mgr/cephadm: clear upgrade health error when upgrade is resumed

The health error is no longer valid if the upgrade has been
removed. If the issue is still present we'll hit it again.

Fixes: https://tracker.ceph.com/issues/57891
Signed-off-by: Adam King <adking@redhat.com>
(cherry picked from commit 23f523dd0fdf079d2e0b41113c524e7b53ed3c0c)

diff --git a/src/pybind/mgr/cephadm/tests/test_upgrade.py b/src/pybind/mgr/cephadm/tests/test_upgrade.py
index 2315511d4473dc6adbe00e91d3fe3310aec7a3c4..107395b93a15bf55c67777bd9808d1856d8fbbe1 100644 (file)
--- a/src/pybind/mgr/cephadm/tests/test_upgrade.py
+++ b/src/pybind/mgr/cephadm/tests/test_upgrade.py
@@ -67,6 +67,19 @@ def test_do_upgrade_offline_hosts(cephadm_module: CephadmOrchestrator):
             cephadm_module.offline_hosts = set([])  # so remove_host doesn't fail when leaving the with_host block
 
 
+@mock.patch("cephadm.serve.CephadmServe._run_cephadm", _run_cephadm('{}'))
+@mock.patch("cephadm.module.CephadmOrchestrator.remove_health_warning")
+def test_upgrade_resume_clear_health_warnings(_rm_health_warning, cephadm_module: CephadmOrchestrator):
+    with with_host(cephadm_module, 'test'):
+        with with_host(cephadm_module, 'test2'):
+            cephadm_module.upgrade.upgrade_state = UpgradeState('target_image', 0, paused=True)
+            _rm_health_warning.return_value = None
+            assert wait(cephadm_module, cephadm_module.upgrade_resume()
+                        ) == 'Resumed upgrade to target_image'
+            calls_list = [mock.call(alert_id) for alert_id in cephadm_module.upgrade.UPGRADE_ERRORS]
+            _rm_health_warning.assert_has_calls(calls_list, any_order=True)
+
+
 @mock.patch("cephadm.serve.CephadmServe._run_cephadm", _run_cephadm('{}'))
 @pytest.mark.parametrize("use_repo_digest",
                          [

diff --git a/src/pybind/mgr/cephadm/upgrade.py b/src/pybind/mgr/cephadm/upgrade.py
index e40d7577299dfb7f64b326ec317b7effb885a066..252d0820de98dfec283eb3cf55e3d4988644c436 100644 (file)
--- a/src/pybind/mgr/cephadm/upgrade.py
+++ b/src/pybind/mgr/cephadm/upgrade.py
@@ -462,6 +462,8 @@ class CephadmUpgrade:
         self.mgr.log.info('Upgrade: Resumed upgrade to %s' % self.target_image)
         self._save_upgrade_state()
         self.mgr.event.set()
+        for alert_id in self.UPGRADE_ERRORS:
+            self.mgr.remove_health_warning(alert_id)
         return 'Resumed upgrade to %s' % self.target_image
 
     def upgrade_stop(self) -> str: