On 28/01/11 07:17, Ming Fu wrote:
Hi Amos,
Thanks for the insight.
Do you remember the bug number? I want to understand the issue especially when unencrypted traffic can be sent.
Ming
It is a bit twisted but involved with this design flaw bug:
http://bugs.squid-cache.org/show_bug.cgi?id=2117
Henriks comment #3 and #4 covers the problem.
In effect Squid handles the CONNECT as a special case whether tunnelling
or bumping. The results is that for a bump the re-CONNECT is done for
the internal request, not a reversion to the original CONNECT. So any
special headers on that first CONNECT may be lost by the bump.
Amos
--
Please be using
Current Stable Squid 2.7.STABLE9 or 3.1.10
Beta testers wanted for 3.2.0.4
