Troubleshooting the "Unable to decode" (IdP 3.3)
Brent Putman
putmanb at georgetown.edu
Fri Aug 25 14:10:44 EDT 2017
On 8/25/17 1:20 PM, O'Dowd, Josh wrote:
>
> Actually, there is a SAMLRequest parameter that I missed, coming after the RelayState parameter, and now I see that only some of the failed requests have a leading RelayState parameter.
> Sample access entry for failed request:
> [25/Aug/2017:09:24:05 -0600] TLSv1.2 ECDHE-RSA-AES256-GCM-SHA384 "GET /idp/profile/SAML2/Redirect/SLO?RelayState=ss:m
> em:18e8f777deec8579d5adfb48fbaf6d9ae296df30c777408c7058488b5509bd70&SAMLRequest=rVbJcqNIEL3rKxzqo8JmEwIUtiNYtLFoAyTBZaIExSKxiSoE6OsHd
> dvT7e6ZCffEXLMyX+Z7mUnyjECaFGM9D/MKb+Glggg/NGmSofHXl5d+VWbjHKAYjTOQQjTG3tgUDX1MP5Hjosxx7uVJ/0Hp4uIM4DjPXvoRxgUaE0SSh3H2VKX4CfoVEfsF0Q
> ...
> ...
> 83szyP7Zvz5rr2fWBOi++VbZD5sXv+gu9+EI8OwggcBF7AM4I40oASK5T0B+EHwfjE/RL0bP/wBvf4J&SigAlg=http://www.w3.org/2000/09/xmldsig HTTP/1.1" 40
> 0 18603
With the full request URL, specifically the SAMLRequest param (which is
the actual SAML protocol message), you can try decoding that with one
of the online SAML decoder/debugger services. I have used in the past:
https://rnd.feide.no/software/saml_2_0_debugger/
Also just found some others via googling, such as:
https://www.samltool.com/decode.php
I'll be a little surprised if those can decode them though, and
OpenSAML can't, at least for valid inputs. The OpenSAML code has
essentially been in use for over a decade, and is not known to have any
issues with valid requests.
Since this is the Redirect binding, what I would initially suspect is
that they aren't encoded properly, wrt to the Deflate part, i.e. maybe
it's just Base64-encoded and not Deflated. That's not valid Redirect
binding and OpenSAML would not handle and would treat as an error. A
decoder tool that attempts to dynamically support Redirect and POST
input simultaneously might be less strict. (I think the FEIDE one does
that). You could also try just handing the SAMLRequest value to a
Base64 decoder and see what happens. If it works, then they aren't
deflating.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://shibboleth.net/pipermail/users/attachments/20170825/9b5dd2cb/attachment.html>
More information about the users
mailing list