Description of problem: The Ceph iSCSI gateway should be able to be deployed within a container (tcmu-runner, ceph-iscsi-gw, ceph-iscsi-api). Version-Release number of selected component (if applicable): RHCS 3.0
Thought: since we are directly tied to the kernel, it seems that putting the user-space portions into a container is only a partial story. For a longer term solution, we could use kata containers (which is a light-weight VM) to truly isolate iSCSI from the host kernel.
We don't have the bandwidth to support this in 3.1
Patch merged upstream, so it'll be in 3.1, although I don't think QE has the bandwidth.
PR 2466 is in master but not stable-3.1, so it's not in any tagged version of ceph-ansible. We need a v3.2.0beta1 tag on master to get this through CentOS.org/RDO and QE testing.
You got it: https://github.com/ceph/ceph-ansible/releases/tag/v3.2.0beta1
I updated the downstream repo so we are waiting for a new container image. Ken, please tell us when the new container image will be ready as nothing is showing up in the BZ. Moving this to POST and will be ready for testing soon with the new container build. Thanks
Created attachment 1498553 [details] ansible.log
The gateways configuration through Ansible is not possible at the moment in a containerized scenario. The reason is simple the Ansible module cannot access the iscsi* libs since it runs on the host and libs are packaged in the container. Presently, we don't have a way to deal with that.
Purge is not a blocker, the BZ is about deploying iSCSI gws in containers, not purging them.
(In reply to leseb from comment #32) > Purge is not a blocker, the BZ is about deploying iSCSI gws in containers, > not purging them. @leseb, so you mean we should just document around this for 3.2 right? If so, I will work with Aron to add some info about this.
Just to clarify, are we talking about purging the whole setup or removing some gateways?
John, this will be in 3.2. Added the doc text.
Thanks. Can you review my updated Doc Text? For my reference here is the old one: Feature – describes the enhancement from the user's point of view No enhancements noticeable, it's just a change on how daemons are being executed (in containers). This is not adding any particular functionality. Reason – why was the enhancement implemented This allows users to deploy a containerized version of their Ceph iSCSI gateways as part of their already containerized Ceph cluster. Result – what is the current user experience (may also be compared to the previous user experience) No difference.
Yes it can go ON_QA
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-2019:0021