I did some more debugging and it seems when going through the proxy the browser does not send "Accept-Encoding: gzip, deflate".
I think the best way for you guys to debug this problem is to give you access to our test environment. We are setting up a temp account for you guys. Please send me an email at [EMAIL PROTECTED] and I will send you the connection information. Alf Høgemark wrote: > > Just as a tip : > If you add the "View Results Tree" listener to the "HTTP Proxy Server", > you might be able to see more details on what data the "HTTP Proxy > Server" sends and receives. > > Alf > > Alf Høgemark wrote: >> Hi >> >> Yes, there could be a bug in the HTTP Proxy Server, for example the >> bug just reported : >> "43625 : system.xml.xmlexception: there is no Unicode byte order". >> It sounds like the HTTP Proxy have some trouble decoding some pages, >> something similar was also reported last week. >> >> A bug report with details on how to reproduce would be great, it would >> help us to solve the problem quicker. >> Can you cwang create such a bug report, please ? >> >> Regards >> Alf >> >> >> sebb wrote: >>> Just tried using IE7 to access GMail, and JMeter works fine for me >>> without needing to eliminate the gzip header. >>> >>> Perhaps there are other types of gzipped content that are not handled >>> so well. >>> >>> S. >>> On 16/10/2007, cwang <[EMAIL PROTECTED]> wrote: >>> >>>> I have already changed the code so that it leaves the >>>> "Content-encoding: >>>> gzip" in. >>>> If "Content-encoding: gzip" is removed IE7 will display a page >>>> cannot be >>>> displayed error. After I made the change the page shows up as it >>>> should and >>>> with the source. >>>> >>>> >>>> >>>> Alf Høgemark wrote: >>>> >>>>> Hi >>>>> >>>>> The http proxy server is unzipping the content it gets from the web >>>>> server, and then passing it on to IE uncompressed, so it has to remove >>>>> the "Content-encoding: gzip" header. >>>>> So if you are having a problem, I do not think that the removal of >>>>> that >>>>> header is the problem. >>>>> >>>>> When you are running the test plan made with the proxy, do you get the >>>>> expected results from the web server then ? >>>>> >>>>> Do you see anything in IE7 if you do a "view source", when you are >>>>> using >>>>> the proxy ? >>>>> >>>>> Regards >>>>> Alf >>>>> >>>>> cwang wrote: >>>>> >>>>>> oh sorry I miss understood your question. >>>>>> >>>>>> The full list of headers >>>>>> >>>>>> HTTP/1.1 200 OK >>>>>> Date: Mon, 15 Oct 2007 15:32:06 GMT >>>>>> Server: WebSphere Application Server/6.1 >>>>>> Keep-Alive: timeout=300, max=70 >>>>>> Connection: Keep-Alive >>>>>> Content-Type: text/html; charset=utf-8 >>>>>> Content-Encoding: gzip >>>>>> Content-Language: en-US >>>>>> >>>>>> it actually just takes out "Content-Encoding: gzip" >>>>>> >>>>>> >>>>>> >>>>>> sebb-2 wrote: >>>>>> >>>>>> >>>>>>> On 13/10/2007, cwang <[EMAIL PROTECTED]> wrote: >>>>>>> >>>>>>> >>>>>>>>> What response header do you get? >>>>>>>>> >>>>>>>>> >>>>>>> What actual response header do you get from the application? >>>>>>> >>>>>>> >>>>>>> >>>>>>>> when I said "original headers" i mean the headers past into >>>>>>>> massageResponseHeaders(SampleResult res, String headers),so >>>>>>>> pretty much >>>>>>>> by >>>>>>>> passed that method. >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>>> I think that's because the JMeter Proxy uses a sampler to get the >>>>>>>>> response, and the sampler will already have unzipped the data. >>>>>>>>> >>>>>>>>> >>>>>>>> That makes sense but then why would removing the "Content-Encoding: >>>>>>>> gzip" >>>>>>>> header cause IE7 to not display the page? >>>>>>>> >>>>>>>> >>>>>>> If it just removes "gzip" from the header, that might leave >>>>>>> something >>>>>>> invalid, which is why I asked what the actual header was. >>>>>>> >>>>>>> >>>>>>> >>>>>>>> I understand adding "Accept-Encoding: gzip, deflate" will probably >>>>>>>> cause >>>>>>>> havoc on trying to proxy pages without that, but on the >>>>>>>> application we >>>>>>>> are >>>>>>>> testing every page has that so it should be ok. Do you have any >>>>>>>> idea >>>>>>>> what >>>>>>>> other problems these changes will cause? I would like to look into >>>>>>>> them >>>>>>>> before we start full scale testing. >>>>>>>> >>>>>>>> >>>>>>> No idea, but I presume the code is there for a reason. >>>>>>> >>>>>>> >>>>>>> >>>>>>>> sebb-2 wrote: >>>>>>>> >>>>>>>> >>>>>>>>> On 12/10/2007, cwang <[EMAIL PROTECTED]> wrote: >>>>>>>>> >>>>>>>>> >>>>>>>>>> when my browser sent a header with "Accept-Encoding: gzip, >>>>>>>>>> deflate" >>>>>>>>>> it >>>>>>>>>> doesn't seem to be picked up by jmeter. It was kind of weird >>>>>>>>>> cause >>>>>>>>>> it >>>>>>>>>> wasn't even in the input stream. >>>>>>>>>> >>>>>>>>>> then when I get the response there was a piece of code in >>>>>>>>>> Jmeter that >>>>>>>>>> removed "Content-Encoding: gzip" >>>>>>>>>> in proxy.java >>>>>>>>>> >>>>>>>>>> /** >>>>>>>>>> * In the event the content was gzipped and unpacked, the >>>>>>>>>> content-encoding >>>>>>>>>> * header must be removed and the content-length header >>>>>>>>>> should >>>>>>>>>> >>>>>>>>>> >>>>>>>> be >>>>>>>> >>>>>>>> >>>>>>>>>> corrected. >>>>>>>>>> * >>>>>>>>>> * @param res >>>>>>>>>> * @param headers >>>>>>>>>> * @return >>>>>>>>>> */ >>>>>>>>>> private String massageResponseHeaders(SampleResult res, >>>>>>>>>> String >>>>>>>>>> headers) { >>>>>>>>>> int encodingHeaderLoc = headers.indexOf(": >>>>>>>>>> gzip"); // >>>>>>>>>> $NON-NLS-1$ >>>>>>>>>> String newHeaders = headers; >>>>>>>>>> if (encodingHeaderLoc > -1) { >>>>>>>>>> int end = headers.indexOf(NEW_LINE, >>>>>>>>>> encodingHeaderLoc); >>>>>>>>>> int begin = headers.lastIndexOf(NEW_LINE, >>>>>>>>>> encodingHeaderLoc); >>>>>>>>>> newHeaders = newHeaders.substring(0, >>>>>>>>>> begin) + >>>>>>>>>> newHeaders.substring(end); >>>>>>>>>> int lengthIndex = >>>>>>>>>> newHeaders.indexOf("ength: >>>>>>>>>> >>>>>>>>>> >>>>>>>> "); >>>>>>>> >>>>>>>> >>>>>>>>>> // $NON-NLS-1$ >>>>>>>>>> end = newHeaders.indexOf(NEW_LINE, >>>>>>>>>> >>>>>>>>>> >>>>>>>> lengthIndex); >>>>>>>> >>>>>>>> >>>>>>>>>> newHeaders = newHeaders.substring(0, >>>>>>>>>> >>>>>>>>>> >>>>>>>> lengthIndex + >>>>>>>> >>>>>>>> >>>>>>>>>> 7) + >>>>>>>>>> res.getResponseData().length >>>>>>>>>> + >>>>>>>>>> newHeaders.substring(end); >>>>>>>>>> } >>>>>>>>>> return newHeaders; >>>>>>>>>> } >>>>>>>>>> >>>>>>>>>> I don't know why this piece of code was added but it cause my >>>>>>>>>> browser, >>>>>>>>>> Internet Explorer 7, to return with a page cannot be displayed >>>>>>>>>> error. >>>>>>>>>> >>>>>>>>>> >>>>>>>>> I think that's because the JMeter Proxy uses a sampler to get the >>>>>>>>> response, and the sampler will already have unzipped the data. >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>>> by hard coding "Accept-Encoding: gzip, deflate" into all my >>>>>>>>>> request >>>>>>>>>> header >>>>>>>>>> and returning the original "headers" from >>>>>>>>>> massageResponseHeaders it >>>>>>>>>> >>>>>>>>>> >>>>>>>> seems >>>>>>>> >>>>>>>> >>>>>>>>>> to >>>>>>>>>> work for my case. >>>>>>>>>> >>>>>>>>>> >>>>>>>>> What response header do you get? >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>>> I was wondering if these changes would cause error or corrupt >>>>>>>>>> data in >>>>>>>>>> other >>>>>>>>>> areas of JMeter. >>>>>>>>>> >>>>>>>>>> >>>>>>>>> Probably. >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>>> Thanks >>>>>>>>>> -- >>>>>>>>>> View this message in context: >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> >>>>>>>> http://www.nabble.com/a-few-problems-I-encountered-gzip-and-header-tf4614182.html#a13177025 >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>>>> Sent from the JMeter - User mailing list archive at Nabble.com. >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> --------------------------------------------------------------------- >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> To unsubscribe, e-mail: >>>>>>>>>> [EMAIL PROTECTED] >>>>>>>>>> For additional commands, e-mail: >>>>>>>>>> [EMAIL PROTECTED] >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> >>>>>>>>> --------------------------------------------------------------------- >>>>>>>>> >>>>>>>>> To unsubscribe, e-mail: [EMAIL PROTECTED] >>>>>>>>> For additional commands, e-mail: >>>>>>>>> [EMAIL PROTECTED] >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>> -- >>>>>>>> View this message in context: >>>>>>>> http://www.nabble.com/a-few-problems-I-encountered-gzip-and-header-tf4614182.html#a13185274 >>>>>>>> >>>>>>>> >>>>>>>> Sent from the JMeter - User mailing list archive at Nabble.com. >>>>>>>> >>>>>>>> >>>>>>>> --------------------------------------------------------------------- >>>>>>>> >>>>>>>> To unsubscribe, e-mail: [EMAIL PROTECTED] >>>>>>>> For additional commands, e-mail: >>>>>>>> [EMAIL PROTECTED] >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>> --------------------------------------------------------------------- >>>>>>> >>>>>>> To unsubscribe, e-mail: [EMAIL PROTECTED] >>>>>>> For additional commands, e-mail: [EMAIL PROTECTED] >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>> >>>>> >>>> -- >>>> View this message in context: >>>> http://www.nabble.com/a-few-problems-I-encountered-gzip-and-header-tf4614182.html#a13235710 >>>> >>>> >>>> Sent from the JMeter - User mailing list archive at Nabble.com. >>>> >>>> >>>> --------------------------------------------------------------------- >>>> To unsubscribe, e-mail: [EMAIL PROTECTED] >>>> For additional commands, e-mail: [EMAIL PROTECTED] >>>> >>>> >>>> >>> >>> --------------------------------------------------------------------- >>> To unsubscribe, e-mail: [EMAIL PROTECTED] >>> For additional commands, e-mail: [EMAIL PROTECTED] >>> >>> >>> >> > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > > > -- View this message in context: http://www.nabble.com/a-few-problems-I-encountered-gzip-and-header-tf4614182.html#a13297622 Sent from the JMeter - User mailing list archive at Nabble.com. --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]