[PATCH libnbd CI] ci: Drop Fedora 33, add Fedora 35
Hi Martin, The pipeline failed, and I'm assuming this is because Fedora 33 is out of support. The log is not very helpful: https://gitlab.com/nbdkit/libnbd/-/jobs/1874368914 Also we don't seem to have lcitool in Fedora anywhere obvious, so instead I just updated the generated files directly. Also, wouldn't it be nice if we could specify "fedora:n" and "fedora:n-1" instead of having to update this every six months ... Rich.
Failed pipeline: https://gitlab.com/nbdkit/libnbd/-/jobs/1874368914 --- ...ora-33.Dockerfile => fedora-35.Dockerfile} | 2 +- ci/gitlab.yml | 32 +++++++++---------- ci/manifest.yml | 4 +-- 3 files changed, 19 insertions(+), 19 deletions(-) diff --git a/ci/containers/fedora-33.Dockerfile b/ci/containers/fedora-35.Dockerfile similarity index 97% rename from ci/containers/fedora-33.Dockerfile rename to ci/containers/fedora-35.Dockerfile index 9c0bfa954..bc63cc845 100644 --- a/ci/containers/fedora-33.Dockerfile +++ b/ci/containers/fedora-35.Dockerfile @@ -4,7 +4,7 @@ # # https://gitlab.com/libvirt/libvirt-ci -FROM registry.fedoraproject.org/fedora:33 +FROM registry.fedoraproject.org/fedora:35 RUN dnf install -y nosync && \ echo -e '#!/bin/sh\n\ diff --git a/ci/gitlab.yml b/ci/gitlab.yml index df3f68ad5..2e0d0f38c 100644 --- a/ci/gitlab.yml +++ b/ci/gitlab.yml @@ -116,13 +116,6 @@ x86_64-debian-sid-container: NAME: debian-sid -x86_64-fedora-33-container: - extends: .container_job - allow_failure: false - variables: - NAME: fedora-33 - - x86_64-fedora-34-container: extends: .container_job allow_failure: false @@ -130,6 +123,13 @@ x86_64-fedora-34-container: NAME: fedora-34 +x86_64-fedora-35-container: + extends: .container_job + allow_failure: false + variables: + NAME: fedora-35 + + x86_64-fedora-rawhide-container: extends: .container_job allow_failure: false @@ -257,15 +257,6 @@ x86_64-debian-sid: NAME: debian-sid -x86_64-fedora-33: - extends: .native_build_job - needs: - - x86_64-fedora-33-container - allow_failure: false - variables: - NAME: fedora-33 - - x86_64-fedora-34: extends: .native_build_job needs: @@ -275,6 +266,15 @@ x86_64-fedora-34: NAME: fedora-34 +x86_64-fedora-35: + extends: .native_build_job + needs: + - x86_64-fedora-35-container + allow_failure: false + variables: + NAME: fedora-35 + + x86_64-fedora-rawhide: extends: .native_build_job needs: diff --git a/ci/manifest.yml b/ci/manifest.yml index 704d4fbb9..d487dda6b 100644 --- a/ci/manifest.yml +++ b/ci/manifest.yml @@ -35,10 +35,10 @@ targets: debian-sid: x86_64 - fedora-33: x86_64 - fedora-34: x86_64 + fedora-35: x86_64 + fedora-rawhide: jobs: - arch: x86_64 -- 2.32.0
On 12/11/21 15:49, Richard W.M. Jones wrote:
Failed pipeline: https://gitlab.com/nbdkit/libnbd/-/jobs/1874368914 --- ...ora-33.Dockerfile => fedora-35.Dockerfile} | 2 +- ci/gitlab.yml | 32 +++++++++---------- ci/manifest.yml | 4 +-- 3 files changed, 19 insertions(+), 19 deletions(-)
diff --git a/ci/containers/fedora-33.Dockerfile b/ci/containers/fedora-35.Dockerfile similarity index 97% rename from ci/containers/fedora-33.Dockerfile rename to ci/containers/fedora-35.Dockerfile index 9c0bfa954..bc63cc845 100644 --- a/ci/containers/fedora-33.Dockerfile +++ b/ci/containers/fedora-35.Dockerfile @@ -4,7 +4,7 @@ # # https://gitlab.com/libvirt/libvirt-ci
-FROM registry.fedoraproject.org/fedora:33 +FROM registry.fedoraproject.org/fedora:35
RUN dnf install -y nosync && \ echo -e '#!/bin/sh\n\ diff --git a/ci/gitlab.yml b/ci/gitlab.yml index df3f68ad5..2e0d0f38c 100644 --- a/ci/gitlab.yml +++ b/ci/gitlab.yml @@ -116,13 +116,6 @@ x86_64-debian-sid-container: NAME: debian-sid
-x86_64-fedora-33-container: - extends: .container_job - allow_failure: false - variables: - NAME: fedora-33 - - x86_64-fedora-34-container: extends: .container_job allow_failure: false @@ -130,6 +123,13 @@ x86_64-fedora-34-container: NAME: fedora-34
+x86_64-fedora-35-container: + extends: .container_job + allow_failure: false + variables: + NAME: fedora-35 + + x86_64-fedora-rawhide-container: extends: .container_job allow_failure: false @@ -257,15 +257,6 @@ x86_64-debian-sid: NAME: debian-sid
-x86_64-fedora-33: - extends: .native_build_job - needs: - - x86_64-fedora-33-container - allow_failure: false - variables: - NAME: fedora-33 - - x86_64-fedora-34: extends: .native_build_job needs: @@ -275,6 +266,15 @@ x86_64-fedora-34: NAME: fedora-34
+x86_64-fedora-35: + extends: .native_build_job + needs: + - x86_64-fedora-35-container + allow_failure: false + variables: + NAME: fedora-35 + + x86_64-fedora-rawhide: extends: .native_build_job needs: diff --git a/ci/manifest.yml b/ci/manifest.yml index 704d4fbb9..d487dda6b 100644 --- a/ci/manifest.yml +++ b/ci/manifest.yml @@ -35,10 +35,10 @@ targets:
debian-sid: x86_64
- fedora-33: x86_64 - fedora-34: x86_64
+ fedora-35: x86_64 + fedora-rawhide: jobs: - arch: x86_64
Acked-by: Laszlo Ersek <lersek@redhat.com>
I pushed this (commit 202d0ecad) in the hope it might fix the tests. However Martin indicates that the error message might be nothing to do with the Fedora version. Let's see. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming and virtualization blog: http://rwmj.wordpress.com virt-top is 'top' for virtual machines. Tiny program with many powerful monitoring features, net stats, disk stats, logging, etc. http://people.redhat.com/~rjones/virt-top
On Mon, Dec 13, 2021 at 02:40:40PM +0000, Richard W.M. Jones wrote:
I pushed this (commit 202d0ecad) in the hope it might fix the tests. However Martin indicates that the error message might be nothing to do with the Fedora version. Let's see.
It looks like everything went swimmingly the next time around even before this patch: https://gitlab.com/nbdkit/libnbd/-/pipelines/428406657 so it clear wasn't anything with Fedora 33, although updating to F35 is definitely worth doing. It did not happen to be before here with the containers for libnbd. I guess it can happen any time if there is an issue on the remote side when pushing a container for example. However I am facing few issues with nbdkit CI that I am still trying to finish and on and off working on it. The issue there is that the tests time out because of all the qemu invocations, especially with TCG. I am trying to think about what tests to skip and maybe split the load between different distros. That's just something I wanted to mention because it still keeps me up at nights while I cannot get out of the loop of finding out why test gets skipped, then either fixing the test or figuring out the missing package name for all distros to properly update libvirt-ci's mapping.yaml, and then looking at what next tests are failing or getting skipped and why.
Rich.
-- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming and virtualization blog: http://rwmj.wordpress.com virt-top is 'top' for virtual machines. Tiny program with many powerful monitoring features, net stats, disk stats, logging, etc. http://people.redhat.com/~rjones/virt-top
On Mon, Dec 13, 2021 at 04:07:27PM +0100, Martin Kletzander wrote:
On Mon, Dec 13, 2021 at 02:40:40PM +0000, Richard W.M. Jones wrote:
I pushed this (commit 202d0ecad) in the hope it might fix the tests. However Martin indicates that the error message might be nothing to do with the Fedora version. Let's see.
It looks like everything went swimmingly the next time around even before this patch:
Yes, it does look like a temporary docker pull problem.
so it clear wasn't anything with Fedora 33, although updating to F35 is definitely worth doing.
Yup.
It did not happen to be before here with the containers for libnbd. I guess it can happen any time if there is an issue on the remote side when pushing a container for example.
However I am facing few issues with nbdkit CI that I am still trying to finish and on and off working on it. The issue there is that the tests time out because of all the qemu invocations, especially with TCG. I am trying to think about what tests to skip and maybe split the load between different distros.
So you can disable the libguestfs-based tests if TCG is too slow: https://src.fedoraproject.org/rpms/nbdkit/blob/rawhide/f/nbdkit.spec#_695 We probably ought to formalise that into a configure flag, but at the moment the above trick should be reliable because we use the same in Fedora builds on slow arches. Also re nbdkit CI, it turns out Windows builds were broken for a while (I fixed it a few days ago). Would be nice to CI this! https://gitlab.com/nbdkit/nbdkit/-/issues/3
That's just something I wanted to mention because it still keeps me up at nights while I cannot get out of the loop of finding out why test gets skipped, then either fixing the test or figuring out the missing package name for all distros to properly update libvirt-ci's mapping.yaml, and then looking at what next tests are failing or getting skipped and why.
:-) Thanks, Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming and virtualization blog: http://rwmj.wordpress.com virt-df lists disk usage of guests without needing to install any software inside the virtual machine. Supports Linux and Windows. http://people.redhat.com/~rjones/virt-df/
On Mon, Dec 13, 2021 at 03:17:56PM +0000, Richard W.M. Jones wrote:
On Mon, Dec 13, 2021 at 04:07:27PM +0100, Martin Kletzander wrote:
On Mon, Dec 13, 2021 at 02:40:40PM +0000, Richard W.M. Jones wrote:
I pushed this (commit 202d0ecad) in the hope it might fix the tests. However Martin indicates that the error message might be nothing to do with the Fedora version. Let's see.
It looks like everything went swimmingly the next time around even before this patch:
Yes, it does look like a temporary docker pull problem.
so it clear wasn't anything with Fedora 33, although updating to F35 is definitely worth doing.
Yup.
It did not happen to be before here with the containers for libnbd. I guess it can happen any time if there is an issue on the remote side when pushing a container for example.
However I am facing few issues with nbdkit CI that I am still trying to finish and on and off working on it. The issue there is that the tests time out because of all the qemu invocations, especially with TCG. I am trying to think about what tests to skip and maybe split the load between different distros.
So you can disable the libguestfs-based tests if TCG is too slow:
https://src.fedoraproject.org/rpms/nbdkit/blob/rawhide/f/nbdkit.spec#_695
We probably ought to formalise that into a configure flag, but at the moment the above trick should be reliable because we use the same in Fedora builds on slow arches.
Also re nbdkit CI, it turns out Windows builds were broken for a while (I fixed it a few days ago). Would be nice to CI this!
Whuen using libvirt-ci manifest to generate gitlab config and dockerfiles, it is pretty trivial to include mingw builds in the matrix. Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
On Mon, Dec 13, 2021 at 03:17:56PM +0000, Richard W.M. Jones wrote:
So you can disable the libguestfs-based tests if TCG is too slow:
https://src.fedoraproject.org/rpms/nbdkit/blob/rawhide/f/nbdkit.spec#_695
We probably ought to formalise that into a configure flag, but at the moment the above trick should be reliable because we use the same in Fedora builds on slow arches.
https://gitlab.com/nbdkit/nbdkit/-/commit/c09ae98ff3b4b786565de4aa173274531a Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming and virtualization blog: http://rwmj.wordpress.com virt-builder quickly builds VMs from scratch http://libguestfs.org/virt-builder.1.html
On Sat, Dec 11, 2021 at 02:49:29PM +0000, Richard W.M. Jones wrote:
Hi Martin,
The pipeline failed, and I'm assuming this is because Fedora 33 is out of support. The log is not very helpful:
That should not have anything related to it.
It seems that "docker push" timed out, there aren't many logs to look at.
Also we don't seem to have lcitool in Fedora anywhere obvious, so instead I just updated the generated files directly.
It is not shipped anywhere because it wouldn't make sense to use anything older than current master, just clone it from: https://gitlab.com/libvirt/libvirt-ci/
Also, wouldn't it be nice if we could specify "fedora:n" and "fedora:n-1" instead of having to update this every six months ...
Good idea, but I think it would be more difficult than changing one line and running a script. But don't hesitate to suggest it upstream, maybe Andrea or Daniel are working on something like that already.
Rich.
participants (4)
-
Daniel P. Berrangé -
Laszlo Ersek -
Martin Kletzander -
Richard W.M. Jones