[OE-core] [PATCH V3] dnf.py: installroot support usrmerge

[Randy MacLeod]

On 9/4/19 4:36 AM, Changqing Li wrote:
>
> On 9/4/19 4:15 PM, Richard Purdie wrote:
>> On Wed, 2019-09-04 at 16:04 +0800, Changqing Li wrote:
>>> Ping
>>> Test with/without  usrmege feature passed.
>>> With usrmerge:
>>> NOTE: test_dnf_installroot (dnf.DnfRepoTest)
>>> DEBUG: Checking if 'DISTRO_FEATURES' value contains 'usrmerge' to
>>> skip the test
>>> NOTE:  ... skipped 'Test run when not enable usrmerge'
>>> Test run when not enable usrmerge
>>> NOTE: test_dnf_installroot_usrmerge (dnf.DnfRepoTest)
>>> DEBUG: Checking if 'DISTRO_FEATURES' value contains 'usrmerge' to run
>>> the test
>>> DEBUG: [Running]$ ssh -l root -o UserKnownHostsFile=/dev/null -o
>>> StrictHostKeyChecking=no -o LogLevel=ERROR 192.168.7.16 export
>>> PATH=/usr/sbin:/sbin:/usr/bin:/bin; mkdir -p
>>> /home/root/chroot/test/etc
>>> ....
>>>
>>> Without usrmerge:
>>> NOTE: test_dnf_installroot (dnf.DnfRepoTest)
>>> DEBUG: Checking if 'DISTRO_FEATURES' value contains 'usrmerge' to
>>> skip the test
>>> DEBUG: [Running]$ ssh -l root -o UserKnownHostsFile=/dev/null -o
>>> StrictHostKeyChecking=no -o LogLevel=ERROR 192.168.7.10 export
>>> PATH=/usr/sbin:/sbin:/usr/bin:/bin; mkdir -p
>>> /home/root/chroot/test/etc
>>> DEBUG: time: 1567404974.8490767, endtime: 1567406474.8424258
>>> DEBUG: [Command returned '0' after 0.13 seconds]
>>> DEBUG: Command: mkdir -p /home/root/chroot/test/etc
>>> Output:
>>> ...
>>> NOTE:  ... ok
>>> NOTE: test_dnf_installroot_usrmerge (dnf.DnfRepoTest)
>>> DEBUG: Checking if 'DISTRO_FEATURES' value contains 'usrmerge' to run
>>> the test
>>> NOTE:  ... skipped 'Test run when enable usrmege'
>>> Test run when enable usrmege
>> In a previous version of this patch, it introduced a failure in the
>> parselogs test as an error was generated into the logs. I haven't
>> tested the new version of the patch, was that issue addressed?
>>
>> Cheers,
>>
>> Richard
>
> I didn't met failure in test_parselog during test,  the whole 
> testimage is passed.
>
> Could you help to send me the link of the failure test if the new 
> version still
>
> met test_parselog failure? Thanks.

Sandy,

Richard shouldn't have to do that since many people have such requests and

if he helped everyone he would (and has in the past) spend/t too much 
time helping people.

We should be able to setup the autobuilder locally to reproduce the problem.


Also during the YP bug review call, someone suggested that you check if the

dnf install tests are running in a deterministic order. If they are not, 
then that

might explain why the AB sees the problem but you do not.


../Randy

>
>>
>>
>>

-- 
# Randy MacLeod
# Wind River Linux