On 3/8/2012 5:12 PM, René Berber wrote:
> On 3/7/2012 4:42 PM, René Berber wrote:
>
>>>> I understand, from reading other posts, that the CURLOPT_TIMEOUT which
>>>> in this case is set to 30 sec is being divided by 8, so the real
>>>> time-out used is 3.75 sec.
>>>
>>> This particular problem was addressed with commit f4d3c0cbfb648917a0 but
>>> it is not yet present in any release. You could try it out though,
>>> either by cherry-picking the commit or by running a recent daily
>>> snapshot or git build.
>>
>> I'll test the snapshot.
>
> After 1 day testing the result is not different than before. It looks
> like I was wrong, its not a timeout problem, but something else.
>
> I'm seeing those CURLE_FAILED_INIT with varying times from start of the
> operation, most errors come at 1 sec, which is strange, but some go as
> much as 8 sec, still much less than it takes if I repeat the operation
> that failed using curl.
>
> Another clue is that its not always the same host with the 8 IP
> addresses, sometimes the same error happens with a host that only has 1
> or 2 IPs, but most (10 : 1) of the problems are with the former.

Found the problem, after a couple of hours testing a new build I can say
it was my own mistake.

I had built the library with the option --enable-threaded-resolver and
that was the cause of all those CURLE_FAILED_INIT, and sockets leaked in
CLOSE_WAIT state.

With a new build that enabled c-ares, the program is working perfectly.

-- 
René Berber
-------------------------------------------------------------------
List admin: http://cool.haxx.se/list/listinfo/curl-library
Etiquette:  http://curl.haxx.se/mail/etiquette.html