Open Stack

Hey guys,
   I started this effort some time ago and ran into several technical
issues.  First of all,
the ceph iscsi driver requires several dependent packages for it to work
properly, end to end.
The biggest problem I ran into was getting the ceph devstack plugin working
and acceptable to
the community.   In order to do ceph iscsi CI, the existing ceph devstack
plugin had to be modified
to install several required packages to properly do iscsi exports.   As it
turns out most distros don't have
packages for those requirements.  (tcmu-runner, targetcli-fb, ceph-iscsi)
all had to be installed from source,
which was unacceptable to the community.  There are no associated packages
for those for ubuntu, and
so it was deemed "unreleased" software.  So the project stalled out.

In my spare time I have been looking at writing a new devstack ceph plugin
purely for ceph-iscsi using
ceph-ansible, as it has native support for installing all deps for fedora,
redhat, centos.   This is a ton of work
and is somewhat a duplication of the existing devstack ceph plugin.   I've
been running into a host of issues
with ceph-ansible, simply being broken and not really maintained to work
properly either.
https://github.com/ceph/ceph-ansible/issues/5382
https://github.com/ceph/ceph-ansible/issues/5383


At the time I got blocked on the CI issue, the volume driver worked and was
able to run tempest against it manually
and attach/detach volumes.   It just required a lot of manual steps to get
everything in place.   I even wrote a
python client that took care of a lot of the details of talking to the ceph
target gw:
https://github.com/hemna/rbd-iscsi-client


So the bottom line is, I need a way to do CI, which means having an
acceptable mechanism for deploying
ceph + all the ceph iscsi toolchain on an acceptable distro.  Once I can
get that working, then we can get
a tempest job in place to do the cinder volume driver CI.

Walt


On Mon, Apr 20, 2020 at 5:36 AM kevinz <kevinzs2048 at gmail.com> wrote:

> Hi Cinder,
>
> We are looking forward to the IRONIC boot from Cinder Ceph Volume. I went
> through the mailist and found the conversation last year, the main gap for
> the Ironic boot from Cinder volume(Ceph BACKEND) is lacking Cinder Ceph
> iscsi driver.
> http://lists.openstack.org/pipermail/openstack-discuss/2019-May/006041.html
>
> So I'm sending to ask the current status for this. We are from Linaro, all
> our machines are under Ironic and we use Ceph as backend storage. Recently
> we are looking at the Diskless server boot plan from the Ceph cluster.
>
> Any info or update or advice, I really appreciate it. Look forward to
> hearing from you.
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-discuss/attachments/20200602/83a4cc96/attachment.html>