pungi status

[Dennis Gilmore]

Hi all,

wanted to give you all a status update as to where we stand with pungi and 
Fedora 23.  the current blocker is 
https://bugzilla.redhat.com/show_bug.cgi?id=1208092  there is a bug in mock 
where it forcibly unmounts anything that is mounted in the chroot.  I have 
proposed it as a blocker for F23 Alpha.

leading up to this we hit an issue where the mounting was not happening 
correctly. it turns out that when you pass os.path.join two variables the 
first one is silently dropped. not sure if this is a bug in python or the way 
that os.path.join is supposed to work.  we have a workaround deployed in the 
runroot plugin.


Right now we are working in a non optional fashion but it is one that works 
with pungi. http://koji.fedoraproject.org/koji/taskinfo?taskID=10328143 is a 
failed runroot task if you look in the mock log you can see the mount being 
forcibly unmounted. 

we could force the use of bind mounts only in runroot and add the mounts to 
the mock config, which would be an additional step to take and one that goes 
against what was intended and limits the options  unless we then mount them 
elsewhere and bind mount.  it is kind of messy. 

We have branching on Tuesday and are just over two weeks from Alpha change 
freeze, our window to get pungi working is rapidly shrinking. we do have other 
deliverables to sort out yet also

Regards

Dennis

[Mathieu Bridon]

-- 
Mathieu
On Thu, 2015-07-09 at 20:54 -0500, Dennis Gilmore wrote:
>  leading up to this we hit an issue where the mounting was not
>  happening correctly. it turns out that when you pass os.path.join two 
>  variables the first one is silently dropped. not sure if this is a 
>  bug in python or the way that os.path.join is supposed to work. 
Do you mean this?

> >> os.path.join('/foo', '/bar') ==
> '/bar' 
    True

If this is what you mean, then it indeed is the expected behaviour:

>  If a component is an absolute path, all previous components are
>  thrown away and joining continues from the absolute path component. 
    https://docs.python.org/2/library/os.path.html#os.path.join


-- 
Mathieu

[Dennis Gilmore]

On Friday, July 10, 2015 09:58:59 AM Mathieu Bridon wrote:
>  > leading up to this we hit an issue where the mounting was not
>  > happening correctly. it turns out that when you pass os.path.join two
>  > variables the first one is silently dropped. not sure if this is a
>  > bug in python or the way that os.path.join is supposed to work.
>  
>  Do you mean this?
>  
>      >>> os.path.join('/foo', '/bar') == '/bar'
>  
>      True
>  
>  If this is what you mean, then it indeed is the expected behaviour:
>  > If a component is an absolute path, all previous components are
>  > thrown away and joining continues from the absolute path component.
>  
>      https://docs.python.org/2/library/os.path.html#os.path.join 
This is the case, so we will have to get a patch into koji to not use 
os.path.join 

Dennis

[Mathieu Bridon]

On Fri, 2015-07-10 at 09:03 -0500, Dennis Gilmore wrote:
>  On Friday, July 10, 2015 09:58:59 AM Mathieu Bridon wrote:
>  > > leading up to this we hit an issue where the mounting was not
>  > > happening correctly. it turns out that when you pass os.path.join 
>  > > two
>  > > variables the first one is silently dropped. not sure if this is 
>  > > a
>  > > bug in python or the way that os.path.join is supposed to work.
>  > 
>  > Do you mean this?
>  > 
>  >     >>> os.path.join('/foo', '/bar') == '/bar'
>  > 
>  >     True
>  > 
>  > If this is what you mean, then it indeed is the expected behaviour:
>  > > If a component is an absolute path, all previous components are
>  > > thrown away and joining continues from the absolute path 
>  > > component.
>  > 
>  >     https://docs.python.org/2/library/os.path.html#os.path.join
>  
>  This is the case, so we will have to get a patch into koji to not us
>  os.path.join  
Or rather, only pass it absolute paths as the first argument.

What code in Koji does that?


-- 
Mathieu

[Adam Miller]

On Fri, Jul 10, 2015 at 9:58 AM, Mathieu Bridon <bochecha(a)daitauha.fr&gt; wrote:
>  On Fri, 2015-07-10 at 09:03 -0500, Dennis Gilmore wrote:
> > On Friday, July 10, 2015 09:58:59 AM Mathieu Bridon wrote:
> > > > leading up to this we hit an issue where the mounting was not
> > > > happening correctly. it turns out that when you pass os.path.join
> > > > two
> > > > variables the first one is silently dropped. not sure if this is
> > > > a
> > > > bug in python or the way that os.path.join is supposed to work.
> > >
> > > Do you mean this?
> > >
> > >     >>> os.path.join('/foo', '/bar') == '/bar'
> > >
> > >     True
> > >
> > > If this is what you mean, then it indeed is the expected behaviour:
> > > > If a component is an absolute path, all previous components are
> > > > thrown away and joining continues from the absolute path
> > > > component.
> > >
> > >     https://docs.python.org/2/library/os.path.html#os.path.join
> >
> > This is the case, so we will have to get a patch into koji to not us
> > os.path.join
>
>  Or rather, only pass it absolute paths as the first argument.
>
>  What code in Koji does that? 
This is specific to the runroot plugin, the code is in
./plugins/runroot.py (relative path at root of git clone dir).

-AdamM

>  --
>  Mathieu
>
>  _______________________________________________
>  rel-eng mailing list
>  rel-eng(a)lists.fedoraproject.org
>  https://admin.fedoraproject.org/mailman/listinfo/rel-eng