On Thu, Sep 8, 2011 at 8:59 PM, Amos Jeffries wrote:
On 09/09/11 07:02, alexus wrote:

Is there a way to analyze somehow deeper what's going on with this?

tss# grep -c 'http://ecs.amazonaws.com/onca/xml?' access.log
3065
tss# grep 'http://ecs.amazonaws.com/onca/xml?' access.log | tail -1
66.55.138.70 - - [08/Sep/2011:18:59:26 +0000] "GET
http://ecs.amazonaws.com/onca/xml? HTTP/1.1" 200 135861 "-"
"Mozilla/4.1" TCP_MISS:DIRECT
tss#

I'd like to capture it somehow so I can look what kind of request is that.


Data retrieval from an amazon online API. You have stripped the query
parameters from the logged information so there is no way to tell how many
different requests are being bunched together in that 3065 count.

The above line count has about the same meaning as:
  grep -c 'http://ecs.amazonaws.com/' access.log

(and probably a similar count.)


On 10/09/11 02:15, alexus wrote:
> Sorry I wasn't clear ... All lines are the same just different time stamp
>

The log lines are all cropped down to the "?".

Fetching that URL produces a message far shorter than 135KB:

"<?xml version="1.0" encoding="UTF-8"?><Errors><Error><Code>AWS.MissingServiceParameter</Code><Message>Your request is missing the Service parameter. Please add the Service parameter to your request and retry.</Message></Error></Errors>
"

You need the full original URL to do much useful analysis about what the HTTP details for a particular object are.

FWIW: the message above comes back with the header "Vary: Accept-Encoding,User-Agent".

User-Agent is pretty much a free-form field these days where browsers and other tools get to put any text they like. There are over 14 million unique User-Agent strings known today with hundreds being added to the databases every month. So that Vary: header could be essentially forcing MISS on almost every request.

Beyond that can't tell from the given incomplete URL.

Amos
--
Please be using
  Current Stable Squid 2.7.STABLE9 or 3.1.15
  Beta testers wanted for 3.2.0.11

Reply via email to