Re: [RFC PATCH 2/2] GitLab CI: crude mapping of PMM's scripts to jobs

[Thomas Huth]

On 07/02/2020 09.37, Thomas Huth wrote:
> On 03/02/2020 04.23, Cleber Rosa wrote:
>> This is a crude and straightforward mapping of Peter's
>> "remake-merge-builds" and "pull-buildtest" scripts.
>>
>> Some characteristics were removed for simplicity sake (but eventually
>> will), including:
>>  * number of simultaneous make jobs
>>  * make's synchronous output, not needed because of previous point
>>  * out-of-tree builds
>>
>> This covers the "x86-64 Linux with a variety of different build
>> configs"[1].  I've personally tested all of them, and only had
>> issues with the "notcg" job[2], but it seems to be a test specific
>> issue with the nested KVM I was using.
>>
>> [1] - https://wiki.qemu.org/Requirements/GatingCI#Current_Tests
>> [2] - https://paste.centos.org/view/1dd43a1c
>>
>> Signed-off-by: Cleber Rosa <address@hidden>
>> ---
>>  .gitlab-ci.yml | 116 +++++++++++++++++++++++++++++++++++++++++++++++++
>>  1 file changed, 116 insertions(+)
> 
> Thanks for doing this! The patch looks basically fine to me, but some
> comments below...
> 
>> diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
>> index d2c7d2198e..eb4077e2ab 100644
>> --- a/.gitlab-ci.yml
>> +++ b/.gitlab-ci.yml
>> @@ -2,6 +2,8 @@ include:
>>    - local: '/.gitlab-ci-edk2.yml'
>>  
>>  build-system1:
>> + rules:
>> + - if: '$CI_COMMIT_REF_NAME != "staging"'
>>   before_script: &before_scr_apt
>>   - apt-get update -qq
>>   - apt-get install -y -qq flex bison libglib2.0-dev libpixman-1-dev 
>> genisoimage
>> @@ -17,6 +19,8 @@ build-system1:
>>   - make -j2 check
>>  
>>  build-system2:
>> + rules:
>> + - if: '$CI_COMMIT_REF_NAME != "staging"'
>>   before_script:
>>    *before_scr_apt
>>   script:
>> @@ -31,6 +35,8 @@ build-system2:
>>   - make -j2 check
>>  
>>  build-disabled:
>> + rules:
>> + - if: '$CI_COMMIT_REF_NAME != "staging"'
>>   before_script:
>>    *before_scr_apt
>>   script:
>> @@ -47,6 +53,8 @@ build-disabled:
>>   - make -j2 check-qtest SPEED=slow
>>  
>>  build-tcg-disabled:
>> + rules:
>> + - if: '$CI_COMMIT_REF_NAME != "staging"'
>>   before_script:
>>    *before_scr_apt
>>   script:
>> @@ -67,6 +75,8 @@ build-tcg-disabled:
>>              248 250 254 255 256
>>  
>>  build-user:
>> + rules:
>> + - if: '$CI_COMMIT_REF_NAME != "staging"'
>>   before_script:
>>    *before_scr_apt
>>   script:
>> @@ -78,6 +88,8 @@ build-user:
>>   - make run-tcg-tests-i386-linux-user run-tcg-tests-x86_64-linux-user
>>  
>>  build-clang:
>> + rules:
>> + - if: '$CI_COMMIT_REF_NAME != "staging"'
>>   before_script:
>>    *before_scr_apt
>>   script:
>> @@ -92,6 +104,8 @@ build-clang:
>>   - make -j2 check
>>  
>>  build-tci:
>> + rules:
>> + - if: '$CI_COMMIT_REF_NAME != "staging"'
>>   before_script:
>>    *before_scr_apt
>>   script:
> 
> Question to Peter/Alex/Stefan/Howevermergespullreqsinthefuture:
> 
> Should the above jobs really be skipped for pull requests, or would it
> be ok to include them there, too? (in the latter case, the above changes
> could just be dropped)

At least most of the hunks could be dropped - we should likely keep the
one for the job that runs additional iotests (i.e.
"build-tcg-disabled"), since the block layers folks don't want to see
additional iotests as a blocker for pull requests (we've had a lengthy
discussion about this last summer...)

>> +
>> +ubuntu-18.04.3-x86_64-notcg:
[...]
>> Question for Peter: Would it be ok to drop this job and simply always
>> use the "build-tcg-disabled" job that is already available in
>> .gitlab-ci.yml ?

If we do not run "build-tcg-disabled" for PRs, then this job should not
be dropped, of course.

 Thomas