Mailing-List: contact httpclient-users-help@hc.apache.org; run by ezmlm
Precedence: bulk
Reply-To: "HttpClient User Discussion" <httpclient-users@hc.apache.org>
Received-SPF: pass (nike.apache.org: domain of bratislav1983@gmail.com
 designates 209.85.213.174 as permitted sender)
MIME-Version: 1.0
In-Reply-To: <5314C3AF.3070408@elyograg.org>
References: 
 <CAAWvMYOf_sZ5i9hyf8E6V32RG0+1R1_bFc6CD4zMMK5vx8CH8Q@mail.gmail.com>
	<CAAWvMYP4KkoaK8jvjSPT2JXgnOq-t42XZb27Uj1tsStOWx43yw@mail.gmail.com>
	<5314C3AF.3070408@elyograg.org>
Date: Mon, 3 Mar 2014 19:36:44 +0100
Message-ID: 
 <CAAWvMYPqg03_1T1R84mxWGfc=otKzJX-Nc6xmKawErRp5n6TdA@mail.gmail.com>
Subject: Re: Fwd: Connection timeout doesn't behave the same on windows and
 linux
From: Bratislav Stojanovic <bratislav1983@gmail.com>
To: HttpClient User Discussion <httpclient-users@hc.apache.org>
Content-Type: multipart/alternative; boundary=001a1134428ec7800d04f3b811bb

--001a1134428ec7800d04f3b811bb
Content-Type: text/plain; charset=ISO-8859-1

Mmmm...good point. I seems to me now this is more architectural problem in
my software, not a httpclient issue.

What I'm trying to achieve here is basically I need to fire a db query (via
REST) during startup of my webapp. DB runs
in a separate process and needs around 10 secs to start being responsive.
If I fire this query too early, I get connection timeout (on windows).
If I fire query at the time when db is just about to finish initialization,
my request gets accepted, but db returns 404. If I fire the same query when
db finished its initialization procedure, I get of course a proper response.

So I figured, I should use retry handler to poke db every second until I
get 200, but that's obviously a bad approach, since db can bring up REST
interface, but the other parts still haven't completed booting. It's
working fine on Windows since it runs on a physical machine,
but on Linux it simply skips timeout, since it's a vmware vm (and has
limited hw resources).

I temporarily solved this by manually adding configurable Thread.sleep
timeout so db can pick up. Not sure if there's a better solution.

Thanx for your input.

P.S. I'm using eXist XML database (exist-db.org)

On Mon, Mar 3, 2014 at 7:02 PM, Shawn Heisey <solr@elyograg.org> wrote:

> On 3/3/2014 9:00 AM, Bratislav Stojanovic wrote:
>
>> It seems that connection timeout parameter doesn't work on Linux Mint 16
>> (which is
>> basically Ubuntu). *Code runs perfectly fine on Windows.*Here's my setup :
>>
>
> <snip>
>
>  *.setConnectTimeout(1000)*
>>
>
> <snip>
>
>  *The problem : when I run this code on linux, there's no 1 sec delay
>> between consecutive attempts. It just
>> *
>> *goes over 10 attempts and finished within ~200ms. Why?
>>
>> *
>> *Am I doing something wrong or this is a bug? Please help.
>> *
>> *
>> P.S. I'm using http client 4.3.2.*
>>
>
> I assume that the service it's trying to connect to is down or in some way
> unavailable.  What is the exact status?  If the server is reachable on that
> port but the http server is not running, a connection attempt will be
> instantly refused, there's no reason to wait for the timeout period.
>
> In order for a full second to pass, the port must be completely
> unreachable, which means the server either needs to be down or it needs to
> be firewalled in a way that offers no response at all rather than a
> refusal.  Is this the case with the server that the Windows version is
> trying to contact?
>
> Thanks,
> Shawn
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: httpclient-users-unsubscribe@hc.apache.org
> For additional commands, e-mail: httpclient-users-help@hc.apache.org
>
>


-- 
Bratislav Stojanovic, M.Sc.

--001a1134428ec7800d04f3b811bb--