You are viewing a plain text version of this content. The canonical link for it is here.
Posted to httpclient-users@hc.apache.org by "Domet De Mont, Dominique (d3m)" <do...@hp.com> on 2008/12/10 14:33:33 UTC
Httpclient 3.1 encodes null=null parameters under stressing
conditions
Dear List,
I am facing a strange httpclient 3.1 behaviour under stressing conditions while encoding a post: the set of parameters is "nullified", whereas the number of parameters is correct.
I have turned on the httpclient wire traces to get more clues, and added in my own code a dump of the PostMethod object before and after the httpclient.execute call: I see the actual parameters correctly set in the PostMethod object, but actually the httpclient encoding "nullified" parameters.
Here is the code excerpt:
synchronized(this) {
// Check parameters before call
log.debug("About to call httpclient.executeMethod on object: " + postMethodM + " containing parameters: " );
for(int i=0; i < postMethodM.getParameters().length; i++)
log.debug("\t" + i + "/ " + postMethodM.getParameters()[i].getName() + "=" + postMethodM.getParameters()[i].getValue() );
// actually call
statusCodeM = httpclient.executeMethod(postMethodM);
// check parameters after call
log.debug("Called httpclient.executeMethod on object: " + postMethodM + " with status: " + statusCodeM + " containing parameters: " );
for(int i=0; i < postMethodM.getParameters().length; i++)
log.debug("\t" + i + "/ " + postMethodM.getParameters()[i].getName() + "=" + postMethodM.getParameters()[i].getValue() );
}
And here are the related traces:
Dec 10 JST 2008 09:05:48 AM.647063| 16301 | EventSender | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 About to call httpclient.executeMethod on object: org.apache.commons.httpclient.methods.PostMethod@210a500 containing parameters:
Dec 10 JST 2008 09:05:48 AM.647205| 16301 | EventSender | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 0/ eventname=undefined
Dec 10 JST 2008 09:05:48 AM.647340| 16301 | EventSender | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 1/ session.id=http://130.211.1.96/ccxmlsession_96_830_110700
Dec 10 JST 2008 09:05:48 AM.647757| 16301 | iThreadedHttpConnectionManager| 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 HttpConnectionManager.getConnection: config = HostConfiguration[host=http://localhost, proxyHost=http://127.0.0.1:3128], timeout = 10000
Dec 10 JST 2008 09:05:48 AM.647946| 16301 | nnectionManager$ConnectionPool| 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 Getting free connection, hostConfig=HostConfiguration[host=http://localhost, proxyHost=http://127.0.0.1:3128]
Dec 10 JST 2008 09:05:48 AM.648178| 16301 | HttpConnection | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 Connection is stale, closing...
Dec 10 JST 2008 09:05:48 AM.648484| 16301 | HttpConnection | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 Open connection to 127.0.0.1:3128
Dec 10 JST 2008 09:05:48 AM.649115| 16301 | Wire | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 >> "POST http://localhost/PreFetchServlet?src=/rVoiceMail_041/vxmlFiles/VoiceMail_0000.vxml&src=/rVoiceMail_041/vxmlFiles/VoiceMail_0001.vxml&src=/rVoiceMail_041/vxmlFiles/VoiceMail_0002.vxml.jsp&src=/rVoiceMail_041/vxmlFiles/VoiceMail_0003.vxml&src=/rVoiceMail_041/vxmlFiles/VoiceMail_0004.vxml&src=/rVoiceMail_041/vxmlFiles/VoiceMail_0005.vxml&src=/rVoiceMail_041/vxmlFiles/VoiceMail_0006.vxml&src=/rVoiceMail_041/vxmlFiles/VoiceMail_0007.vxml.jsp&src=/rVoiceMail_041/vxmlFiles/VoiceMail_0008.vxml.jsp&src=/rVoiceMail_041/vxmlFiles/VoiceMail_C005.jsf&src=/rVoiceMail_041/vxmlFiles/VoiceMail_C008.jsp&src=/rVoiceMail_041/grammarFiles/VoiceMail_C000.grxml&src=/rVoiceMail_041/grammarFiles/VoiceMail_C002.grxml&src=/rVoiceMail_041/ECMAScripts/VoiceMail_0000.es&src=/rVoiceMail_041/ECMAScripts/VoiceMail_C000.es&src=/rVoiceMail_041/promptFiles/VoiceMail_C000.jsf&src=/rVoiceMail_041/promptFiles/VoiceMail_C001.jsf&src=/rVoiceMail_041/promptFiles/VoiceMail_C002.jsf&src=/rVoiceMail_041/promptFiles/VoiceMail_0000.ulaw.jsp&src=/rVoiceMail_041/promptFiles/VoiceMail_0001.ulaw.jsp&src=/rVoiceMail_041/promptFiles/VoiceMail_C009.ulaw.jsp&src=/rVoiceMail_041/promptFiles/VoiceMail_C010.ulaw.jsp&src=/rVoiceMail_041/promptFiles/VoiceMail_C011.ulaw.jsp&src=/rVoiceMail_041/promptFiles/VoiceMail_C012.ulaw.jsp&src=/rVoiceMail_041/promptFiles/00430058.ulaw&src=/rVoiceMail_041/promptFiles/00420001.ulaw&src=/rVoiceMail_041/promptFiles/00440000.ulaw&src=/rVoiceMail_041/promptFiles/10430058.ulaw&src=/rVoiceMail_041/promptFiles/10420005.ulaw&chk HTTP/1.1[\r][\n]"
Dec 10 JST 2008 09:05:48 AM.649328| 16301 | HttpMethodBase | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 Adding Host request header
Dec 10 JST 2008 09:05:48 AM.649588| 16301 | HttpMethodBase | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 Default charset used: ISO-8859-1
Dec 10 JST 2008 09:05:48 AM.649825| 16301 | HttpMethodBase | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 Default charset used: ISO-8859-1
Dec 10 JST 2008 09:05:48 AM.650013| 16301 | Wire | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 >> "User-Agent: OCMP-BRE-CCXML[\r][\n]"
Dec 10 JST 2008 09:05:48 AM.650194| 16301 | Wire | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 >> "Host: localhost[\r][\n]"
Dec 10 JST 2008 09:05:48 AM.650373| 16301 | Wire | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 >> "Proxy-Connection: Keep-Alive[\r][\n]"
Dec 10 JST 2008 09:05:48 AM.650555| 16301 | Wire | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 >> "Content-Length: 19[\r][\n]"
Dec 10 JST 2008 09:05:48 AM.650738| 16301 | Wire | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 >> "Content-Type: application/x-www-form-urlencoded[\r][\n]"
Dec 10 JST 2008 09:05:48 AM.650925| 16301 | Wire | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 >> "[\r][\n]"
Dec 10 JST 2008 09:05:48 AM.651111| 16301 | Wire | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 >> "null=null&null=null"
Dec 10 JST 2008 09:05:48 AM.651701| 16301 | EntityEnclosingMethod | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 Request body sent
Dec 10 JST 2008 09:05:48 AM.679966| 16301 | Wire | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 << "HTTP/1.0 200 OK[\r][\n]"
Dec 10 JST 2008 09:05:48 AM.680165| 16301 | Wire | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 << "HTTP/1.0 200 OK[\r][\n]"
Dec 10 JST 2008 09:05:48 AM.680370| 16301 | Wire | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 << "Server: Apache-Coyote/1.1[\r][\n]"
Dec 10 JST 2008 09:05:48 AM.680559| 16301 | Wire | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 << "Content-Type: application/x-www-form-urlencoded[\r][\n]"
Dec 10 JST 2008 09:05:48 AM.680747| 16301 | Wire | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 << "Content-Length: 0[\r][\n]"
Dec 10 JST 2008 09:05:48 AM.680947| 16301 | Wire | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 << "Date: Wed, 10 Dec 2008 00:05:48 GMT[\r][\n]"
Dec 10 JST 2008 09:05:48 AM.681143| 16301 | Wire | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 << "X-Cache: MISS from ocmpsquid[\r][\n]"
Dec 10 JST 2008 09:05:48 AM.681331| 16301 | Wire | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 << "X-Cache-Lookup: MISS from ocmpsquid:3128[\r][\n]"
Dec 10 JST 2008 09:05:48 AM.681518| 16301 | Wire | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 << "Via: 1.0 ocmpsquid:3128 (squid/2.6.STABLE17)[\r][\n]"
Dec 10 JST 2008 09:05:48 AM.681704| 16301 | Wire | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 << "Proxy-Connection: keep-alive[\r][\n]"
Dec 10 JST 2008 09:05:48 AM.682050| 16301 | Wire | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 << "[\r][\n]"
Dec 10 JST 2008 09:05:48 AM.682229| 16301 | EventSender | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 Called httpclient.executeMethod on object: org.apache.commons.httpclient.methods.PostMethod@210a500 with status: 200 containing parameters:
Dec 10 JST 2008 09:05:48 AM.682373| 16301 | EventSender | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 0/ eventname=undefined
Dec 10 JST 2008 09:05:48 AM.682517| 16301 | EventSender | 0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 1/ session.id=http://130.211.1.96/ccxmlsession_96_830_110700
The wireshark capture is consistent as well:
POST http://localhost/PreFetchServlet?src=/rVoiceMail_041/vxmlFiles/VoiceMail_0000.vxml&src=/rVoiceMail_041/vxmlFiles/VoiceMail_0001.vxml&src=/rVoiceMail_041/vxmlFiles/VoiceMail_0002.vxml.jsp&src=/rVoiceMail_041/vxmlFiles/VoiceMail_0003.vxml&src=/rVoiceMail_041/vxmlFiles/VoiceMail_0004.vxml&src=/rVoiceMail_041/vxmlFiles/VoiceMail_0005.vxml&src=/rVoiceMail_041/vxmlFiles/VoiceMail_0006.vxml&src=/rVoiceMail_041/vxmlFiles/VoiceMail_0007.vxml.jsp&src=/rVoiceMail_041/vxmlFiles/VoiceMail_0008.vxml.jsp&src=/rVoiceMail_041/vxmlFiles/VoiceMail_C005.jsf&src=/rVoiceMail_041/vxmlFiles/VoiceMail_C008.jsp&src=/rVoiceMail_041/grammarFiles/VoiceMail_C000.grxml&src=/rVoiceMail_041/grammarFiles/VoiceMail_C002.grxml&src=/rVoiceMail_041/ECMAScripts/VoiceMail_0000.es&src=/rVoiceMail_041/ECMAScripts/VoiceMail_C000.es&src=/rVoiceMail_041/promptFiles/VoiceMail_C000.jsf&src=/rVoiceMail_041/promptFiles/VoiceMail_C001.jsf&src=/rVoiceMail_041/promptFiles/VoiceMail_C002.jsf&src=/rVoiceMail_041/promptFiles/VoiceMail_0000.ulaw.jsp&src=/rVoiceMail_041/promptFiles/VoiceMail_0001.ulaw.jsp&src=/rVoiceMail_041/promptFiles/VoiceMail_C009.ulaw.jsp&src=/rVoiceMail_041/promptFiles/VoiceMail_C010.ulaw.jsp&src=/rVoiceMail_041/promptFiles/VoiceMail_C011.ulaw.jsp&src=/rVoiceMail_041/promptFiles/VoiceMail_C012.ulaw.jsp&src=/rVoiceMail_041/promptFiles/00430058.ulaw&src=/rVoiceMail_041/promptFiles/00420001.ulaw&src=/rVoiceMail_041/promptFiles/00440000.ulaw&src=/rVoiceMail_041/promptFiles/10430058.ulaw&src=/rVoiceMail_041/promptFiles/10420005.ulaw&chk HTTP/1.1
User-Agent: OCMP-BRE-CCXML
Host: localhost
Proxy-Connection: Keep-Alive
Content-Length: 19
Content-Type: application/x-www-form-urlencoded
null=null&null=null
One more strange point: the problem requires intensive traffic to happen, but once the it has happened, all http post suffer the same problem, encoding null parameters ...
Would this problem ring any bell? Should I turn other traces on to get more clues?
Any feedback would be greatly appreciated.
Thanks a lot for your help.
Dominique Domet de Mont
Re: Httpclient 3.1 encodes null=null parameters under stressing conditions
Posted by Oleg Kalnichevski <ol...@apache.org>.
Domet De Mont, Dominique (d3m) wrote:
> Hi Oleg,
>
>
>
> Actually, I turned the wire log on to confirm the 'nullified' parameters encoding: I got this trace:
>
> Wire|0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 >> "null=null&null=null"
>
>
>
> So I added a PostMethod dump before and after the httpclient.executeMethod(postMethodM) invocation to check the parameter validity: I got the expected values:
>
> EventSender|0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 0/ eventname=undefined
>
> EventSender|0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 1/ session.id=http://130.211.1.96/ccxmlsession_96_830_110700
>
>
>
> As per multithreading support, I understand from this page http://hc.apache.org/httpclient-3.x/threading.html that MultiThreadedHttpConnectionManager brings thread safe http clients; am I wrong?
>
The HttpClient is threading safe, when used with a threading safe
connection manager, but not HTTP methods.
Hope this helps
Oleg
---------------------------------------------------------------------
To unsubscribe, e-mail: httpclient-users-unsubscribe@hc.apache.org
For additional commands, e-mail: httpclient-users-help@hc.apache.org
RE: Httpclient 3.1 encodes null=null parameters under stressing
conditions
Posted by "Domet De Mont, Dominique (d3m)" <do...@hp.com>.
Hi Oleg,
Actually, I turned the wire log on to confirm the 'nullified' parameters encoding: I got this trace:
Wire|0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 >> "null=null&null=null"
So I added a PostMethod dump before and after the httpclient.executeMethod(postMethodM) invocation to check the parameter validity: I got the expected values:
EventSender|0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 0/ eventname=undefined
EventSender|0| DEBUG | http://130.211.1.96/ccxmlsession_96_830_110700/event_288797 1/ session.id=http://130.211.1.96/ccxmlsession_96_830_110700
As per multithreading support, I understand from this page http://hc.apache.org/httpclient-3.x/threading.html that MultiThreadedHttpConnectionManager brings thread safe http clients; am I wrong?
Thanks again.
Dominique
-----Original Message-----
From: sebb [mailto:sebbaz@gmail.com]
Sent: Thursday, December 11, 2008 5:15 PM
To: HttpClient User Discussion
Subject: Re: Httpclient 3.1 encodes null=null parameters under stressing conditions
On 11/12/2008, Oleg Kalnichevski <ol...@apache.org> wrote:
> On Wed, 2008-12-10 at 13:33 +0000, Domet De Mont, Dominique (d3m) wrote:
> > Dear List,
> >
> > I am facing a strange httpclient 3.1 behaviour under stressing conditions while encoding a post: the set of parameters is "nullified", whereas the number of parameters is correct.
> >
> > I have turned on the httpclient wire traces to get more clues, and added in my own code a dump of the PostMethod object before and after the httpclient.execute call: I see the actual parameters correctly set in the PostMethod object, but actually the httpclient encoding "nullified" parameters.
> >
> > Here is the code excerpt:
> > synchronized(this) {
> > // Check parameters before call
> > log.debug("About to call httpclient.executeMethod on object: " + postMethodM + " containing parameters: " );
> > for(int i=0; i < postMethodM.getParameters().length; i++)
> > log.debug("\t" + i + "/ " + postMethodM.getParameters()[i].getName() + "=" + postMethodM.getParameters()[i].getValue() );
> > // actually call
> > statusCodeM = httpclient.executeMethod(postMethodM);
> > // check parameters after call
> > log.debug("Called httpclient.executeMethod on object: " + postMethodM + " with status: " + statusCodeM + " containing parameters: " );
> > for(int i=0; i < postMethodM.getParameters().length; i++)
> > log.debug("\t" + i + "/ " + postMethodM.getParameters()[i].getName() + "=" + postMethodM.getParameters()[i].getValue() );
> > }
> >
>
>
> Do not do that. HttpMethods are not threading safe.
Not sure I follow - surely it is OK to call the methods in the same
thread, as in the sample above?
> Just turn on the
> wire log if you need to examine what gets sent across the wire.
>
> Oleg
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: httpclient-users-unsubscribe@hc.apache.org
> For additional commands, e-mail: httpclient-users-help@hc.apache.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: httpclient-users-unsubscribe@hc.apache.org
For additional commands, e-mail: httpclient-users-help@hc.apache.org
Re: Httpclient 3.1 encodes null=null parameters under stressing conditions
Posted by sebb <se...@gmail.com>.
On 11/12/2008, Oleg Kalnichevski <ol...@apache.org> wrote:
> On Wed, 2008-12-10 at 13:33 +0000, Domet De Mont, Dominique (d3m) wrote:
> > Dear List,
> >
> > I am facing a strange httpclient 3.1 behaviour under stressing conditions while encoding a post: the set of parameters is "nullified", whereas the number of parameters is correct.
> >
> > I have turned on the httpclient wire traces to get more clues, and added in my own code a dump of the PostMethod object before and after the httpclient.execute call: I see the actual parameters correctly set in the PostMethod object, but actually the httpclient encoding "nullified" parameters.
> >
> > Here is the code excerpt:
> > synchronized(this) {
> > // Check parameters before call
> > log.debug("About to call httpclient.executeMethod on object: " + postMethodM + " containing parameters: " );
> > for(int i=0; i < postMethodM.getParameters().length; i++)
> > log.debug("\t" + i + "/ " + postMethodM.getParameters()[i].getName() + "=" + postMethodM.getParameters()[i].getValue() );
> > // actually call
> > statusCodeM = httpclient.executeMethod(postMethodM);
> > // check parameters after call
> > log.debug("Called httpclient.executeMethod on object: " + postMethodM + " with status: " + statusCodeM + " containing parameters: " );
> > for(int i=0; i < postMethodM.getParameters().length; i++)
> > log.debug("\t" + i + "/ " + postMethodM.getParameters()[i].getName() + "=" + postMethodM.getParameters()[i].getValue() );
> > }
> >
>
>
> Do not do that. HttpMethods are not threading safe.
Not sure I follow - surely it is OK to call the methods in the same
thread, as in the sample above?
> Just turn on the
> wire log if you need to examine what gets sent across the wire.
>
> Oleg
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: httpclient-users-unsubscribe@hc.apache.org
> For additional commands, e-mail: httpclient-users-help@hc.apache.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: httpclient-users-unsubscribe@hc.apache.org
For additional commands, e-mail: httpclient-users-help@hc.apache.org
Re: Httpclient 3.1 encodes null=null parameters under stressing
conditions
Posted by Oleg Kalnichevski <ol...@apache.org>.
On Wed, 2008-12-10 at 13:33 +0000, Domet De Mont, Dominique (d3m) wrote:
> Dear List,
>
> I am facing a strange httpclient 3.1 behaviour under stressing conditions while encoding a post: the set of parameters is "nullified", whereas the number of parameters is correct.
>
> I have turned on the httpclient wire traces to get more clues, and added in my own code a dump of the PostMethod object before and after the httpclient.execute call: I see the actual parameters correctly set in the PostMethod object, but actually the httpclient encoding "nullified" parameters.
>
> Here is the code excerpt:
> synchronized(this) {
> // Check parameters before call
> log.debug("About to call httpclient.executeMethod on object: " + postMethodM + " containing parameters: " );
> for(int i=0; i < postMethodM.getParameters().length; i++)
> log.debug("\t" + i + "/ " + postMethodM.getParameters()[i].getName() + "=" + postMethodM.getParameters()[i].getValue() );
> // actually call
> statusCodeM = httpclient.executeMethod(postMethodM);
> // check parameters after call
> log.debug("Called httpclient.executeMethod on object: " + postMethodM + " with status: " + statusCodeM + " containing parameters: " );
> for(int i=0; i < postMethodM.getParameters().length; i++)
> log.debug("\t" + i + "/ " + postMethodM.getParameters()[i].getName() + "=" + postMethodM.getParameters()[i].getValue() );
> }
>
Do not do that. HttpMethods are not threading safe. Just turn on the
wire log if you need to examine what gets sent across the wire.
Oleg
---------------------------------------------------------------------
To unsubscribe, e-mail: httpclient-users-unsubscribe@hc.apache.org
For additional commands, e-mail: httpclient-users-help@hc.apache.org