Thanks for finding the issue. Shouldn't we set this system property in HTTPJavaImpl directly as it's mandatory for Jmeter.
Thanks On Monday, February 27, 2017, sebb <seb...@gmail.com> wrote: > On 27 February 2017 at 00:30, sebb <seb...@gmail.com <javascript:;>> > wrote: > > http://bugs.java.com/bugdatabase/view_bug.do?bug_id=6996110 > > > > requires > > > > -Dsun.net.http.allowRestrictedHeaders=true > > > > to be set. > > > > This is done in system.properties. > > > > system.properties is enabled in jmeter.properties. > > > > So there must be some JMeter code that causes the Java HTTP code to > > pick up the setting between the initial > > -ptestfiles/jmeter-batch.properties and the -qjmeter.properties > > > > A temporary fix is to add > > > > system.properties=system.properties > > > > to > > > > jmeter-batch.properties > > > > But ideally the code that wakes up the Java HTTP code should be found > > and moved so it occurs later. > > Otherwise it may cause other strange behaviour. > > Actually the problem is that the system.properties value is only > checked after loading the initial properties file. > The secondary property files are processed later. > So the system properties are not processed. > > The default log level is now WARN, so the initial > -ptestfiles/jmeter-batch.properties is no longer needed. > > > > > On 26 February 2017 at 22:42, Philippe Mouawad > > <philippe.moua...@gmail.com <javascript:;>> wrote: > >> Strangely: > >> conn.setRequestProperty(HTTPConstants.HEADER_CONNECTION, > >> HTTPConstants.KEEP_ALIVE); > >> > >> has no effect, calling conn.getRequestProperties() returns an empty map. > >> If the other order of options is used , then it is ok. > >> > >> This is really weird. > >> > >> > >> > >> On Sun, Feb 26, 2017 at 11:28 PM, Philippe Mouawad < > >> philippe.moua...@gmail.com <javascript:;>> wrote: > >> > >>> With current order of properties (-p / -q) it seems : > >>> Connection: keep-alive is missing > >>> > >>> On Sun, Feb 26, 2017 at 11:15 PM, Philippe Mouawad < > >>> philippe.moua...@gmail.com <javascript:;>> wrote: > >>> > >>>> Shouldn't it be : > >>>> -pjmeter.properties > >>>> -q /testfiles/jmeter-batch.properties > >>>> > >>>> instead of > >>>> -qjmeter.properties > >>>> -p /testfiles/jmeter-batch.properties > >>>> > >>>> > >>>> > >>>> On Sun, Feb 26, 2017 at 11:14 PM, Philippe Mouawad < > >>>> philippe.moua...@gmail.com <javascript:;>> wrote: > >>>> > >>>>> I've found out that it's due to: > >>>>> -p /testfiles/jmeter-batch.properties > >>>>> > >>>>> On Sun, Feb 26, 2017 at 11:10 PM, Philippe Mouawad < > >>>>> philippe.moua...@gmail.com <javascript:;>> wrote: > >>>>> > >>>>>> Hello, > >>>>>> If anybody understands this failure I'll be happy to get some help. > >>>>>> > >>>>>> When I run test in GUI and NON GuI mode it succeeds and generates > the > >>>>>> correct csv. > >>>>>> But when ran from ant through batchtest it fails. > >>>>>> > >>>>>> Any idea ? > >>>>>> Regards > >>>>>> Philippe > >>>>>> > >>>>>> On Sun, Feb 26, 2017 at 11:07 PM, <build...@apache.org > <javascript:;>> wrote: > >>>>>> > >>>>>>> The Buildbot has detected a new failure on builder jmeter-trunk > while > >>>>>>> building . Full details are available at: > >>>>>>> https://ci.apache.org/builders/jmeter-trunk/builds/2407 > >>>>>>> > >>>>>>> Buildbot URL: https://ci.apache.org/ > >>>>>>> > >>>>>>> Buildslave for this Build: bb_slave1_ubuntu > >>>>>>> > >>>>>>> Build Reason: The AnyBranchScheduler scheduler named > >>>>>>> 'on-jmeter-commit' triggered this build > >>>>>>> Build Source Stamp: [branch jmeter/trunk] 1784486 > >>>>>>> Blamelist: pmouawad > >>>>>>> > >>>>>>> BUILD FAILED: failed shell_3 > >>>>>>> > >>>>>>> Sincerely, > >>>>>>> -The Buildbot > >>>>>>> > >>>>>>> > >>>>>>> > >>>>>>> > >>>>>> > >>>>>> > >>>>>> -- > >>>>>> Cordialement. > >>>>>> Philippe Mouawad. > >>>>>> > >>>>>> > >>>>>> > >>>>> > >>>>> > >>>>> -- > >>>>> Cordialement. > >>>>> Philippe Mouawad. > >>>>> > >>>>> > >>>>> > >>>> > >>>> > >>>> -- > >>>> Cordialement. > >>>> Philippe Mouawad. > >>>> > >>>> > >>>> > >>> > >>> > >>> -- > >>> Cordialement. > >>> Philippe Mouawad. > >>> > >>> > >>> > >> > >> > >> -- > >> Cordialement. > >> Philippe Mouawad. > -- Cordialement. Philippe Mouawad.