]> git-server-git.apps.pok.os.sepia.ceph.com Git - teuthology.git/log
teuthology.git
6 weeks agotask/install: don't attempt to handle strings for extra_system_packages
Ilya Dryomov [Tue, 4 Nov 2025 12:02:46 +0000 (13:02 +0100)]
task/install: don't attempt to handle strings for extra_system_packages

To the best of my knowledge, extra_system_packages has never accepted
a string as a value -- whether at the top level (i.e. on the install
task itself) or on a particular project (e.g. ceph).

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
6 weeks agotask/install: extra_system_packages at the top level can be a list
Ilya Dryomov [Tue, 4 Nov 2025 12:02:46 +0000 (13:02 +0100)]
task/install: extra_system_packages at the top level can be a list

Commit fc5a3197f22e ("task/install: allow both list and dicts for
extra_system_packages") introduced a regression for job definitions
where extra_system_packages at the top level (i.e. on the install
task itself) is a list.  For example

  overrides:
    install:
      ceph:
        flavor: default
        sha1: b7d57fefbec68934ae5e6e7d7c48ac09182397a5
      extra_system_packages:
      - pv

produces AssertionError on

  assert isinstance(b, dict)

in deep_merge() when it's called from _override_extra_system_packages().

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
6 weeks agoorchestra/test: drop opensuse 42.2 from opsys
Kyrylo Shatskyy [Sat, 25 Oct 2025 01:05:02 +0000 (03:05 +0200)]
orchestra/test: drop opensuse 42.2 from opsys

Signed-off-by: Kyrylo Shatskyy <kyrylo.shatskyy@gmail.com>
6 weeks agopackaging: fix test for distro version
Kyrylo Shatskyy [Sat, 25 Oct 2025 00:37:14 +0000 (02:37 +0200)]
packaging: fix test for distro version

Signed-off-by: Kyrylo Shatskyy <kyrylo.shatskyy@clyso.com>
6 weeks agoorchestra/opsys: update suse related versions
Kyrylo Shatskyy [Sat, 25 Oct 2025 00:23:14 +0000 (02:23 +0200)]
orchestra/opsys: update suse related versions

Signed-off-by: Kyrylo Shatskyy <kyrylo.shatskyy@clyso.com>
6 weeks agocargo-cult some packaging updates for rocky
Greg Farnum [Fri, 24 Oct 2025 16:32:10 +0000 (16:32 +0000)]
cargo-cult some packaging updates for rocky

Signed-off-by: Greg Farnum <gfarnum@redhat.com>
6 weeks agotask/install/rpm: don't use builder when using custom repos
Kyr Shatskyy [Thu, 4 Sep 2025 22:07:06 +0000 (00:07 +0200)]
task/install/rpm: don't use builder when using custom repos

Signed-off-by: Kyr Shatskyy <kyrylo.shatskyy@clyso.com>
6 weeks agotask/install/rpm: add/remove yum repos
Kyr Shatskyy [Wed, 3 Sep 2025 21:14:52 +0000 (23:14 +0200)]
task/install/rpm: add/remove yum repos

Signed-off-by: Kyr Shatskyy <kyrylo.shatskyy@clyso.com>
6 weeks agoorchestra/opsys: alma and rocky are rpm distro
Kyr Shatskyy [Wed, 3 Sep 2025 11:22:40 +0000 (13:22 +0200)]
orchestra/opsys: alma and rocky are rpm distro

Signed-off-by: Kyr Shatskyy <kyrylo.shatskyy@clyso.com>
6 weeks agotask/install: allow both list and dicts for extra_system_packages
Kyr Shatskyy [Wed, 15 Oct 2025 22:41:21 +0000 (00:41 +0200)]
task/install: allow both list and dicts for extra_system_packages

Fixes: https://tracker.ceph.com/issues/73527
Signed-off-by: Kyr Shatskyy <kyrylo.shatskyy@clyso.com>
6 weeks agoOpenstack: handle attached_volumes field when fetching volumes
deepssin [Thu, 25 Sep 2025 11:22:32 +0000 (07:22 -0400)]
Openstack: handle attached_volumes field when fetching volumes

Some clouds use attached_volumes instead of os-extended-volumes:volumes_attached

Signed-off-by: deepssin <deepssin@redhat.com>
6 weeks agoorchestra/console.py: increase 'power status' timeout
Dan Mick [Mon, 29 Sep 2025 17:29:39 +0000 (10:29 -0700)]
orchestra/console.py: increase 'power status' timeout

in set_power() we check to see if the power operation worked by
examining ipmi power status.  Give longer than 1s for the
status command to complete.

Signed-off-by: Dan Mick <dan.mick@redhat.com>
2 months agolock.ops.unlock_one_safe: Invert run-match logic 2036/head
Zack Cerza [Wed, 19 Mar 2025 18:35:11 +0000 (12:35 -0600)]
lock.ops.unlock_one_safe: Invert run-match logic

When unlock_one_safe is called with run_name, the caller means to express
"unlock this node if it belongs to this run".
When it is called with run_name and job_id, it means "unlock this node if it
belongs to this job in this run".
We had inverted the logic, causing leaks on reimage failures.

Signed-off-by: Zack Cerza <zack@cerza.org>