On 14/10/10 22:32, Tsantilas Christos wrote:
On 10/14/2010 02:35 AM, Amos Jeffries wrote:
On Wed, 13 Oct 2010 23:09:43 +0300, Tsantilas Christos
<chtsa...@users.sourceforge.net> wrote:
Hi all,
I am working on adding support for client-side bandwith limiting on
squid3. The related description of this feature is at
http://wiki.squid-cache.org/Features/ClientBandwidthLimit
The feature implemented by Alex Rousskov for his 3p1-rock branch. In
this mail I am including a first port for squid-trunk.
The patch needs more testing and probably farther development but I am
posting the current patch for preview and initial feedback.
A short description included in the patch.
Regards,
Christos
Can't see anything obvious in the code. :)
I know :-)
I am still trying to clarify some parts of the code.
Maybe there are comments for the configuration parameters, or other
comments about how good policy is to use the comm handlers
(commHandleWrite function) etc...
Can you at least make the new functions into members where possible. In
this case clientdbSetWriteLimiter into ClientInfo::setWriteLimiter.
OK.
In hunk @@ -3125,40 +3129,82 @@ it looks like "ClientDelayPools&
pools(Config.ClientDelay.pools);" could be reduced in scope to inside the
if().
+1 from me.
Whats the plan for committing this?
Currently I am running some basic tests using 2-3 clients..
Probably it will be ready next week.
Thanks. Okay then depending on Alex we may directly clash between the is
and the Comm::Io::Write polishing.
I'm unsure how the commHandleWrite function will fit into that layout.
There is a matching read delay system we will have to match it to
somehow. We currently have the same queue pattern in three places under
three names: AcceptLimiter for incoming connection acceptances, read
"Delays" for delay pools, and now write "Quota" for client pools.
Amos
--
Please be using
Current Stable Squid 2.7.STABLE9 or 3.1.8
Beta testers wanted for 3.2.0.2
