Revisiting priorities for Odum Release:
- Gluster services under Kubernetes (in labeled data nodes)
- Volume declarations, automation, and operational integration
- Startup, provisioning in IAAS, initialization, and operations/mgmt
- Related ticket: NDS-223 - Getting issue details... STATUS
- Ability to assign quotas to project directories
- Option 1: Manual process with documentation
- Option 2: Ability to call some Pod
- Related tickets
- NDS-210 - Getting issue details... STATUS
- Container to mount Gluster volumes on nodes
- NDS-227 - Getting issue details... STATUS
- Yaml to create daemon set to mount volumes
- Run on each compute node
- Testing
- Unified Ansible Deployment scripts
- Openstack and kubernetes with NDSLabs reproducible deploy
- Add OpenStack provisioning playbook/roles
- currently lives in ndslabs-deploy-tools?
- NDS-235 - Getting issue details... STATUS
- Add OpenStack provisioning playbook/roles
- GlusterFS
- NDS-236 - Getting issue details... STATUS
- System boot process:
- Add systems-level services - monitoring, admin-tools, introspection tools ... cadvisor,
- NDS-238 - Getting issue details... STATUS
- Related ticket:
- Openstack and kubernetes with NDSLabs reproducible deploy
- Resource limits
- Test multinode with realistic Dataverse deploy
- OOM, High CPU process testing
- Understand dataverse resource requirements, update specs
- Cluster admin tools for monitoring
- Console access
- Load balancer
- NDS-234 - Getting issue details... STATUS
- TLS
- New: TLS HTTPS/WSS support in UI and Apiserver
- Feedback
- NDS-183 - Getting issue details... STATUS
Cluster boot sequence:
- Decide on cluster size. How many of each type of node?
- kubernetes etcd
- kubernetes master
- kubernetes compute
- kubernetes ingress
- gluster storage
- NDS-243 - Getting issue details... STATUS - Allocate desired nodes with Ansible playbook. Nodes should now be online with Kubernetes running.
-
NDS-251
-
Getting issue details...
STATUS
- Label all nodes with their associated role/type from above
- kubectl label nodes <nodeName> ndslabs-role=<nodeType>
- NDS-223 - Getting issue details... STATUS - Deploy GlusterFS server Kubernetes service on nodes with ndslabs-role=storage
- NDS-223 - Getting issue details... STATUS - Deploy GlusterFS client Kubernetes service on nodes with ndslabs-role=compute
- NDS-214 - Getting issue details... STATUS - Deploy "default" Kubernetes LoadBalancer / Ingress rules for UI / API / cluster-wide services (i.e. TERRA toolserver)
-
NDS-244
-
Getting issue details...
STATUS
- Deploy kube-system Kubernetes services
- Grafana, dashboard, ELK, etc
-
NDS-238
-
Getting issue details...
STATUS
- Depoy NDS Labs Kubernetes services
- UI / API
- TERRA toolserver?
-
NDS-252
-
Getting issue details...
STATUS
- Create any necessary NDS Labs projects using the API
- Each project needs:
- Public IP / CNAME
- We need to get ours hands on these IPs soon if we want them in time for the Odum event
- a storage volume (limited by the above quota)
- Ansible task? Script? What is the process for this?
- resource limits/quotas
- Project in etcd under /ndslabs/projects
- Existing script
- Kubernetes Namespace / Secret / LoadBalancer / Ingress ruleset
- API Server creates namespace
- I don't think we use per-project secrets yet
- each project may need its own LoadBalancer and ingress rules (they are namespace-specific)
- Public IP / CNAME
- Each project needs:
- Decide on cluster size. How many of each type of node?
2016-04-29
Completed?:
- Console Access
Priorities:
- Load Balancer
- prototype done in sprint 1
- need to create follow on tickets for testing, TLS, ...
- Unified Ansible Deployment Scripts