On 28/03/2012 6:41 p.m., Colin Coe wrote:
Hi all
I'm running squid 3.1.10 on a RHEL6.2 box. When I point my clients at
it, the clients experience about a 2 minute 20 second delay between
sending the request to squid and the request being fulfilled.
Might be client problems with "Expect:100-continue" HTTP/1.1-only
feature being sent to HTTP/1.0 software?
That is still happening fairly frequently. We offer admin the choice
of ignore_expect_100 which changes the HTTP/1.0 proxy from sending 417
(which tells the client it can retry immediately without Expect:) to
ignoring it and waiting for the full request to arrive (to which the
client has no choice but wait some timeout then send the POST data anyway).
Guess which one can result in several minutes delay?
If you want to be sure you will need to grab a packet trace for the HTTP
headers between the cleint and proxy.
Amos
