Red Hat Bugzilla – Bug 1466873
[Intel OSP13] Discover and compose storage drives on Intel RSD nodes
Last modified: 2017-10-19 11:28:37 EDT
Description of feature:
RSD platforms currently support the Redfish API, and are expected to soon support the Swordfish and Yang-to-Redfish APIs. Red Hat has already written a Redfish library (Sushy) to provide access from within OpenStack to the RSD platforms. Functionality is needed within OpenStack (eg. through the CLI), to view and manage the storage drives through the open APIs. Specifically:
- discover and list RSD storage drives – PCIe devices
- IB and OOB discovery of SAS JBODs (iSCSI targets)
- Allow composition of nodes with full drive PCIe devices.
- Allow decomposition of nodes with erase and no erase option for data on drives
- List storage drive metric
Version-Release number of selected component (if applicable):
OpenStack Client version (potentially) in OpenStack Queens release
2. Business Justification:
a) Why is this feature needed?
RSD is a new architecture that realizes an agile infrastructure where the hardware resources can be pooled according to application needs. It also enables a more easily scaled infrastructure, so CPU, memory, network and storage resources can be added as needed, without the need to do complete replacements of nodes
b) What hardware does this enable?
New platforms based on RSD architecture
c) Is this hardware on-board in a system (eg, LOM) or an add-on card?
RSD nodes are disaggregated (CPU, memory, storage, accelerators etc.)
d) Business impact? N/A
e) Other business drivers: N/A
3. Primary contact at Partner, email, phone (chat)
Priyank Durugkar - firstname.lastname@example.org
4. Expected results:
- DC admin installs the undercloud
- Admin then uses the CLI to discover PCIe and iSCSI storage devices (along with other resources covered in other RFEs)
- Admin then composes a node on a RSD rack with the required PCIe and iSCSI storage resources
- TripleO then deploys the composed node in the overcloud
- Once a node is no longer needed, admin decomposes the node
This bugzilla has been removed from the release and needs to be reviewed and Triaged for another Target Release.
We've added iSCSI and PCIe-attached NVMe support so far. Links:
We've merged all the code into https://github.com/openstack/rsd-lib and https://github.com/openstack/python-rsdclient. We're now working on a final piece - Ironic drivers to recognize the remote storage attached to a composed node - and we're also setting up a CI/CD environment on an RSD rack to test and submit the Ironic changes. This is high priority for Queens release
Hi Krish, could you share the patches related to Ironic mentioned in comment #3?
We understand that this RFE will depend on adding the python-rsdclient to RDO and OSP, then finishing the work you mention with the Ironic to recognise the remote storage.
Adding it to OSP 13, pending the usual reviewing of dependencies, security, legal, etc. for adding python-rsdclient and its dependencies to OSP.
(In reply to Ramon Acedo from comment #5)
> Hi Krish, could you share the patches related to Ironic mentioned in comment
> We understand that this RFE will depend on adding the python-rsdclient to
> RDO and OSP, then finishing the work you mention with the Ironic to
> recognise the remote storage.
> Adding it to OSP 13, pending the usual reviewing of dependencies, security,
> legal, etc. for adding python-rsdclient and its dependencies to OSP.
Ramon, here's been some conversation in Gerrit between Dmitry and our engineers on the Ironic driver. It looks like that driver enhancement may not be needed ie. the manual work-around to attach remote storage is fine. See https://review.openstack.org/#/c/506400/6