Re: [OE-core] [PATCH 1/4] recipetool: Set master branch only as fallback

[Richard Purdie <richard.purdie@linuxfoundation.org>]

On Wed, 2021-12-08 at 08:45 +0100, Stefan Herbrechtsmeier wrote:
> Hi Richard,
> 
> Am 07.12.2021 um 16:45 schrieb Richard Purdie:
> > On Mon, 2021-12-06 at 09:04 +0100, Stefan Herbrechtsmeier wrote:
> > > From: Stefan Herbrechtsmeier <stefan.herbrechtsmeier@weidmueller.com>
> > > 
> > > The commit 'meta/scripts: Manual git url branch additions (dc53fe75cc)'
> > > sets the branch= parameter too early to master and thereby breaks the
> > > -B/--srcbranch option.
> > > 
> > > ERROR: branch= parameter and -B/--srcbranch option cannot both be specified - use one or the other
> > > 
> > > Signed-off-by: Stefan Herbrechtsmeier <stefan.herbrechtsmeier@weidmueller.com>
> > > ---
> > > 
> > >   scripts/lib/recipetool/create.py | 15 +++++++--------
> > >   1 file changed, 7 insertions(+), 8 deletions(-)
> > 
> > I think something in this series is causing:
> > 
> > https://autobuilder.yoctoproject.org/typhoon/#/builders/79/builds/2926/steps/14/logs/stdio
> 
> Sorry I missed the adaption of the test test_devtool_add_fetch_git. It 
> doesn't expect a "branch=master" which is wrong.
> 
> - gitsm://git.yoctoproject.org/mraa;branch=master
> ?                                  --------------
> + gitsm://git.yoctoproject.org/mraa

I suspected something like that, thanks.

> I will update my patch series but need some time to test it locally. The 
> devtool test build a lot of recipe on every run because it use a private 
> sstate cache. How does the autobuilder overcome this problem?

The autobuilder doesn't. We stopped devtool using the public shared cache as it
is shared between autobuilder workers and one worker could delete artefacts
whilst another is using them. This caused all kinds of weird races and there
isn't much extra build time from the autobuilder perspective. The autobuilder
never deletes things from the shared cache outside of the timeout cron job which
removes objects that haven't been accessed in a month or so.

> > 
> > (the other selftest builds also failed)
> What do you mean by this?

I mean that we run oe-selftest on 4 different distros and they all failed the
same way so there were four failures in the build and I just pasted a link to
one of them.

Cheers,

Richard