Hi,

I'm slightly puzzled over the date warnings as they all seem to be
correct. As the date changes appear to be different based on their
location is it reasonable to assume  the varnish reverse proxy doing
something to alter the results? if so how is the time warning
calculated as different to that shown in the header response?

to replicate

Click http://redbot.org/?uri=http://creativecommons.org/about/licenses

redbot view headers response  (edited)
    Date: Sat, 10 Apr 2010 13:23:55 GMT
    X-Varnish: 1977484888
    Age: 0
    Via: 1.1 varnish

no warning, date is accurate

Click view assets

redbot response
The server's clock is 7 sec ahead.
The server's clock is 20 sec ahead.
The server's clock is 1 day 1 hr ahead.
The server's clock is 6 hr 33 min ahead.

multiple warnings on the image files so
click the 4th image link
    Date: Sat, 10 Apr 2010 13:37:07 GMT
    X-Varnish: 1977584744 1967571076
    Age: 92796
    Via: 1.1 varnish
    Connection: keep-alive

The server's clock is 1 day 1 hr ahead.

click the last image link
http://redbot.org/?uri=http://i.creativecommons.org/l/by/3.0/88x31.png

redbot view headers response  (edited)
    Date: Sat, 10 Apr 2010 13:27:56 GMT
    X-Varnish: 2112650971 2109301108
    Age: 23729
    Via: 1.1 varnish
    Connection: keep-alive

redbot response (edited)
The server's clock is 6 hr 35 min ahead.

one warning, date is accurate

Regards,
Mark

Hi,

Varnish is sending an Age header and also re-writing the Date header to be the current time, instead of leaving it as the original time the message was created (as HTTP requires). As such, from the outside, it looks like there's clock skew happening.

This is a problem when using some reverse proxies; a good paper about the effects of doing so is:
  http://www2.research.att.com/~edith/Papers/HTML/usits01/index.html

I need to check for this condition in RED and highlight it specifically, rather than just identifying it as clock skew.

Cheers,

On 10/04/2010, at 11:41 PM, marked wrote:

--
Mark Nottingham     http://www.mnot.net/

Hi

On 11 April 2010 00:45, Mark Nottingham <m...@mnot.net> wrote:

OK, I thought that it was something like that, but couldn't entirely work it
out without also seeing the config of creativecommons varnish. I am curious
as to the difference in clock skew based on images between different
locations, and not the html content.

I'll have a read of this later. Thanks

Cheers,
Mark

2nd para "cause it inefficiency" should be "cause inefficiency"

On 11 April 2010 04:34, Mark Nottingham <m...@mnot.net> wrote:

Thanks, I've reported it to creativecommons as a bug.

Regards,
Mark

Fixed; thanks.

On 11/04/2010, at 6:30 PM, Mark Dickman wrote:

--
Mark Nottingham     http://www.mnot.net/