  The images are *.png and size ranges from 6000-8000 bytes.  There are several other images on the page but they never seem to be an issue. 

    This problem only occurs using HTTPS and IE (tried versions 8, 9, 10, 11).  IE works fine over HTTP and the web application works fine in Chrome and Firefox (both HTTP and HTTPS).  This affects all users (small user base at this time).  Most users are running Windows 7 but this also fails when IE client is run from Windows Server 2008 R2.

    In working with the web application vendor they were not able to replicate the issue in their local test environment.  This problem occurs in all of our environments (test, uat, prod).  Another out-of-the-box application provided by same vendor deployed to same web servers does not have the issue (but doesn’t have as many images).

    While working the web application vendor, we were looking at a Wireshark trace that showed the client (IE) was cancelling the request for image(s) just milliseconds after the initial request was made for no apparent reason.

    We have tried numerous IE setting changes (including IE Reset Settings).

    Any suggestions on what could be causing this issue or how to further troubleshoot the issue to identify the problem?

    Thanks in advance for any help you can provide.

    Tuesday, July 15, 2014 1:28 PM

Answers

All replies

  • While working the web application vendor, we were looking at a Wireshark trace that showed the client (IE) was cancelling the request for image(s) just milliseconds after the initial request was made for no apparent reason.

    Is it a DNS lookup which can be cached?  The most common reason for such a cancellation seems to be a lookup failure but methods for extending the timeout for that process seem either non-existent or very obscure.  E.g. I have seen some which apparently can be applied to a server, so if you control the client's DNS server you might be able to change your symptom that way; otherwise I don't know of any way that the client can change this (e.g. seems to be hardcoded).  As a test you could try putting the necessary lookups in the client machine's HOSTS file.  Then they get cached by dnscache, effectively permanently, e.g. as seen by  ipconfig  /displaydns.

    In any case, use the Developer Tools, Network capture tool to test this hypothesis, since of course you could then see something else happening which would contradict it or provide more compelling evidence for a better one.

     

    Good luck



    Robert Aldwinckle
    ---

    Wednesday, July 16, 2014 1:41 PM
    Answerer
  • Robert,

    Thank you for your suggestion.  I did see other posts with similar symptoms that was DNS related.  So I did try to update the client machine's hosts file with the web server information but I got the same results. :(

    Tuesday, July 22, 2014 9:15 PM
  • Gotta open that port 443 in your external-facing firewall.
    Friday, July 25, 2014 10:39 AM