Description of problem: The existing documentation for multisite configuration [1] does not include the ceph-ansible playbook [2]. All of the steps outlined in section 3.6.1 of the doc mentioned are included in the playbook. As it was for us, if we used the instructions to do it all on the cmdline rather than the playbook, the procedure is broken by having the user limiting the playbook rerun to just the RGW nodes, which creates new keyrings and breaks the existing cluster. [1] https://access.redhat.com/documentation/en-us/red_hat_ceph_storage/3/html-single/installation_guide_for_red_hat_enterprise_linux/#configuring-a-multisite-ceph-object-gateway-install [2] https://github.com/ceph/ceph-ansible/blob/master/README-MULTISITE.md Version-Release number of selected component (if applicable): 7.6 (Maipo) 3.10.0-957.el7.x86_64 ceph-base.x86_64 2:12.2.8-128.el7cp How reproducible: Consistent. We were never able to successfully configure multisite with existing documentation and had far more success using the playbook.
Correction, section 3.6.1 does reference the playbook's use but is missing some of the info included in the upstream readme (e.g., rgw_multisite_endpoints_list) and it does have the user limit the playbook to the RGWs which failed for me for the reason mentioned above.
Updating the QA Contact to a Hemant. Hemant will be rerouting them to the appropriate QE Associate. Regards, Giri