hc-httpclient-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Oleg Kalnichevski <ol...@apache.org>
Subject Re: HttpConnection.flushRequestOutputStream() sometimes lasts too long
Date Sat, 25 Oct 2008 12:35:55 GMT
On Thu, 2008-10-23 at 18:43 +0200, Furmaniak Christophe wrote:
> Hi,
> 
> we've developped a web application performance tool. Our product intensivly uses httpclient
(currently 3.1) to measure response time of these web applications (single url or sequence
of url to copy a human being navigation).
> 
> The main principle of our tool is to perform the same requests at a fixed frequency (5
minutes most of the times) to be able to compute Service Level Agreements.
> 
> We've detected some "strange" behaviours on some monitors (=sequence of url request retreived
with httpclient). Sometimes (1% of the executions), some requests take more time (like 3 sec
instead of 100ms) to be performed.
> 
> We cannot say for sure for the moment that the network elements of the architecture are
not concerned but let's assume first that our problems are not network related (and I'm not
saying that the guilty part is httpclient ;], if you can help me to find the proof that it's
network related, let's go!)
> 
> Logs have been added on the web server side (apache) using a %D  in the LogFormat specification.
The response times measured on apache side don't match the ones we get using httpclient, not
even closer (we see 3 seconds on httpclient side, we see less that 1 ms on apache side).
> 
> We've traced the httpclient call and we've been able to detect 2 different behaviours:
> 
> the first one occurs during the connection opening, inside HttpConnection.open() call.
I've added the "OPENED" trace that comes just after the isOpen=true.
> 
> =============================
>                   isOpen = true;
>                   if (LOG.isDebugEnabled()) {
>                         LOG.debug("connection to " + host + ":" + port + " OPENED");
>                   }
> =============================
> 
> our logs:
> ==============================
> [DEBUG] [22/10/08 12:38:26.250] [httpclient.HttpConnection]<7747> Open connection
to 160.92.110.240:443 (HttpConnection.java:717)
> [DEBUG] [22/10/08 12:38:29.246] [httpclient.HttpConnection]<7747> connection to
160.92.110.240:443OPENED (HttpConnection.java:771)
> ==============================
> 
> I'm aware that there can be a lot of reason why the HttpConnection.open() lasts so long,
but does it exist explanation NOT related to network only issues?
> 
> 
> 
> the second one occurs during the flush of the request OutputStream
> 
> ==============================
> [DEBUG] [22/10/08 14:00:26.234] [httpclient.HttpConnection]<16536> enter HttpConnection.flushRequestOutputStream()
(HttpConnection.java:851)
> [DEBUG] [22/10/08 14:00:29.240] [httpclient.HttpConnection]<16536> leaving HttpConnection.flushRequestOutputStream()
(HttpConnection.java:854)
> ==============================
> 
> my modifications are:
> ==============================
>       public void flushRequestOutputStream() throws IOException {
>             LOG.trace("enter HttpConnection.flushRequestOutputStream()");
>             assertOpen();
>             outputStream.flush();
>             LOG.trace("leaving HttpConnection.flushRequestOutputStream()");
>       }
> ==============================
> 
> what can explain such a long time to flush the request OutputStream?
> Am I wrong or does this output stream only "contains" the request built by httpclient?
That would mean that the web server is taking a (too) long time to accept the request?
> 
> 
> some more details: we have these behaviours using different hosts for our performance
tool, so we can assume that these response times are not related to the load of the host that
is running the httpclient.
> 

Christophe,

HttpClient has no control over TCP/IP connections other than
java.net.Socket API provided by JRE. The problem can be related to the
particular JVM implementation, OS, network configuration, or all
combined.

Oleg   


> Thanks for reading :)
> 
> Christophe
> 
> ________________________________
> 
> Ce message et les pi?ces jointes sont confidentiels et r?serv?s ? l'usage exclusif de
ses destinataires. Il peut ?galement ?tre prot?g? par le secret professionnel. Si vous recevez
ce message par erreur, merci d'en avertir imm?diatement l'exp?diteur et de le d?truire. L'int?grit?
du message ne pouvant ?tre assur?e sur Internet, la responsabilit? du groupe Atos Origin ne
pourra ?tre recherch?e quant au contenu de ce message. Bien que les meilleurs efforts soient
faits pour maintenir cette transmission exempte de tout virus, l'exp?diteur ne donne aucune
garantie ? cet ?gard et sa responsabilit? ne saurait ?tre recherch?e pour tout dommage r?sultant
d'un virus transmis.
> 
> This e-mail and the documents attached are confidential and intended solely for the addressee;
it may also be privileged. If you receive this e-mail in error, please notify the sender immediately
and destroy it. As its integrity cannot be secured on the Internet, the Atos Origin group
liability cannot be triggered for the message content. Although the sender endeavours to maintain
a computer virus-free network, the sender does not warrant that this transmission is virus-free
and will not be liable for any damages resulting from any virus transmitted.


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


Mime
View raw message