Am 26. Januar 2017 16:57:50 MEZ schrieb Philippe Mouawad <philippe.moua...@gmail.com>: >On Thursday, January 26, 2017, Felix Schumacher < >felix.schumac...@internetallee.de> wrote: > >> Am 26.01.2017 14:19, schrieb Philippe Mouawad: >> >>> Thanks Felix. >>> Maybe we should ensure it only affect this part. >>> >> >> Do you have something special in mind? > >I meant that, are we sure user is in this case.
I still don't get it. Do you mean if we are sure that this is the OPs problem? >I agree with patch although shouldn't it close BoundedStream also ? Is it not? It should both be in the try clause. Felix > > >> >> I don't think the patch has more impact than not to leak fds. >> >> Yes > >> On the other hand, I think we should look at all other instances of >> IOUtils.copy, whether we leak fds there, too. > > >IOUtils.toString also I suppose ? > >> >> Felix >> >> >>> Thanks >>> >>> On Thursday, January 26, 2017, Felix Schumacher < >>> felix.schumac...@internetallee.de> wrote: >>> >>> Am 26.01.2017 12:09, schrieb Felix Schumacher: >>>> >>>> Am 26. Januar 2017 09:02:55 MEZ schrieb bugzi...@apache.org: >>>>> >>>>> https://bz.apache.org/bugzilla/show_bug.cgi?id=60607 >>>>>> >>>>>> --- Comment #12 from Philippe Mouawad ><p.moua...@ubik-ingenierie.com> >>>>>> --- >>>>>> (In reply to Sean Chang from comment #10) >>>>>> >>>>>> And some request failed due to file not found, but actually the >file >>>>>>> >>>>>>> is >>>>>> >>>>>> there and succeed for some other requests. >>>>>>> >>>>>>> java.io.FileNotFoundException: >/home/cosben/cos_upload/2bill/200K.ts >>>>>>> >>>>>>> (Too >>>>>> >>>>>> many open files) >>>>>>> at java.io.FileInputStream.open0(Native Method) >>>>>>> at >java.io.FileInputStream.open(FileInputStream.java:195) >>>>>>> at >java.io.FileInputStream.<init>(FileInputStream.java:138) >>>>>>> at >org.apache.http.entity.FileEntity.getContent(FileEntity. >>>>>>> java:88) >>>>>>> at >>>>>>> org.apache.jmeter.protocol.http.sampler.HTTPHC4Impl. >>>>>>> sendEntityData(HTTPHC4Impl.java:1459) >>>>>>> at >>>>>>> >>>>>>> >>>>>> I think we are leaking file descriptors here. I will look into >it. >>>>> >>>>> >>>> You can see the fd leak, if you use httpclient 4 with PUT and a >file from >>>> disk to upload. >>>> The attached patch should fix it. I will submit a bug this evening >and >>>> commit the patch. >>>> >>>> Felix >>>> >>>> >>>> Felix >>>>> >>>>> >>>>> org.apache.jmeter.protocol.http.sampler.HTTPHC4Impl.handleMe >>>>>>> >>>>>> thod(HTTPHC4Impl. >>>>>> >>>>>> java:585) >>>>>>> at >>>>>>> >>>>>>> org.apache.jmeter.protocol.http.sampler.HTTPHC4Impl.sample( >>>>>>> >>>>>> HTTPHC4Impl.java: >>>>>> >>>>>> 399) >>>>>>> at >>>>>>> org.apache.jmeter.protocol.http.sampler.HTTPSamplerProxy. >>>>>>> sample(HTTPSamplerProxy.java:74) >>>>>>> at >>>>>>> org.apache.jmeter.protocol.http.sampler.HTTPSamplerBase. >>>>>>> sample(HTTPSamplerBase.java:1176) >>>>>>> at >>>>>>> org.apache.jmeter.protocol.http.sampler.HTTPSamplerBase. >>>>>>> sample(HTTPSamplerBase.java:1165) >>>>>>> at >>>>>>> >>>>>>> org.apache.jmeter.threads.JMeterThread.executeSamplePackage( >>>>>>> >>>>>> JMeterThread. >>>>>> >>>>>> java:473) >>>>>>> at >>>>>>> >>>>>>> org.apache.jmeter.threads.JMeterThread.processSampler(JMeter >>>>>>> >>>>>> Thread.java:416) >>>>>> >>>>>> at org.apache.jmeter.threads.JMet >>>>>>> erThread.run(JMeterThread.java >>>>>>> :252) >>>>>>> at java.lang.Thread.run(Thread.java:745) >>>>>>> >>>>>>> >>>>>> This does not mean the file is not here, it means you reach a >linux >>>>>> limit on >>>>>> file descriptors. >>>>>> If you don't hit this with the other version of JMeter then it >could be >>>>>> a >>>>>> jmeter issue but it's not sure. >>>>>> https://easyengine.io/tutorials/linux/increase-open-files-limit/ >>>>>> >>>>>> >>>>>> 1/ Run ulimit -a with the user that runs jmeter . >>>>>> 2/ How many threads are you running ? >>>>>> Did you compare versions of JMeter with exactly the same script >and >>>>>> user.properties/jmeter.properties ? >>>>>> Thank you >>>>>> >>>>>> >>>>>
