Greetings There is one more version that's just been released to http://pogstest.theskynet.org/pogstest. This one is built from a recent commit to BOINC repo, which excludes continuous cpu time calculation for Android to overcome a bug that crashes the wrapper. It's pretty much identical to version 0.09 that was recently released. I ask that all users testing to update/reset to receive version 0.10 app tasks. This will probably be the one released to http://pogs.theskynet.org/pogs.
Additionally, I am going to keep an instance of these wrapper tasks up at my project site here <http://akira.onburde.net/burdetest/>, as we still need to get to the bottom of the problem. Anyone that has this issue and willing to spare an Android device out of hours please attach to the project so I can collect relevant debugging output for analysis. Thanks to all that have helped thus far. Cheers Daniel On Wed, Aug 7, 2013 at 4:18 PM, Daniel Carrion <[email protected]> wrote: > Hello Again > > I have pushed out yet another version that outputs during task poll loop > for 60 counts. If you can, please attach so we can see if this reveals any > patterns as to where it's exiting. > > Regards > > Daniel > > On Tue, Aug 6, 2013 at 4:42 PM, Daniel Carrion <[email protected]>wrote: > >> Hello >> >> Thanks to everyone that has attached so far. I can see there's errors in >> the task out...hope there's something useful in that logcat dump :). >> >> I have released another version that removes redirection to child >> specific stderr output file. Please update/reset project on your phones to >> get this version. >> >> Cheers >> >> Daniel >> >> >> On Mon, Aug 5, 2013 at 5:02 PM, Daniel Carrion <[email protected]>wrote: >> >>> Hello All >>> >>> We've set up a test project site to troubleshoot problems with POGS >>> tasks on Android. We invite everyone currently experiencing computation >>> error problems to help us test. The wrapper being distributed for this >>> project has debugging flags turned on and more logging. >>> >>> The test project site to attach to: >>> http://pogstest.theskynet.org/pogstest. >>> >>> You will need to create a new account. >>> >>> For those that are able, we ask that you run logcat in the background >>> before POGS tasks begin running. You can do this from a terminal emulator >>> on your Android or by invoking adb from the SDK and then running: >>> >>> logcat -s DEBUG > /sdcard/Download/logcat.txt >>> >>> If the tasks crash, there should be a some useful dump information in >>> that logcat.txt file you can send me. >>> >>> We appreciate any assistance in helping us debug this problem. If you >>> are helping out, please post over at >>> http://pogs.theskynet.org/pogs/forum_thread.php?id=250. >>> >>> Kind Regards >>> >>> Daniel >>> >>> On Thu, Aug 1, 2013 at 10:49 PM, Daniel Carrion <[email protected]>wrote: >>> >>>> I apologize if I've confused the situation. I think for now it's >>>> probably best if users experiencing the same errors on their devices stop >>>> crunching POGS until we get this issue resolved. It would be great if >>>> those that are and are willing to help troubleshoot could head over to POGS >>>> forums <http://pogs.theskynet.org/pogs/> and post. >>>> >>>> From Iris' assistance troubleshooting, we know that there is a definite >>>> problem with the wrapper exiting with SIGILL and SIGSEGV on some devices. I >>>> can see these errors on other Android hosts crunching POGS. The dumps Iris >>>> provided confirm this, however, it's not clear as to why it's happening. >>>> >>>> I am now preparing a wrapper version that will have better debugging >>>> and logging to see what exactly causes it to crash. Once ready I will post >>>> for willing users experiencing the problem to help me test it and then >>>> review the stderr logs once the tasks report back. >>>> >>>> At the same time I've also contacted the NativeBOINC developer to see >>>> if he can provide some insight. I'm also having a poke around other Android >>>> projects to see if they are getting similar problems. >>>> >>>> I again apologize for the confusion and making the thread convoluted. >>>> It may be better if this troubleshooting happen on the POGS forum instead >>>> of this group as it's hard to follow as an email correspondence. >>>> >>>> Regards >>>> >>>> Daniel >>>> >>>> On Thu, Aug 1, 2013 at 10:18 PM, Iris Ailin-Pyzik >>>> <[email protected]>wrote: >>>> >>>>> OK, just to clear somethings up with this, because I think the thread >>>>> is convoluted. >>>>> >>>>> I was running BOINC and having all these errors on POGS, so Daniel >>>>> asked me to try NativeBOINC to see if he could make more sense from the >>>>> bugcatcher feature. The dumps below are from bugcatcher. >>>>> >>>>> To complicate further - >>>>> >>>>> Last night I left my phone running NativeBOINC and now it will not run >>>>> POGS at all. It downloads a task, and almost immediately says it is done >>>>> but there is no output file, and fetches another. I rebooted the phone - >>>>> same deal. See >>>>> >>>>> http://pogs.theskynet.org/pogs/results.php?hostid=10438 >>>>> http://pogs.theskynet.org/pogs/results.php?hostid=10529 >>>>> >>>>> It seems to be treating the host as a new one after I re-started >>>>> NativeBOINC. >>>>> >>>>> I suspended the POGS project and deleted the tasks, added Einstein >>>>> which ran well under BOINC, and it seems to be happily chugging away on >>>>> that one. >>>>> >>>>> Just by way of contrast, this is an old Dell Streak that lives in our >>>>> kitchen for internet radio, and has two errors but otherwise is running >>>>> POGS under BOINC with no problems - >>>>> >>>>> http://pogs.theskynet.org/pogs/results.php?hostid=9083 >>>>> >>>>> Iris >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> On Thu, Aug 1, 2013 at 2:11 AM, Daniel Carrion >>>>> <[email protected]>wrote: >>>>> >>>>>> So just an update on this. Iris has provided me with some output from >>>>>> NativeBOINC BugCatcher mode. It seems that it is the wrapper bombing out. >>>>>> Extract from log files >>>>>> >>>>>> One with SIGILL >>>>>> >>>>>> ---- NATIVEBOINC BUGCATCH REPORT HEADER ---- >>>>>> BugReportTime:1375312087.187 >>>>>> CommandLine: >>>>>> "../../projects/pogs.theskynet.org_pogs/wrapper_arm-android-linux-gnu_340" >>>>>> Manufacturer:motorola >>>>>> ModelName:XT897 >>>>>> AndroidVer:4.1.2 >>>>>> KernelVer:3.0.42-gbd030d1 >>>>>> NativeBOINCVer:0.4.4.2.2b >>>>>> Pid:5990 >>>>>> Signal:4 >>>>>> >>>>>> Other with SIGSEGV: >>>>>> >>>>>> ---- NATIVEBOINC BUGCATCH REPORT HEADER ---- >>>>>> BugReportTime:1375317764.520 >>>>>> CommandLine: >>>>>> "../../projects/pogs.theskynet.org_pogs/wrapper_arm-android-linux-gnu_340" >>>>>> Manufacturer:motorola >>>>>> ModelName:XT897 >>>>>> AndroidVer:4.1.2 >>>>>> KernelVer:3.0.42-gbd030d1 >>>>>> NativeBOINCVer:0.4.4.2.2b >>>>>> Pid:6000 >>>>>> Signal:11 >>>>>> >>>>>> At first glance of the stack trace, seems to be when wrapper goes to >>>>>> start new task. I've contacted Mateusz from NativeBOINC for his thoughts >>>>>> on >>>>>> the output. >>>>>> >>>>>> In the meantime I will be whacking in a whole lot of debug output >>>>>> into the wrapper code to see exactly where it exits out. I would like to >>>>>> attach GDB to the running process but at the moment I don't want to have >>>>>> to >>>>>> bother people in doing this. It's unfortunate I can't reproduce this. >>>>>> >>>>>> Thank you Iris for your assistance thus far. >>>>>> >>>>>> Regards >>>>>> >>>>>> Daniel >>>>>> >>>>>> On Thu, Aug 1, 2013 at 4:47 AM, Iris Ailin-Pyzik >>>>>> <[email protected]>wrote: >>>>>> >>>>>>> Daniel - >>>>>>> >>>>>>> I let another POGS task load, and I'll let it run. If it is true to >>>>>>> form, I'll get an error, and I'll check tombstones again and let you >>>>>>> know. >>>>>>> >>>>>>> Iris >>>>>>> >>>>>>> >>>>>>> >>>>>>> On Wed, Jul 31, 2013 at 11:50 AM, Daniel Carrion < >>>>>>> [email protected]> wrote: >>>>>>> >>>>>>>> Hi Iris >>>>>>>> >>>>>>>> Thanks for responding. I had a look and noticed that a couple of >>>>>>>> your Asteroid tasks also exited with SIGSEGV. >>>>>>>> >>>>>>>> Bugger. I was hoping /data/tombstones would contain something but I >>>>>>>> guess that might just be for core applications? This was on CM10.1 >>>>>>>> where I >>>>>>>> found stuff in mine. >>>>>>>> >>>>>>>> I'm going to build an application around the BOINC wrapper to >>>>>>>> simulate these errors to find the best way of getting users to provide >>>>>>>> full >>>>>>>> dumps/traces for applications exiting like that. >>>>>>>> >>>>>>>> Regards >>>>>>>> >>>>>>>> Daniel >>>>>>>> >>>>>>>> On Thu, Aug 1, 2013 at 1:12 AM, Iris Ailin-Pyzik < >>>>>>>> [email protected]> wrote: >>>>>>>> >>>>>>>>> Daniel - >>>>>>>>> >>>>>>>>> I had those repeated errors on POGS, to the point where I quit >>>>>>>>> running those tasks. Asteroids ran with an occasional computation >>>>>>>>> error, >>>>>>>>> and I'd never had one on Einstein until yesterday. >>>>>>>>> >>>>>>>>> I just looked in tombstones (not sure if it is truly the right >>>>>>>>> directory or not) and found nothing at all there. >>>>>>>>> >>>>>>>>> I'm willing to help, how do we proceed? >>>>>>>>> >>>>>>>>> Iris >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>> On Wednesday, July 31, 2013 11:05:14 AM UTC-4, Daniel Carrion >>>>>>>>> wrote: >>>>>>>>>> >>>>>>>>>> Hello >>>>>>>>>> >>>>>>>>>> There seems to be a bug when running on BOINC for Android where >>>>>>>>>> POGS is exiting due to SIGILL and SIGSEGV errors. The problem is >>>>>>>>>> that this >>>>>>>>>> is not happening on all devices, including my Samsung Galaxy S3 and >>>>>>>>>> ODROID-U2. I'm hoping that the dumps in /data/tombstones will be >>>>>>>>>> able to >>>>>>>>>> tell me what was going on when the application exited. It's clearly >>>>>>>>>> memory >>>>>>>>>> related but it's not clear as to why. >>>>>>>>>> >>>>>>>>>> If there's anyone out there who's running POGS and getting these >>>>>>>>>> errors who is willing to assist in troubleshooting please contact me >>>>>>>>>> or >>>>>>>>>> jump over onto the forums at http://pogs.theskynet.org/pogs**. >>>>>>>>>> If I can find someone with a device experiencing this problem and >>>>>>>>>> willing >>>>>>>>>> to troubleshoot then I can hopefully get to the bottom of this. >>>>>>>>>> >>>>>>>>>> Regards >>>>>>>>>> >>>>>>>>>> Daniel >>>>>>>>>> >>>>>>>>>> -- >>>>>>>>> You received this message because you are subscribed to the Google >>>>>>>>> Groups "BOINC Android Testing" group. >>>>>>>>> To unsubscribe from this group and stop receiving emails from it, >>>>>>>>> send an email to >>>>>>>>> [email protected]. >>>>>>>>> To post to this group, send email to >>>>>>>>> [email protected]. >>>>>>>>> Visit this group at >>>>>>>>> http://groups.google.com/group/boinc-android-testing. >>>>>>>>> To view this discussion on the web visit >>>>>>>>> https://groups.google.com/d/msgid/boinc-android-testing/b4b479cd-0df1-4d31-a4fd-89a65ca673fa%40googlegroups.com >>>>>>>>> . >>>>>>>>> For more options, visit https://groups.google.com/groups/opt_out. >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>> >>>>>>>> >>>>>>> >>>>>> >>>>> >>>> >>> >> > _______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
