Mailing-List: contact users-help@httpd.apache.org; run by ezmlm
Precedence: bulk
Reply-To: users@httpd.apache.org
Received-SPF: pass (athena.apache.org: local policy)
Message-ID: <552D64D1.1070709@epimorphics.com>
Date: Tue, 14 Apr 2015 20:04:49 +0100
From: Brian McBride <brian@epimorphics.com>
User-Agent: Mozilla/5.0 (X11; Linux x86_64;
 rv:31.0) Gecko/20100101 Thunderbird/31.5.0
MIME-Version: 1.0
To: users@httpd.apache.org
References: <551D55A2.7030309@epimorphics.com>
 <552BC4A3.5010301@epimorphics.com>
 <1F08C88D-357D-470B-BD90-3786AB13C617@jaguNET.com>
In-Reply-To: <1F08C88D-357D-470B-BD90-3786AB13C617@jaguNET.com>
Content-Type: text/plain; charset=windows-1252; format=flowed
Content-Transfer-Encoding: 7bit
Subject: Re: [users@httpd] mod_reqtimeout question:

Hi Jim,

Thanks for responding.

Short answer: yes - the client is using http keep alive to keep a 
persist connection open to the server.

Brian

On 14/04/15 19:38, Jim Jagielski wrote:
> By persistent, do you mean one that was held open via
> standard HTTP keepalive?
>
> Thx
>> On Apr 13, 2015, at 9:29 AM, Brian McBride <brian@epimorphics.com> wrote:
>>
>> Not having had a response to this question I'm left wondering whether I've screwed up maybe by not asking a clear question or asking in the wrong place.  Or maybe I just asked it at a bad time or I'm just out of luck and no one is motivated to respond.
>>
>> Brian
>>
>>
>> On 02/04/15 15:43, Brian McBride wrote:
>>> Given an apache 2.4 server with:
>>>
>>>   ReadRequestTimeout header=60-65,rate=500
>>>
>>> When a client opens a persistent connection to that server, does the request timeout apply to
>>>
>>>   a) each request sent down the connection individually, or
>>>   b) the whole connection, i.e. the connection will close after 65 seconds of reading requests even if there
>>>       are more requests ready to send
>>>
>>> Background:
>>>
>>> I have two apache 2.4 servers, one acting as a reverse proxy for the other.  Both have mod_reqtimeout enabled.
>>>
>>> The reverse proxy is using persistent connections to talk to the backend.
>>>
>>> I have applied a steady test load of 1 request per second; requests are quick to process in the backend
>>>
>>> I'm getting 502 errors from the front end.  It appears that the backend is closing the connections, though its hard to be sure which end is.
>>>
>>> The rate of 502 errors reduces the longer I make the ReadRequestTimeout.  I get no errors if I disable mod_reqtimeout. It looks like the backend is closing then connections after about N seconds where N is the timeout value, even in the presence of a steady load.
>>>
>>> MaxKeepAliveRequests and KeepAliveTimeout are set very high.
>>>
>>> I'm pretty new to this sort of thing.  I'd appreciate any thoughts or guidance.
>>>
>>> Brian
>>>
>>>
>>>
>>>
>>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
>> For additional commands, e-mail: users-help@httpd.apache.org
>>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
> For additional commands, e-mail: users-help@httpd.apache.org
>


---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org