Re: Some changes to curl-loader

[Val S. <va...@nv...>]

Hi Robert,

   Thank you for accepting this.  I've downloaded svn.566 and I am running both fixed and non-fixed rate.  So far no problem.

I am not sure what  you mean by the "riskiness" of the fixed rate.  The worst thing I saw happening is that the fixed rate is not maintained.  This happened to me in two cases:
1. Very slow server response to opening a connection (SYN). Seems libcurl does not return until the connection is established.
2.  Very slow server response to HTTP requests. Here the number of available curl-loader clients gets exhausted and the rate falls to the level they can provide, my code complains about the absence of free clients.

In both of those cases the rate will be low in the non-fixed rate mode as well, although that mode will adapt to the load better (will not complain :-)).  But of course, it all depends on what load you really want to model.

Best regards,
/Val


>
>From: Robert Iakobashvili <cor...@gm...>
>To: curl-loader-devel <cur...@li...>
>Sent: Thursday, August 20, 2009 11:33:54 PM
>Subject: Re: Some changes to curl-loader
>
>>Hi Val,
>
>
>On Fri, Aug 21, 2009 at 1:12 AM, Val Shkolnikov <va...@nv...> wrote:
>
>>>Hi, Robert,
>>>>   Well, I am sorry to say, but the README file does contain all the
>>stuff that the FAQ does, do not know how I missed it.  I've updated it
>>for both versions and I am attaching here the new patches that contain
>>those changes.  Sorry for the inconvenience.
>>>>/Val
>Thanks, applied with some minor changes.
>Added you to our THANKS list.
>
>Could you, please, take the version from svn and test it a bit? 
>Thanks.
>Fixed-rate load is a risky feature, therefore, please, try to test
>>without fixed-rate and see, if it does not break something.
>
>Best wishes!
>-- 
>Truly,
>Robert Iakobashvili, Ph.D.
>
>