The cores will make teuthology fail the job--and we don't want them for
this test, where we are deliberately causing crashes.
Fixes: https://tracker.ceph.com/issues/43653
Signed-off-by: Sage Weil <sage@redhat.com>
* refs/pull/32006/head:
mgr/cephadm: revert to OrchestratorClientMixin child
mgr/cephadm: initial upgrade logic
cephadm: return 'cephadm pull' result as JSON
mgr/cephadm: upgrade start/stop/pause/resume
mgr/cephadm: track ceph version as well as container in on 'upgrade check'
mgr/orchestrator_cli: upgrade {start,pause,resume,stop}
mgr/cephadm: fix upgrade check
mon/ConfigKeyService: remove config-key set warning
Reviewed-by: Sebastian Wagner <swagner@suse.com>
* refs/pull/32685/head:
cephadm: remove -crash container before starting it
cephadm: stop the crash container the nice way
cephadm: no need for podman rm when passing --rm to run
ceph-crash: exit code 0 on SIGINT, SIGTERM
Reviewed-by: Michael Fritch <mfritch@suse.com>
see also #32632. before we have the complete fix, we should disable this
test to avoid more noise from this test.
Signed-off-by: Kefu Chai <kchai@redhat.com>
* refs/pull/32673/head:
crush/CrushWrapper: behave with empty weight vector
cmake: no need to link libglobal and libblkid for testing crush
test/crush: no need to use libglobal for testing crush
Reviewed-by: Sage Weil <sage@redhat.com>
* refs/pull/32713/head:
qa/tasks/cephadm: start watching initial daemons before bootstrap
qa/tasks/cephadm: create /etc/ceph if it doesn't exist
qa/tasks/cephadm: fix log whitelist when there is no whitelist
Reviewed-by: Josh Durgin <jdurgin@redhat.com>
Existing pools might have too many/few PGs and produce a warning that
prevents us from getting to HEALTH_OK.
Signed-off-by: Sage Weil <sage@redhat.com>
This lets us see output while bootstrap is happening.
(Depends on the teuthology change to use journalctl, see
teuthology commit 4fa83040b05b604280789459f095d6f2ad1b0d01.)
Signed-off-by: Sage Weil <sage@redhat.com>
* refs/pull/32698/head:
common: drop the unused mutable_item_history container.
common: drop the unused prune() of {safe,mutable}_item_history.
Reviewed-by: Kefu Chai <kchai@redhat.com>
Reviewed-by: Sage Weil <sage@redhat.com>
self.mgr does not exist--I can't figure out what this 'mgr' reference
is and where it comes from. So reverting to being a child for the time
being; we can clean this up later.
Signed-off-by: Sage Weil <sage@redhat.com>
The presence of these never-used methods obscure bug investigation.
For instance, `safe_item_history` is used to record all messenger's
addresses. As we never lock reading from it in `_conn_prefix()`, we
expect they live as long their corresponding messenger instances.
When such an item becomes corrupted, presence of `prune()` imposes
verification the method is never called – this effort could have
been avoided if we hadn't have the dangerous method.
Signed-off-by: Radoslaw Zarzynski <rzarzyns@redhat.com>
