Thanks for your feedback.
Terence Kelly wrote:
> I like what you've done so far. The ability to log all
> request headers and all reply headers is a very important
> enhancement. In addition, it would be very useful if Squid
> could optionally log the following:
That we do today, if requested, and also covered by the custom format
specification suggestion.
> 1) microsecond-resolution timestamps
You need higher resolution than there already is? (1/1000 sec)
Entirely possible to do. Internal measurements are down to 1/1000000 (or
as close as possible, given the granulatiry of the OS clock).
> 2) a digest (MD5, SHA1, or similar) of message bodies
> (both request and reply bodies)
Not planned unless someone volonteers in implementing the sums..
> 3) entire requests and replies, including all headers
> and bodies, for a) the request received from the
> client by Squid, b) the request (if any) issued
> by Squid to the origin server, c) the origin
> server's reply (if any), and d) the reply sent
> to the client by Squid
As above, and not very practical to do as part of the access log. Data
logging needs to be done on-the-fly.
> One problem with your existing documentation is that it's
> not clear what is meant by "Request header": Are you
> talking about the request *received* by Squid from a client,
> or the request *sent* by Squid to an origin server? It
> should be possible to log both! The same goes for "Reply
> header". At the very least, make it clear *whose* request
> and reply headers you're talking about.
Squid access.log covers requests as received by Squid and their
responses, with some minimal exceptions if you choose to mangle the
request (i.e. redirectors may change the request URL).
> The basic principle is simple: If verbose proxy logging is
> enabled, it should be possible to recover from the proxy log
> all information available to the proxy during its operation.
Which can be done, but most likely will only get done if one interested
in getting these logs makes sure it gets done. To the general Squid user
most of your suggestions is of very limited use.
Please note that we are only tinkering on the specification. The job of
getting it implemented is not yet assigned, and neither is a timeframe
on when it should be completed.
Regards
Henrik Nordström
Squid Hacker
Received on Mon Nov 19 2001 - 18:43:50 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:14:38 MST