Past internship ideas
Contents
- 1 Documentation
- 2 Community
- 3 Coding
- 3.1 Murano - Murano package validation tool
- 3.2 Neutron - ovsdb client monitor for Windows
- 3.3 Glance - Extended support for requests library
- 3.4 Glance - Develop a python based GLARE (GLance Artifacts REpository) client library and shell API
- 3.5 Searchlight - Extend automated functional testing for Searchlight plugins / Improve existing plugins
- 3.6 OSprofiler - Implement new storage drivers for OSprofiler / add OSprofiler support to other OpenStack projects
- 3.7 Magnum - Container Service for Magnum's Kubernetes Orchestration Engine
- 3.8 Horizon- Context-sensitive help in openstack-dashboard
- 3.9 Keystone- Long Term Tokens
- 3.10 Neutron - metering agent add port statistics
- 3.11 Redis jobboards
- 3.12 Policy | Policy enforcement in OpenStack: Nova Scheduling and Policy integration
- 3.13 Test results dashboard
- 3.14 Policy | Policy-based intent-driven OpenStack Cloud Infrastructure
- 3.15 Rich Network Services in a Federated OpenStack Public Cloud
- 3.16 Understand OpenStack Operations via Insights from Logs and Metrics: A Data Science Perspective
- 3.17 Keystone | Dynamic Policy
- 3.18 Keystone | Tokenless Operations
- 3.19 Ceilometer
- 3.20 Trove
- 3.21 Glance - Swift ranged uploads
- 3.22 Swift - storage server OPTIONS support and checker tool
- 3.23 Swift/Swift3 - Improve S3 compatibility layer
- 3.24 Proposed by applicants
- 4 Design
Documentation
Performance-docs - Add missing sections to http://docs.openstack.org/developer/performance-docs/# and identify the documentation gaps
Description:
Performance-docs is quite new initiative leaded and pushed by OpenStack Performance Working Group - https://wiki.openstack.org/wiki/Performance_Team - and we really need your help to work on adding test results, topologies and environments description, etc. to make this source valuable for all community.
Difficulty | Medium |
Topics | Performance-docs |
Required skills | Good English and great communication skills to collect the information |
Extra skills | |
Mentor | DinaBelova |
Mentors:
- loquacities on IRC / openstack at lanabrindley dot com
- asettle on IRC/ alexandra dot settle at rackspace dot com
- Assist with the RST conversion of the Admin User Guide
- Help with the Networking Guide: either with content, copyediting, or the RST conversion.
- General bug work (find something interesting and work on it). This would look great on her resume as it's a solid visible contribution, but would also hopefully pay down some of our technical debt.
- Choose a book/section of interest, and research and update that section. A good candidate here would be the API guides.
Community
To be announced
Mentors:
- hodepodge on IRC / chris at openstack org
- EmilienM on IRC / emilien at redhat com
- reed in IRC / stefano at openstack org
Coding
See also Ideas pages.
Murano - Murano package validation tool
Description:
Murano cloud-ready applications are written in MuranoPL language and are packaged into zip packages, with certain prerequisites obligatory (manifest file, package structure, etc.). Having a murano package verification tool would speed up development and debugging of such apps greatly. Package verification tool should include tools for verifying that package structure is correct, all the class files mentioned in manifest file are present. It should also include MuranoPL linting code, to speed up MuranoPL-app development. Finally after the tool is ready — we should implement checks at import level (when importing a package to murano-api) and jobs at commit-time for murano-apps
Difficulty | Medium |
Topics | Murano |
Required skills | Python |
Extra skills | YAML |
Mentor | kzaitcev |
Neutron - ovsdb client monitor for Windows
Description:
The OVS agent monitors the ports that are added in the compute host to be able to wire them correctly. In Linux it uses the class InterfacePollingMinimizer that notifies the agent when a new port is plugged or unplugged and passes the related events (port added or deleted). For Windows it uses the class AlwaysPoll that doesn't notify any specific event, it returns always true. The OVS agent in Windows is forced to rescan the devices currently in the machine to infer which were added. This is because the current Windows implementation of the interface polling manager doesn't use ovsdb client monitor. The aim of this project is to use ovsdb client monitor also for Windows and make sure that the events are passed correctly to the OVS agent. This will improve the performance and will enable some clean up in the OVS agent code. The first step is getting familiar with the OVS agent and with Neutron in general. Then you will approach openvswitch tools and investigate how to use ovsdb monitor client in Windows. Neutron folks are pretty active on #openstack-neutron channel most of the time and would be willing to share their opinions on this or any other project. You'll submit your code upstream and address the comments you get till your patch gets merged.
Difficulty | Medium |
Topics | Neutron |
Required skills | Python |
Extra skills | Networking, OVS |
Mentor | rossella_s |
Glance - Extended support for requests library
Description:
You would be learning about glance-replicator, glance_store drivers and if time permits other modules. You are then expected to add support for requests library ( https://pypi.python.org/pypi/requests ) starting with glance-replicator. Although, supporting more than glance-replicator is not expected, more support you add the better your internship will be. You will also help with bug triage and optimizations around that code base as you add more support.
Difficulty | Medium |
Topics | Glance |
Required skills | Python |
Extra skills | Good communication skills |
Mentor | nikhil |
Glance - Develop a python based GLARE (GLance Artifacts REpository) client library and shell API
Description:
You will learn how python based clients are developed in the Openstack realm. You will be responsible for closely working with the Glare drivers to understand the requirements, API evolution and contribute ideas to the development of the Glare API. You should be able to set up the basic build structure, common interfaces, setup configs and infra jobs for the glareclient. Co-ordinate with the Glare drivers and infra team to setup repositories, documentation and test jobs for releases of this client. Also, based on the outcome and feedback from the Glare API discussions you will be responsible for keep evolving the client library.
Difficulty | Medium-Advanced |
Topics | Glance |
Required skills | Python |
Extra skills | Shell scripting & packaging, good communication skills |
Mentor | nikhil |
Searchlight - Extend automated functional testing for Searchlight plugins / Improve existing plugins
Description:
Searchlight is intended to dramatically improving the search capabilities and performance of various OpenStack cloud services by indexing their data into ElasticSearch using plugins. You will be learning the Searchlight fundamentals including indexing, searching, faceting, security, etc. You will also learn the APIs and data models of various other OpenStack services that are indexed into Searchlight. You will understand plugin design then explore all of the functional testing aspects mentioned above. You will be responsible for implementing complete coverage of functional testing for one big or two medium sized plugins. You will improve the existing plugins for any bugs or improvements you discover.
Difficulty | Medium-Advanced |
Topics | Searchlight |
Required skills | Python |
Extra skills | Basic understanding of ElasticSearch, good communication skills |
Mentor | nikhil |
OSprofiler - Implement new storage drivers for OSprofiler / add OSprofiler support to other OpenStack projects
Description:
OSprofiler is an Oslo library allowing to trace cross-project requests and identify the OpenStack performance bottlenecks via understanding what time was spent on each request stage, how many requests were used, etc. Lots of developing efforts might be found here, including writing new storage drivers for OSprofiler and adding its integration to other OpenStack projects.
Difficulty | Medium-Advanced |
Topics | OSprofiler |
Required skills | Python |
Extra skills | |
Mentor | DinaBelova |
Magnum - Container Service for Magnum's Kubernetes Orchestration Engine
Description:
Magnum's client has several actions (create/delete/exec/logs/pause/reboot/start/stop/unpause) for containers, currently these commands work only for Docker Swarm COE. When a operator deploys a bay with either Kubernetes COE or the Mesos COE, these command line functionality is not available for the operator as there is not backend support for these operations. In this project, we will first add a concrete implementation for the Container Service that calls Kubernetes API appropriately, then we make sure that the magnum's client command lines work properly against this, just like this works when the operator deploys a bay using swarm COE.
Difficulty | Medium |
Topics | Magnum |
Required skills | Python |
Extra skills | |
Mentor | Dims |
Horizon- Context-sensitive help in openstack-dashboard
Description:
Context-sensitive Help allows the openstack-dashboard to dynamically provide users with links to relevant content, depending on the user's location in the dashboard (ie. their 'user context). This implementation should also allow developers to easily define links to serve for each specific context. For more details please check this blueprint
Difficulty | Medium |
Topics | Horizon |
Required skills | Python |
Extra skills | Good communication skills |
Mentor | sayalilunkad |
Keystone- Long Term Tokens
Description:
We want to make a form of validation that will ignore expiration of tokens in order to support long running tasks
Difficulty | Easy |
Topics | keystone |
Required skills | Python |
Extra skills | Keystone is its own thing, but we'll teach you |
Mentor | ayoung |
Neutron - metering agent add port statistics
Description:
In Neutron the metering agent [1] collects statistics regarding bandwidth usage. Right now it only measure the bandwidth used by routers. The idea is to extend it and provide statistics also for ports. In the first implementation only openvswitch will be supported, since we will use openvswitch tools to get the port statistics. The first step will be getting familiar with the metering agent and with Neutron in general. Then you will approach openvswitch tools and think about how to use them for this project. After that you can reach out to the community to collect and discuss ideas. Neutron folks are pretty active on #openstack-neutron channel most of the time and would be willing to share their opinions on this or any other project. You'll submit your code upstream and address the comments you get till your patch gets merged.
Related blueprint: https://blueprints.launchpad.net/neutron/+spec/port-statistics
Expected results: Patchsets submitted for review, reviewed by community members.
Knowledge prerequisites: programming knowledge
Nice-to-have knowledge: Python, Openvswitch
Mentors: Rossella Sblendido
Redis jobboards
The spec @ https://review.openstack.org/#/c/105298/ basically explains all of this (the why, what and where).
Difficulty | |
Topics | TaskFlow, Distributed Systems |
Required skills | Python |
Extra skills | Distributed Systems |
Mentor | Joshua Harlow (harlowja) |
Policy | Policy enforcement in OpenStack: Nova Scheduling and Policy integration
Difficulty | |
Topics | Policy, Nova |
Required skills | |
Extra skills | |
Mentor |
Test results dashboard
Tasks:
* Create an infra hosted dashboard view of testing results (both check/gate and periodic)
Difficulty | |
Topics | |
Required skills | |
Extra skills | |
Mentor |
Policy | Policy-based intent-driven OpenStack Cloud Infrastructure
See an example case study here: Meta-data Driven Cloud - Running OpenStack @ Scale Using a Policy Framework
Tasks:
* Extension of intent-based policy framework to Compute and Storage * Policy-based service composition integration with OpenDaylight
Difficulty | Moderate to Advanced |
Topics | Functional and scenario testing framework |
Required skills | OpenStack, Python |
Extra skills | |
Mentor | Sumit Naiksatam, Magesh GV |
Rich Network Services in a Federated OpenStack Public Cloud
L4-7 Services' integration, NFV, service function chaining, OpenDaylight
Difficulty | Moderate to Advanced |
Topics | Functional and scenario testing framework |
Required skills | OpenStack, Python |
Extra skills | |
Mentor | Sumit Naiksatam, Magesh GV |
Understand OpenStack Operations via Insights from Logs and Metrics: A Data Science Perspective
Tasks:
* How do we make sense out of logs? * How do we set up alarms based on specific events (identify events) * Anomaly detection from Logs
Difficulty | |
Topics | |
Required skills | Experience/interest in algorithms |
Extra skills | |
Mentor |
Keystone | Dynamic Policy
The general idea is described at the blog post Dynamic Policy in Keystone, by Adam Young.
Difficulty | Medium |
Topics | Keystone |
Required skills | Python |
Extra skills | |
Mentor | David Stanek (dstanek), Samuel de Medeiros Queiroz (samueldmq) |
Keystone | Tokenless Operations
The general idea is described at the blog post Who can sign for what?, by Adam Young. In few words, the fact of the signature on a document be valid does not mean that the signer was authorized to sign it.
The proposed challenge is to create, based on existing mechanisms, an extension to Keystone client able to check if the the signer of a token was effectively authorized to sign tokens.
Tasks:
* Extract signer data out of the certificates; * Fetch the complete list of certificate from Keystone using the OS-SIMPLE-CERT extension; * Match the signer to the cert to validate the signature and extract the domain data for the token; * Fetch the mapping info from the Federation extension; * Use the mapping info to convert from the signing cert to a keystone user and groups; * Fetch the effective roles from Keystone for the user/groups for that domain; * Fetch policy from Keystone; * Execute the policy check to validate that the signer could sign for the data.
Difficulty | Medium |
Topics | Keystone |
Required skills | Python |
Extra skills | |
Mentor | David Stanek (dstanek), Samuel de Medeiros Queiroz (samueldmq) |
Ceilometer
Mentor: Dina Belova
Ceilometer team is currently working on new Time-Series storage concept for metrics. For now Gnocchi (Telemetry Stackforge project where we're trying to implement this kind of effort) lacks of the backend support (only Swift in place, InfluxDB and OpenTSDB still in progress). There is some interest to use Ceph directly instead of it. So we need direct-to-ceph usage from Gnocchi, possibly via the rados gateway REST API (as opposed to ceph-sitting-behind-swift-proxy as an alternative storage driver for Swift itself).
Trove
Mentor: Iccha Sethi
Description: MySQL replication enhancements
Work on trove enhancements for mysql replication.
Glance - Swift ranged uploads
Note(nikhil_k): This is a really challenging project. It would need a lot of research and motivation from candidate to perform the tasks. The tasks would also need to be defined by the candidate themselves to have a good "concurrent" distributed system as a end-result.
Description: We currently retry the entire upload process if it fails. Need to add the ranged uploads logic from swift store to improve performance. Target repo - openstack/glance_store
Glance folks are pretty active on #openstack-glance channel most of the time and would be willing to share their opinions on this or any other project. Please reach out on IRC to get a detailed information on this.
Related blueprint: TBA
Expected results: Patchsets submitted for review, reviewed by community members.
Knowledge prerequisites: programming knowledge
Nice-to-have knowledge: Python
Mentors: Nikhil Komawar
Swift - storage server OPTIONS support and checker tool
Many times new deployers get mysterious errors after first setting up their Swift clusters. Most of the time, the errors are because the values in the ring are incorrect (e.g. a bad port number). We need a way to validate that the rings actually match the deployment.
http://tools.ietf.org/html/rfc7231#section-4.3.7 talks about the OPTIONS verb. Swift's proxy server already supports OPTIONS, but the storage nodes do not. The first task is to implement OPTIONS on the account, container, and object servers.
Once the "OPTIONS *" functionality is implemented, swift-recon can be updated to include a ring checker. This checker would look at the ring and validate that there is something running on the server endpoints listed in the ring. Furthermore it would be able to actually check that the endpoint, if it's a real endpoint, is the right kind of endpoint (e.g. actually check that it's an object server). swift-recon would then generate a report of any found issues in the cluster.
Expected results: patches written, submitted for review, and reviewed by the community
Knowledge prerequisites: Python programming knowledge
Nice-to-have knowledge: familiarity with HTTP protocols
Mentors: John Dickinson
WIP Schedule: https://etherpad.openstack.org/p/iaR4TQZ7NP
Swift/Swift3 - Improve S3 compatibility layer
Improve the Swift3 middleware to get better S3 API coverage
https://github.com/stackforge/swift3
Mentors: John Dickinson
Proposed by applicants
Keystone - Implementation of Attribute and Graph Based Access Control Model (AGBAC) for Openstack
Proposed by: Tahmina Ahmed - Tahmina
Description: The Core Openstack Access Control Module according to identity API V3 (user, token,project, domain, group, role association) [0] can be abstracted as a graph:
Attribute of different entities and the relationship between them and attribute of the relationship can be expressed through property graph [2] where entities are nodes and relationships are edges and both nodes and edges has attribute. Using entities and their attributes and relationship between entities and attribute of relationship to specify authorization policy will allow a system to have more finer grained access control model. For this representation, OpenStack entities (user, group, role, project, domain) are represented as nodes in the graph and the attributes and association between any two of them can be depicted as attributed edges. Given that a role is associated with a user and a project where the association is temporal, if we can say that the association is only active from 8am to 5pm this is something to say about the user- project- role association. We can express this as an attribute of the user- project and project - role association edge and use this for authorization to findout active roles.
This implementation plan is to make minimum impact on services outside keystone.So the plan is to like computing authorization path specification for a certain time when user requests for a token and and return a list of active roles. In that case this extension to the authorization model is transparent to all other openstack services like nova, glance, cynder etc.
Steps to Implement AGBAC in Openstack
1. Define API and Storage for Specifying Attributes of different entities 2. Define storage Assignment Attributes. 3. Define API to set the entity attributes 4. Define API to set the association attributes 5. A Policy specification storage that would specify path based policy to compute roles. 6. A Compute function that would compute the roles using entites, their attributes relationship between entities attribute of relationships and policy.
[0] Bo Tang and Ravi Sandhu, Extending Openstack Access Control With Domain Trust. In Proceedings 8th International Conference on Network and System Security (NSS 2014), Xi'an, China, October 15-17, 2014, 15 pages
[2] Rodriguez, Marko A., and Peter Neubauer. "Constructions from dots and lines." Bulletin of the American Society for Information Science and Technology 36.6 (2010): 35-41.
Design
Persona research and design for Horizon UI
Description: There has been some initial work done in a few different areas around generating Personas for the Horizon UI. This task/project would allow someone to lead the effort to pull the research together, perhaps perform further end-user interviews or surveys, analyze data, form groups of personas, and put together an initial set of target Horizon personas that we can use during requirements gathering, design phases, and general discussion.
Related blueprint: https://blueprints.launchpad.net/openstack-ux/+spec/horizon-personas
Expected results: Initial set of Personas for people to use during requirements gathering, design phases, and general discussion of our target Horizon users.
Knowledge prerequisites: basic interviewing skills, willingness to talk with a lot of new people, readiness to listen and report back findings
Nice-to-have knowledge: User Experience Design basics
Mentors: Ju Lim
Redesign for the Object Storage -> Containers section in Horizon
Description:
Currently, there is a section under the Object Storage panel that allows users (e.g. Consumers and/or Project Administrators) to Create, Delete, and Edit Containers. It would be great to do a little bit of research into identifying the use cases around these features and then proposing a redesign of these features based on the findings. This will include learning some basic Usability best practices and applying them to a design. The design would be done in wireframe format, so it could be done in any tool that allows for this. The design would be reviewed on the UX AskBot site, a new blueprint would need to be created as well. This would take a new designer through the process that currently exists for proposing an updated design to a current feature in Horizon.
Related blueprint: N/A (To be created by intern as part of process)
Expected results: Redesign of "Containers" section in Object Storage panel proposed and approved by Horizon community.
Knowledge prerequisites: basic sketching or wireframing skills
Nice-to-have knowledge: User Experience Design basics
Mentor: Ju Lim
Horizon Concept Review and Usability Testing
Description: There has been very little work done around Horizon UI concept reviews and usability testing. This task/project would allow someone to lead the effort to put together a test plan, identify customers as test participants, conduct (contextual) inquiry / interview, survey users, analyze data to share results / findings with the community findings regarding how current production OpenStack users are using the software, their use cases, and whether the Horizon UI meets their needs. Along the way, you may identify areas of improvement, concepts that need further refinement, and new use cases or opportunities that have not yet been addressed.
Related blueprint: https://blueprints.launchpad.net/openstack-ux/+spec/baseline-horizon-usability-testing
Expected results: (1) Identify a community of users willing to participate in Horizon UI usability testing and participate in various OpenStack concept reviews. (2) Share usability test findings with the community that will be used for design and general Horizon discussions.
Knowledge prerequisites: usability testing skills, basic interviewing skills, willingness to talk with a lot of new people, readiness to listen and report back findings, ability to analyze data to find patterns.
Nice-to-have knowledge: User Experience Design and/or User Research basics
Mentors: Ju Lim