Ceph

ceph CLI does not propagate orchestrator error codes to process exit status

When an orchestrator command returns an error (e.g. OrchestratorValidationError), the mgr reply includes the correct retval (e.g. 22 for EINVAL), but the ceph CLI process exits with 0 instead of that error code.

Impact
QA test tasks.mgr.test_orchestrator_cli.TestOrchestratorCli.test_error fails with AssertionError: 0 != 22.
Scripts that rely on ceph exit codes for error handling will not see orchestrator errors.

Environment
Teuthology run: rocky-test-2026-03-02_12:47:57-orch:cephadm-wip-rocky10-branch-of-the-day-2026-02-24-1771941190-distro-default-openstack
Job ID: 103
Ceph SHA: 506b402a3d3a240d0bfe073b4769eaffdf84497b
Rocky Linux 10.1, OpenStack VMs

The test expects raw_cluster_cmd_result (which uses the ceph process exit status) to be 22, but the assertion fails with AssertionError: 0 != 22, so the ceph process exited with 0.