Updateding and building the latest code base.

amir.nathoo at yahoo.com amir.nathoo at yahoo.com
Fri Jan 15 13:31:26 EST 2016


Hi Art,
See my comments below. 

    On Friday, January 15, 2016 9:52 AM, "McGee, Art" <amcgee7 at jaguarlandrover.com> wrote:
 

 See, below.
Art McGeeInfotainment Engineer


Jaguar Land Rover North America, LLC1419 NW 14th Ave, Portland, Oregon, 97209
Jaguar.com  |  LandRover.com

On 15 January 2016 at 09:36, Andersson, Gunnar <gunnar.x.andersson at volvocars.com> wrote:

On Fri, 2016-01-15 at 09:10 -0800, McGee, Art wrote:
> I still continue to have this error. 
>
> amcgee7 at tizen-Precision-T5600:/media/amcgee7/Datapart/gdp/genivi-demo-
> platform/gdp-src-build$ bitbake genivi-demo-platform
> ERROR: Command execution failed: Traceback (most recent call last):
>   File "/media/amcgee7/Datapart/gdp/genivi-demo-
> platform/poky/bitbake/lib/bb/command.py", line 99, in runAsyncCommand
>     self.cooker.updateCache()
>   File "/media/amcgee7/Datapart/gdp/genivi-demo-
> platform/poky/bitbake/lib/bb/cooker.py", line 1383, in updateCache
>     self.add_filewatch(self.data.getVar("__base_depends"),
> self.configwatcher)
>   File "/media/amcgee7/Datapart/gdp/genivi-demo-
> platform/poky/bitbake/lib/bb/cooker.py", line 211, in add_filewatch
>     watcher.add_watch(f, self.watchmask, quiet=False)
>   File "/media/amcgee7/Datapart/gdp/genivi-demo-
> platform/poky/bitbake/lib/pyinotify.py", line 1977, in add_watch
>     raise WatchManagerError(err, ret_)
> WatchManagerError: add_watch: cannot watch
> /media/amcgee7/Datapart/gdp/genivi-demo-platform/gdp-src-
> build/conf/bblayers.conf WD=-1, Errno=No space left on device (ENOSPC)
>
>
> Summary: There was 1 ERROR message shown, returning a non-zero exit code.
>
>
> and I don't believe my files system is encrypted.


So what is /media/amcgee7/Datapart ?  

This is a separate partition mounted at boot. 
Is it removable media?  

Not Removable 
What is the file system being used?

ext4 
Or is this on a machine running a large number of Yocto builds from multiple
people?

Only I run the builds  been only working with GDP mostly. 
Did you check (just for elimination) that there truly is enough disk space

Yes not running out of disk space.  +170GB free
so that we know it is adding watch that causing the problem?

Yes,  the build is Adding watches and not cleaning them up to my knowlege.  I had 65535 and that only worked for one build and after that I've increased to 512k and 1024k and each time I get a bit more usage and then it stops.  I have to reboot otherwise to clean this up. 
[Amir] I was able to build GDP multiple times with the default Ubuntu max_user_watches (8192), but I was using a newer version of Yocto/Poky (1.7). I do think there is a bug in how the cooker is using the notify as mentioned here: http://patchwork.openembedded.org/patch/100515/ and here: http://patches.openembedded.org/patch/86831/
So unless you are able to upgrade your current version or merge these patches, you may have to reboot your machine after each build.Is this happening after fully completed builds or even when you try multiple times with the same code base?



If indeed there is a resource leak causing the max-watches to run out,rebooting might help.  I saw you mentioned it before, did you try?

HTH
- Gunnar

>
>
> Art McGee
> Infotainment Engineer
>
>
>
> Jaguar Land Rover North America, LLC
> 1419 NW 14th Ave, Portland, Oregon, 97209
> Jaguar.com  |  LandRover.com
>
>
> On 9 January 2016 at 18:18, Streif, Rudolf <rstreif at jaguarlandrover.com>
> wrote:
> > This error is most commonly caused when the file system is encrypted.
> > Ubuntu, and other distros, offer file system encryption during
> > installation. It is typically a good idea but not for file systems used
> > for Yocto Project build environments.
> > :rjs
> > On Jan 9, 2016 12:27, "Amir Nathoo via genivi-projects" <genivi-projects
> > @lists.genivi.org> wrote:
> > > Hi Art,
> > >
> > > It seems this issue in bitbake is being addressed, but not all the
> > > patches have been contributed yet; at least not in the version I am
> > > using.
> > > The BBCooker class, from what I understand is adding a watch for every
> > > single files in the BBPATH. So that needed to be optimized.
> > > I found 2 bug reports related to this issue: [1] is integrated already
> > > to Poky 1.8, but [2] isn't. 
> > >
> > > I think bumping your watch count should do it. But depending on what
> > > version of Poky you are using, you may still have to reboot your
> > > machine if you reach that limit again.
> > >
> > > The community does not seem to be addressing the core of this issue,
> > > which is how to cleanup the watches once a build is completed. But I
> > > don't know any better :-)
> > >
> > > Amir.
> > >
> > > [1] http://patches.openembedded.org/patch/86831/
> > > [2] [bitbake-devel,03/13] cooker: further limit inotify watches -
> > > Patchwork
> > >
> > >
> > >
> > > On Friday, January 8, 2016 11:41 AM, "McGee, Art" <amcgee7 at jaguarlandr
> > > over.com> wrote:
> > >
> > >
> > > I would really like to know how many are allocated.  But I did bump
> > > mine up.  I suspect that will be a good solution.
> > >
> > > Thanks,
> > >
> > > Art McGee
> > > Infotainment Engineer
> > >
> > >
> > >
> > > Jaguar Land Rover North America, LLC
> > > 1419 NW 14th Ave, Portland, Oregon, 97209
> > > Jaguar.com  |  LandRover.com
> > >
> > >
> > > On 7 January 2016 at 21:30, <amir.nathoo at yahoo.com> wrote:
> > > Hi Art,
> > >
> > > Glad it helped. 
> > > I can see the cooker.py in poky, calling add_watch but never rm_watch.
> > > I am using Poky 1.8 and there are additional log errors which is more
> > > helpful than what you got.
> > > if 'ENOSPC' in str(e):
> > >         providerlog.error("No space left on device or exceeds
> > > fs.inotify.max_user_watches?")
> > >         providerlog.error("To check max_user_watches: sysctl -n
> > > fs.inotify.max_user_watches.")
> > >         providerlog.error("To modify max_user_watches: sysctl -n -w
> > > fs.inotify.max_user_watches=<value>.")
> > >         providerlog.error("Root privilege is required to modify
> > > max_user_watches.")
> > >
> > > On my Ubuntu 14.04 64bits the max is set to 524288 which is probably
> > > why I don't see the problem. But this happened to be a while back on a
> > > different large project.
> > >
> > > Amir.
> > >
> > >
> > >
> > > On Thursday, January 7, 2016 12:53 PM, "McGee, Art" <amcgee7 at jaguarlan
> > > drover.com> wrote:
> > >
> > >
> > > Amir,
> > >
> > > Thanks,  This has helped.  I have to reboot my box to clear this up.
> > >  currently the max watches is 65535.  I suspect the build didn't
> > > cleanup properly.
> > >
> > > Thanks,
> > >
> > > Art McGee
> > > Infotainment Engineer
> > >
> > >
> > >
> > > Jaguar Land Rover North America, LLC
> > > 1419 NW 14th Ave, Portland, Oregon, 97209
> > > Jaguar.com  |  LandRover.com
> > >
> > >
> > > On 7 January 2016 at 09:05, <amir.nathoo at yahoo.com> wrote:
> > > Hi Art,
> > >
> > > It could also be a problem with pyinotify.py script or somehow you
> > > have reached your quota of watches on your system.
> > > Try the following: 
> > > $ sysctl -n  fs.inotify.max_user_watches (shows how much watches you
> > > are allowed) 
> > > $ sysctl -n  -w fs.inotify.max_user_watches <number_of_watches> (you
> > > can increase that number from above)
> > >
> > > Hope this helps.
> > > Amir.
> > >
> > >
> > > On Thursday, January 7, 2016 4:07 AM, Paul Sherwood <paul.sherwood at cod
> > > ethink.co.uk> wrote:
> > >
> > >
> > > Hi Art,
> > > On 2016-01-06 20:28, McGee, Art wrote:
> > > <snip>
> > > > WatchManagerError: add_watch: cannot watch
> > > > /media/amcgee7/Datapart/gdp/genivi2/gdp-src-build/conf/bblayers.conf
> > > > WD=-1, Errno=No space left on device (ENOSPC)
> > >
> > > looks like you may need to clear some space?
> > >
> > > _______________________________________________
> > > genivi-projects mailing list
> > > genivi-projects at lists.genivi.org
> > > https://lists.genivi.org/mailman/listinfo/genivi-projects
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > > _______________________________________________
> > > genivi-projects mailing list
> > > genivi-projects at lists.genivi.org
> > > https://lists.genivi.org/mailman/listinfo/genivi-projects
> > >
> _______________________________________________
> genivi-projects mailing list
> genivi-projects at lists.genivi.org
> https://lists.genivi.org/mailman/listinfo/genivi-projects


_______________________________________________
genivi-projects mailing list
genivi-projects at lists.genivi.org
https://lists.genivi.org/mailman/listinfo/genivi-projects


  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.genivi.org/pipermail/genivi-projects_lists.genivi.org/attachments/20160115/0a56d5e4/attachment.html>


More information about the genivi-projects mailing list