Message ID | 20231102072030.1800796-1-liwang@redhat.com |
---|---|
State | Accepted |
Headers | show |
Series | ci: add centos stream support | expand |
Hi Li, > This is base on Peter's patch: > CI: ReAplace CentOS 7 with openSUSE Leap 42.2, Ubuntu xenial > Ci: https://github.com/wangli5665/ltp/actions/runs/6729379565 Reviewed-by: Petr Vorel <pvorel@suse.cz> ATM we have 18 jobs, IMHO we should not add more without removing others just to have reasonable CI runtime. Kind regards, Petr
Hi Petr, On Thu, Nov 2, 2023 at 5:21 PM Petr Vorel <pvorel@suse.cz> wrote: > Hi Li, > > > This is base on Peter's patch: > > CI: ReAplace CentOS 7 with openSUSE Leap 42.2, Ubuntu xenial > > > Ci: https://github.com/wangli5665/ltp/actions/runs/6729379565 > > Reviewed-by: Petr Vorel <pvorel@suse.cz> > Thanks! > ATM we have 18 jobs, IMHO we should not add more without removing others > just to > have reasonable CI runtime. > Your concern makes sense, but I have no idea which one could be removed.
> Hi Petr, > On Thu, Nov 2, 2023 at 5:21 PM Petr Vorel <pvorel@suse.cz> wrote: > > Hi Li, > > > This is base on Peter's patch: > > > CI: ReAplace CentOS 7 with openSUSE Leap 42.2, Ubuntu xenial > > > Ci: https://github.com/wangli5665/ltp/actions/runs/6729379565 > > Reviewed-by: Petr Vorel <pvorel@suse.cz> > Thanks! > > ATM we have 18 jobs, IMHO we should not add more without removing others > > just to > > have reasonable CI runtime. > Your concern makes sense, but I have no idea which one could be removed. I wonder if just adding CentOS 9 would be enough. Because in CI we test just compilation. The point of the public CI is to test the oldest distros, newest distros and something in between, that should cover the most of compilation problems (no need to test *all* used distros to be safe). But sure, both can stay if you want. The runtime is still reasonable. NOTE: I did not care much in case of iputils, which have much faster build than LTP. If we want to add both CentOS 8 and 9, we could remove some of debian:oldstable (bullseye) (similar kernel). And, we could add just openSUSE Leap 42.2, which has older glibc and gcc than Ubuntu 16.04 LTS xenial. For sure there are more people which use xenial than old Leap 42.2, but as I said, the goal is to test the oldest distro we maintainers really care about (and do not have extra work for nothing, which is IMHO keeping CentOS 7 compilable). WDYT? Kind regards, Petr
Hi Petr, On Fri, Nov 3, 2023 at 5:14 PM Petr Vorel <pvorel@suse.cz> wrote: > > Hi Petr, > > > On Thu, Nov 2, 2023 at 5:21 PM Petr Vorel <pvorel@suse.cz> wrote: > > > > Hi Li, > > > > > This is base on Peter's patch: > > > > CI: ReAplace CentOS 7 with openSUSE Leap 42.2, Ubuntu xenial > > > > > Ci: https://github.com/wangli5665/ltp/actions/runs/6729379565 > > > > Reviewed-by: Petr Vorel <pvorel@suse.cz> > > > > Thanks! > > > > > ATM we have 18 jobs, IMHO we should not add more without removing > others > > > just to > > > have reasonable CI runtime. > > > > Your concern makes sense, but I have no idea which one could be removed. > > I wonder if just adding CentOS 9 would be enough. Because in CI we test > just > compilation. The point of the public CI is to test the oldest distros, > newest > distros and something in between, that should cover the most of compilation > problems (no need to test *all* used distros to be safe). > But sure, both can stay if you want. The runtime is still reasonable. > NOTE: I did not care much in case of iputils, which have much faster build > than > LTP. > > If we want to add both CentOS 8 and 9, we could remove some of > debian:oldstable (bullseye) > (similar kernel). > > And, we could add just openSUSE Leap 42.2, which has older glibc and gcc > than > Ubuntu 16.04 LTS xenial. For sure there are more people which use xenial > than > old Leap 42.2, but as I said, the goal is to test the oldest distro we > maintainers really care about (and do not have extra work for nothing, > which is > IMHO keeping CentOS 7 compilable). > > WDYT? > Your explanation is quite useful, thanks! I'm fine to remove the CentOS8 build in this patch, should I resent the new patch or push it directly?
> Hi Petr, > On Fri, Nov 3, 2023 at 5:14 PM Petr Vorel <pvorel@suse.cz> wrote: > > > Hi Petr, > > > On Thu, Nov 2, 2023 at 5:21 PM Petr Vorel <pvorel@suse.cz> wrote: > > > > Hi Li, > > > > > This is base on Peter's patch: > > > > > CI: ReAplace CentOS 7 with openSUSE Leap 42.2, Ubuntu xenial > > > > > Ci: https://github.com/wangli5665/ltp/actions/runs/6729379565 > > > > Reviewed-by: Petr Vorel <pvorel@suse.cz> > > > Thanks! > > > > ATM we have 18 jobs, IMHO we should not add more without removing > > others > > > > just to > > > > have reasonable CI runtime. > > > Your concern makes sense, but I have no idea which one could be removed. > > I wonder if just adding CentOS 9 would be enough. Because in CI we test > > just > > compilation. The point of the public CI is to test the oldest distros, > > newest > > distros and something in between, that should cover the most of compilation > > problems (no need to test *all* used distros to be safe). > > But sure, both can stay if you want. The runtime is still reasonable. > > NOTE: I did not care much in case of iputils, which have much faster build > > than > > LTP. > > If we want to add both CentOS 8 and 9, we could remove some of > > debian:oldstable (bullseye) > > (similar kernel). > > And, we could add just openSUSE Leap 42.2, which has older glibc and gcc > > than > > Ubuntu 16.04 LTS xenial. For sure there are more people which use xenial > > than > > old Leap 42.2, but as I said, the goal is to test the oldest distro we > > maintainers really care about (and do not have extra work for nothing, > > which is > > IMHO keeping CentOS 7 compilable). > > WDYT? > Your explanation is quite useful, thanks! > I'm fine to remove the CentOS8 build in this patch, > should I resent the new patch or push it directly? You can push it directly. If it does not apply, feel free to push my change first. I'll post a note about keeping only openSUSE Leap 42.2 in the patchset. Acked-by: Petr Vorel <pvorel@suse.cz> Kind regards, Petr
Petr Vorel <pvorel@suse.cz> wrote: > > > > If we want to add both CentOS 8 and 9, we could remove some of > > > debian:oldstable (bullseye) > > > (similar kernel). > > > > And, we could add just openSUSE Leap 42.2, which has older glibc and > gcc > > > than > > > Ubuntu 16.04 LTS xenial. For sure there are more people which use > xenial > > > than > > > old Leap 42.2, but as I said, the goal is to test the oldest distro we > > > maintainers really care about (and do not have extra work for nothing, > > > which is > > > IMHO keeping CentOS 7 compilable). > > > > WDYT? > > > > Your explanation is quite useful, thanks! > > > I'm fine to remove the CentOS8 build in this patch, > > should I resent the new patch or push it directly? > > You can push it directly. If it does not apply, feel free to push my change > first. I'll post a note about keeping only openSUSE Leap 42.2 in the > patchset. > Done, and sorry I just merged the V2 then see you post V3 patches.
> Petr Vorel <pvorel@suse.cz> wrote: > > > > If we want to add both CentOS 8 and 9, we could remove some of > > > > debian:oldstable (bullseye) > > > > (similar kernel). > > > > And, we could add just openSUSE Leap 42.2, which has older glibc and > > gcc > > > > than > > > > Ubuntu 16.04 LTS xenial. For sure there are more people which use > > xenial > > > > than > > > > old Leap 42.2, but as I said, the goal is to test the oldest distro we > > > > maintainers really care about (and do not have extra work for nothing, > > > > which is > > > > IMHO keeping CentOS 7 compilable). > > > > WDYT? > > > Your explanation is quite useful, thanks! > > > I'm fine to remove the CentOS8 build in this patch, > > > should I resent the new patch or push it directly? > > You can push it directly. If it does not apply, feel free to push my change > > first. I'll post a note about keeping only openSUSE Leap 42.2 in the > > patchset. > Done, and sorry I just merged the V2 then see you post V3 patches. No problem, thanks for merging. I'll send a patch which removes xenial. Kind regards, Petr
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml index 9a8a66b6d..27cc5fd72 100644 --- a/.github/workflows/ci.yml +++ b/.github/workflows/ci.yml @@ -69,6 +69,18 @@ jobs: METADATA: asciidoc-pdf TREE: out + - container: "quay.io/centos/centos:stream8" + env: + CC: gcc + METADATA: asciidoc-pdf + TREE: out + + - container: "quay.io/centos/centos:stream9" + env: + CC: gcc + METADATA: asciidoc-pdf + TREE: out + - container: "debian:testing" env: CC: gcc diff --git a/ci/centos.sh b/ci/quay.io.sh similarity index 100% rename from ci/centos.sh rename to ci/quay.io.sh
This is base on Peter's patch: CI: ReAplace CentOS 7 with openSUSE Leap 42.2, Ubuntu xenial Ci: https://github.com/wangli5665/ltp/actions/runs/6729379565 Signed-off-by: Li Wang <liwang@redhat.com> Cc: Chao Ye <cye@redhat.com> Cc: Jan Stancek <jstancek@redhat.com> Cc: Petr Vorel <pvorel@suse.cz> --- .github/workflows/ci.yml | 12 ++++++++++++ ci/{centos.sh => quay.io.sh} | 0 2 files changed, 12 insertions(+) rename ci/{centos.sh => quay.io.sh} (100%)