This RFE tracks support for the following pattern which is relevant to running multiple instances of Glance accross multiple edge sites: 1. deploy central site with internal ceph1 and glance1 with rbd backend ceph1 2. deploy edge site with internal ceph2 and glance2 with rbd backend ceph2 3. deploy edge site with internal ceph3 and glance3 with rbd backend ceph3 4. deploy edge site with internal cephN and glanceN with rbd backend cephN 5. update central site with internal ceph1, external ceph2, ceph3, ..., cephN and glance1 with rbd backends ceph1, ceph2, ceph3, ..., cephN Steps 1-4 are already possible through split control plane deployments [0], but step 5 is not possible because a single stack only deals with a single ceph cluster. This RFE is focussed only on step 5's support for multiple external ceph clusters in a single stack update (internal ceph1, external ceph2, ceph3, ..., cephN). Though glance supports multiple backends [1], a separate RFE could track enabling TripleO to configure them (glance1 with rbd backends ceph1, ceph2, ceph3, ..., cephN). More details on configuring multiple external Ceph backends is filed in an upstream Tripleo Blueprint [2]. During step 5 we should support (and document) which ansible tags to pass when using config-download in order to minimize change so that the central stack update minimizes service restarts. [0] https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/features/distributed_compute_node.html [1] https://docs.openstack.org/glance/rocky/admin/multistores.html [2] https://blueprints.launchpad.net/tripleo/+spec/multiple-external-ceph
(In reply to John Fulton from comment #0) > During step 5 we should support (and document) which ansible tags to pass > when using config-download in order to minimize change so that the central > stack update minimizes service restarts. tracked in https://bugzilla.redhat.com/show_bug.cgi?id=1767581 The idea is that we'd have: openstack overcloud deploy --config-download-args "--skip-tags something" for the central stack update
The changes merged in THT + tripleo-ansible: https://review.opendev.org/#/q/status:merged+topic:multiple-external-ceph-train I'm changing the component to THT since that's what one of the fixed-ins will be.
Verified for OSP 16.1
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2020:3148