| Message ID | 20230816080300.5559-1-arisu.tachibana@miraclelinux.com (mailing list archive) |
|---|---|
| State | Changes Requested |
| Headers | show |
| Series | [isar-cip-core] .gitlab-ci: Create amd64 release variations for KernelCI | expand |
On 16.08.23 10:03, Arisu Tachibana wrote: > Force the creation of buster, bullseye and bookworm images > for KernelCI storage with GitLabCI > This images will be tested with the relative linux-cip > kernel version > ``` > buster with linux-4.19.y-cip > bullseye with kernel linux-5.10.y-cip > bookworm with kernel linux-6.1.y-cip > ``` > > Signed-off-by: Arisu Tachibana <arisu.tachibana@miraclelinux.com> > --- > .gitlab-ci.yml | 29 ++++++++++++++++++++++++++++- > 1 file changed, 28 insertions(+), 1 deletion(-) > > diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml > index 892ba92..183efcd 100644 > --- a/.gitlab-ci.yml > +++ b/.gitlab-ci.yml > @@ -101,12 +101,39 @@ build:qemu-amd64-base: > wic_targz: disable > targz: enable > > -build:qemu-amd64-base-kernelci: > +build:qemu-amd64-base-kernelci-buster: > extends: > - .build_base > variables: > target: qemu-amd64 > extension: kernelci > + release: buster > + use_rt: disable > + wic_targz: disable > + targz: enable > + deploy: disable > + deploy_kernelci: enable > + > +build:qemu-amd64-base-kernelci-bullseye: > + extends: > + - .build_base > + variables: > + target: qemu-amd64 > + extension: kernelci > + release: bullseye > + use_rt: disable > + wic_targz: disable > + targz: enable > + deploy: disable > + deploy_kernelci: enable > + > +build:qemu-amd64-base-kernelci-bookworm: > + extends: > + - .build_base > + variables: > + target: qemu-amd64 > + release: bookworm > + extension: kernelci > use_rt: disable > wic_targz: disable > targz: enable Happy to add that - just wondering if adding those two targets may allow dropping one or two others which are implicitly covered by this now. Did you check that as well? Jan
On 16.08.23 14:47, Jan Kiszka wrote: > On 16.08.23 10:03, Arisu Tachibana wrote: >> Force the creation of buster, bullseye and bookworm images >> for KernelCI storage with GitLabCI >> This images will be tested with the relative linux-cip >> kernel version >> ``` >> buster with linux-4.19.y-cip >> bullseye with kernel linux-5.10.y-cip >> bookworm with kernel linux-6.1.y-cip >> ``` >> >> Signed-off-by: Arisu Tachibana <arisu.tachibana@miraclelinux.com> >> --- >> .gitlab-ci.yml | 29 ++++++++++++++++++++++++++++- >> 1 file changed, 28 insertions(+), 1 deletion(-) >> >> diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml >> index 892ba92..183efcd 100644 >> --- a/.gitlab-ci.yml >> +++ b/.gitlab-ci.yml >> @@ -101,12 +101,39 @@ build:qemu-amd64-base: >> wic_targz: disable >> targz: enable >> >> -build:qemu-amd64-base-kernelci: >> +build:qemu-amd64-base-kernelci-buster: >> extends: >> - .build_base >> variables: >> target: qemu-amd64 >> extension: kernelci >> + release: buster >> + use_rt: disable >> + wic_targz: disable >> + targz: enable >> + deploy: disable >> + deploy_kernelci: enable >> + >> +build:qemu-amd64-base-kernelci-bullseye: >> + extends: >> + - .build_base >> + variables: >> + target: qemu-amd64 >> + extension: kernelci >> + release: bullseye >> + use_rt: disable >> + wic_targz: disable >> + targz: enable >> + deploy: disable >> + deploy_kernelci: enable >> + >> +build:qemu-amd64-base-kernelci-bookworm: >> + extends: >> + - .build_base >> + variables: >> + target: qemu-amd64 >> + release: bookworm >> + extension: kernelci >> use_rt: disable >> wic_targz: disable >> targz: enable > > Happy to add that - just wondering if adding those two targets may allow > dropping one or two others which are implicitly covered by this now. Did > you check that as well? > Oh, and why only amd64? There is no need for other archs in that combination? Not that I'm keen on exploding our matrix further... BTW, do we need the kernelci builds on every push, or should they be updated at a different pace? When are they consumed? Jan
> Happy to add that - just wondering if adding those two targets may allow > dropping one or two others which are implicitly covered by this now. Did > you check that as well? > sorry, I'm not sure of the meaning of your question. could you explain it better? what are you willing to drop?
> > Oh, and why only amd64? There is no need for other archs in that > combination? Not that I'm keen on exploding our matrix further... Currently I thought to work one arch at time and at first doing amd64, in the future I could also work on other arches > > BTW, do we need the kernelci builds on every push, or should they be > updated at a different pace? When are they consumed? every push works ok for me thanks, Arisu
On 16.08.23 16:40, Arisu Tachibana wrote: >> Happy to add that - just wondering if adding those two targets may allow >> dropping one or two others which are implicitly covered by this now. Did >> you check that as well? >> > > sorry, I'm not sure of the meaning of your question. > could you explain it better? what are you willing to drop? > Our CI tries to cover representative combinations of isar-cip-core's configuration variants. We can't cover them all - too many. Now, when we add these two, later on also for non-x86 archs, it is worth to think about if those obsolete other CI targets without creating a relevant test coverage gap. We currently have 33 build targets. Makes 33 runners that are started when even just a simple change to isar-cip-core is pushed. That can sum up to relevant costs if we just leave that running, adding more and more targets. Jan
On Sat, Aug 19, 2023 at 8:53 PM Jan Kiszka <jan.kiszka@siemens.com> wrote: > > On 16.08.23 16:40, Arisu Tachibana wrote: > >> Happy to add that - just wondering if adding those two targets may allow > >> dropping one or two others which are implicitly covered by this now. Did > >> you check that as well? > >> > > > > sorry, I'm not sure of the meaning of your question. > > could you explain it better? what are you willing to drop? > > > > Our CI tries to cover representative combinations of isar-cip-core's > configuration variants. We can't cover them all - too many. Now, when we > add these two, later on also for non-x86 archs, it is worth to think > about if those obsolete other CI targets without creating a relevant > test coverage gap. Can you specify what targets do you retain obsolete ? are you saying to skip amd64 and do only the non-x86 arch? > > We currently have 33 build targets. Makes 33 runners that are started > when even just a simple change to isar-cip-core is pushed. That can sum > up to relevant costs if we just leave that running, adding more and more > targets. During the CIP weekly meeting we decided to proceed with KernelCI testing of isar-cip-core in the following way (https://ircbot.wl.linuxfoundation.org/meetings/cip/2023/04/cip.2023-04-06-12.02.log.txt): ``` 12:19:26 <alicefm> sent mail about isar-cip-core on kernelci https://github.com/kernelci/kernelci-core/pull/1656 12:19:38 <alicefm> to cip-dev 12:20:14 <patersonc[m]> Thanks Arisu 12:20:59 <alicefm> about if we want to just move to bullseye or we want to keep also buster to be tested on kernelci 12:21:41 <patersonc[m]> In the gitlab setup, it depends on which kernel version is used 12:21:57 <patersonc[m]> To map with how Debian does it I believe 12:22:10 <patersonc[m]> Do we want to do something similar in kernelci? 12:22:23 <jki> yeah, likely to avoid surprises in userland 12:22:24 <patersonc[m]> Older kernels - buster. Newer - bullseye 12:22:31 <jki> exactly 12:23:16 <alicefm> how much older kernel? 12:23:36 <alicefm> 4.19? 4.4？ 12:23:43 <jki> 4.4 would likely be better done with buster, though it should work with bullseye as well 12:24:15 <patersonc[m]> Arisu: I need to check 12:24:16 <jki> doing 4.4 with bookworm one day could be risky, if possibly at all 12:24:36 <alicefm> ok thanks. ``` Debian is currently doing: ``` buster with linux-4.19.y-cip bullseye with kernel linux-5.10.y-cip bookworm with kernel linux-6.1.y-cip ``` Are you now suggesting to proceed in a different way ? Thanks, Arisu
On 22.08.23 05:38, Arisu Tachibana wrote: > On Sat, Aug 19, 2023 at 8:53 PM Jan Kiszka <jan.kiszka@siemens.com> wrote: >> >> On 16.08.23 16:40, Arisu Tachibana wrote: >>>> Happy to add that - just wondering if adding those two targets may allow >>>> dropping one or two others which are implicitly covered by this now. Did >>>> you check that as well? >>>> >>> >>> sorry, I'm not sure of the meaning of your question. >>> could you explain it better? what are you willing to drop? >>> >> >> Our CI tries to cover representative combinations of isar-cip-core's >> configuration variants. We can't cover them all - too many. Now, when we >> add these two, later on also for non-x86 archs, it is worth to think >> about if those obsolete other CI targets without creating a relevant >> test coverage gap. > > Can you specify what targets do you retain obsolete ? > are you saying to skip amd64 and do only the non-x86 arch? > I'm still scanning for optimization potential for the mentioned reasons, that's all. If you all tell me, we need all the variants, then we need to build them. But looking at our current CI set, there also the test images [1]. Is there a chance to consolidate them with the kernel CI images? Jan [1] https://gitlab.com/cip-project/cip-core/isar-cip-core/-/blob/master/.gitlab-ci.yml?ref_type=heads#L160
diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml index 892ba92..183efcd 100644 --- a/.gitlab-ci.yml +++ b/.gitlab-ci.yml @@ -101,12 +101,39 @@ build:qemu-amd64-base: wic_targz: disable targz: enable -build:qemu-amd64-base-kernelci: +build:qemu-amd64-base-kernelci-buster: extends: - .build_base variables: target: qemu-amd64 extension: kernelci + release: buster + use_rt: disable + wic_targz: disable + targz: enable + deploy: disable + deploy_kernelci: enable + +build:qemu-amd64-base-kernelci-bullseye: + extends: + - .build_base + variables: + target: qemu-amd64 + extension: kernelci + release: bullseye + use_rt: disable + wic_targz: disable + targz: enable + deploy: disable + deploy_kernelci: enable + +build:qemu-amd64-base-kernelci-bookworm: + extends: + - .build_base + variables: + target: qemu-amd64 + release: bookworm + extension: kernelci use_rt: disable wic_targz: disable targz: enable
Force the creation of buster, bullseye and bookworm images for KernelCI storage with GitLabCI This images will be tested with the relative linux-cip kernel version ``` buster with linux-4.19.y-cip bullseye with kernel linux-5.10.y-cip bookworm with kernel linux-6.1.y-cip ``` Signed-off-by: Arisu Tachibana <arisu.tachibana@miraclelinux.com> --- .gitlab-ci.yml | 29 ++++++++++++++++++++++++++++- 1 file changed, 28 insertions(+), 1 deletion(-)