Page History

...

NDS Labs API/CLI requirements

Starting the API server

eval $(docker run --rm -t ndslabs/system-shell usage docker)
git clone https://github.com/craig-willis/ndslabs-startup
cd ndslabs-startup
./kube-up.sh
./ndslabs-up.sh

...

If needed, download the latest

...

copy of

...

ndslabsctl

...

- wget

- https://github.com/nds-org/ndslabs/releases/

...

- download/v1.0-alpha

...

If the API server is running under docker or Kubernetes, you will need to specify the server address. This can be done on the commandline using the --server flag or by creating a .ndslabsctl.yaml file in your home directory. For example:

APISERVER=`kubectl describe svc ndslabs-apiserver | grep ^IP | awk '{print $2}'`

...

- /ndslabsctl-linux-amd64 -O /usr/local/bin/ndslabsctl
- chmod +x ndslabsctl
If testing a new version of API server
- vi /usr/local/lib/ndslabs/apiserver
- Change latest -> test
- docker rmi ndslabs/apiserver:test
ndslabs-up.sh
"ndslabsctl list projects" to confirm things are working

Global preconditions

Kubernetes is running
API Server is running either as a Kubernetes service or as a standalone Docker container
ndslabsctl has been installed and is in your path

...

Preconditions: none
Test steps:
- ndslabsctl
- ndslabsctl -h
- ndslabsctl --help
Expected results
- Any of these commands should display the help/usage information

Authentication/authorization

Preconditions:
- Project has been created or admin user
Test steps:
- ndslabsctl login admin
- ndslabsctl login <user>
- Prompted for password
Expected results:
- Login succeeded
Alternative flows
- Invalid username or password results in login error
Notes:
- Login information is stored in ~/.apictl/.passwd. This files contains the currently logged in user and an authentication token. Removing this file is the equivalent of logging out

Logout

Preconditions
- User is logged in
Test steps
- ndslabsctl logout
- ndslabsctl list services
Expected results:
- List command results in "unauthorized"

Change password

Preconditions:
- User is logged in
Test steps
- ndslabsctl login <user>
- ndslabsctl passwd
  - Enter current password
  - Enter new password
  - Confirm new password
Expected results
- Password is changed

Project administration

...

Account administration

Add account (admin only)

Preconditions:
- Admin is logged in
- Test steps
  - ndslabsctl add

project

- - account <name> <password>
  - ndslabsctl add

project

- - account -f

<project

- - <account.json>
  - ndslabsctl add

project

- - account --file

<project

- - <account.json>
- Expected results
  - Prompted for admin password

Project

- Account is added. User can login with specified name and password
Alternate flows

Project

- - Account exists with same name (409 conflict)

Logged in user is not

- - Incorrect admin password (401 unauthorized)
- Sample

...

- Data:
  - Code Block
    language js
    { "id": "demo", "name": "

...

- - demo account",

...

- - "description": "

...

demo account description",
    "namespace": "

...

- - demo",

...


    "password": "12345",
	"storageQuota": "10",
	"resourceLimits": {
		"cpuMax": "2",
		"cpuDefault": "1",
		"memMax": "8Gi",
		"memDefault": "100Mi",
		"storageQuota": "10"
	}
}

Delete account (admin only)

- Test steps
  - ndslabsctl account project <account>
- Expected results
  - Prompted for admin password
  - Account

Delete project (admin only)

Preconditions
- Admin is logged in
Test steps
- ndslabsctl delete project <project>

Expected resultsProject

- is deleted
Alternate flows:

Project User is not admin

- - Account does not exist (404 not found)
Problem: currently results in 500
- - Incorrect admin password (401 unauthorized)

Get

...

account details (admin or current project only)

- Preconditions
  - Admin is logged in or current

project

- - account user is logged in
- Test steps
  - ndslabsctl get

project <project>

- - account <account>
- Expected results
  - Displays

project details

- account details (as json)
- Password is redacted
Alternate flows:

Project

- - Account does not exist (404)

User is not admin and project is

- - Account is not current project (401 unauthorized)

List

...

Admin is logged in

accounts

Test steps
- ndslabsctl list

...

- accounts
Expected results

...

- Prompted for admin password
- Accounts are listed

...

Alternate flows:

...

- Incorrect admin password (401 unauthorized)

Authentication/authorization

Preconditions:
- Account has been created
Test steps:
- ndslabsctl login <user>
- Prompted for password
Expected results:
- Login succeeded
Alternative flows
- Invalid username / password (401 unauthorized)
Notes:
- Login information is stored in ~/.apictl/.passwd. This files contains the currently logged in user and an authentication token. Removing this file is the equivalent of logging out
Possible Issues:
- It is possible to execute login while already logged in.
  - This may be intentional to allow you to easily switch namespaces.
  - Won't fix, for now

Logout

Preconditions
- User

...

Service administration

Add service (admin only)

Preconditions:
- Admin is logged in
Test steps
- ndslabsctl add service -f <service.json>
- ndslabsctl add service --file <service.json>
Expected results
- Service is added. User can add service/stack.
Alternate flows
- Service exists with same key (409 conflict)
- Logged in user is not admin (401 unauthorized)
Notes:
- See ndslabs-specs repos for sample services

Delete service (admin only)

Preconditions
- Admin is logged in
Test steps
- ndslabsctl delete service <key>logout
- ndslabsctl list services
Expected results:
- List command results in "unauthorized"

Change password

Preconditions:
- User is logged in
Test steps
- ndslabsctl login <user>
- ndslabsctl passwd
  - Enter current password
  - Enter new password
  - Confirm new password
- Service is deleted
Alternate flows:
- Project does not exist (404 not found)
- User is not admin logged in (401 unauthorized)
- Not logged in (401)
- Not admin user (401)
- Service is in use by projects (409)
- Service required by other services (409)
- No such service (404)
- Invalid password
- Password mismatch yield error message: "Passwords do not match"
Expected results
- Password is changed

Service administration

Get service details

- Preconditions
  - User is logged in
- Test steps
  - ndslabsctl get service <key>
- Expected results
  - Displays service spec (as json). Defaults to user catalog
- Alternate flows:
  - Service does not exist (404)

List

...

services

Preconditions
- User is logged in
Test steps
- ndslabsctl list services
- ndslabsctl list services -c system
- ndslabsctl list services -c user
- ndslabsctl list services -c all
Expected results
- Services are listed

...

Volume administration

...

- for the specified catalog (Default: user)
Alternate flows:
- User not logged in (401 unauthorized)

Add service

Test steps
- ndslabsctl add service -f <service.json> -c <catalog>
Expected results
- Prompted for admin password
- Service is added to the specified catalog (default: user). Only admin user can add to system catalog
- "ndslabsctl list services" now shows added service
Alternate flows
- Service exists with same key (409 conflict)
  - Problem: value is replaced in etcd instead of error thrown - this may be intentional
    - Should not replace service if being used in a stack
    - Prompt for confirmation on replace?
- Incorrect admin password (401 unauthorized)

Sample Data:

Code Block

language	js

{
    "key": "whalesay",
    "label": "Docker Whalesay Example",
    "image": "ndslabs/cowsay-php",
    "description": "An example of a custom spec loaded into NDSLabs",
    "isStack": true,
    "isService": true,
    "isPublic": true,
    "isStandalone": true,
    "ports": [
        { "port": 80, "protocol": "http" }
    ]
}

Delete service

- Test steps
  - ndslabsctl delete service <key> -c <catalog>
- Expected results
  - Prompted for admin password
  - Service is deleted from the specified catalog
  - Default catalog is user
  - Admin user required to delete from system catalog
  - "ndslabsctl list services" no longer shows deleted service
- Alternate flows:
  - Incorrect admin password (401 unauthorized)
  - Service is in use by projects (409)
  - Service required by other services (409)
  - No such service (404)

Volume administration

Create volume

- Preconditions
  - Logged in
  - Sufficient quota remaining (not implemented)
  - Service exists with no existing attachments
  - Service not running
- Test steps
  - ndslabsctl

create

- - add volume <vol name> <vol size GB>
    - Creates a detached volume
  - ndslabsctl

create

- - add volume <vol name> <vol size GB> <sid>
    - Creates and attaches volume
- Expected results
  - Volume is created
  - Volume is created and attached
- Alternate

flow

- flows
  - Not logged in (401)
  - Insufficient quota (409) (not implemented)
  - ~~Duplicate name (409)~~
    - NDS-181 now allows duplicate volume names
  - Stack service already has a volume attached (409)
  - Stack service doesn't exist (404)

Attach volume

- Preconditions:
  - Logged in
  - Volume

exits

- - exists
  - Service exists with no existing attachments
  - Service not running
- Test steps
  - ndslabsctl attach <vol name> <sid>
- Expected results
  - Volume is attached to specified service
- Alternate

flow

- flows
  - Not logged in (401)
  - Volume doesn't exist (404)
  - Service doesn't exist (404)
  - Service is running (409)
  - Volume is attached (409)
    - Problem: Reattaches volume to new service without throwing error
- Issues:
  - It is possible to attach a single volume to any kind of service. This may be intentional.

List volumes

- Preconditions
  - Logged in
  - Volumes exist
- Test steps
  - ndslabsctl list volumes
- Expected results
  - Volume details are displayed

Alterate flow

- Alternate flows
  - Not logged in (401)
  - No volumes exist (200)

Detach volume

- Preconditions
  - Logged in
  - Volume exists and is attached
  - Stack not running
- Test steps
  - ndslabsctl detach <vol name>
- Expected results
  - Volume is detached
- Alternative

flow

- flows
  - Not logged in (401)
  - Volume doesn't exist (404)
  - Stack is running (409)
    - Problem: Was able to detach from running stack without error
    - Resolved
  - Volume isn't attached (409)
    - Problem: Was able to detach multiple times without error
    - Resolved

Delete volume

- Preconditions
  - Logged in
  - Volume exists
  - Stack not running
- Test case:
  - ndslabsctl delete <vol name>
- Expected results:
  - Volume is deleted
- Alternative

flow

- flows
  - Not logged in (401)
  - Volume doesn't exist (404)
  - Stack is running (409)

Stack administration

Admin user can create / start / stop stacks. We will need to revisit the workflow for admin.

Add stack

...

Preconditions
Logged in
Service exists
Test steps
ndslabsctl add stack <key> <name>
ndslabsctl add stack <key> <name> --opt=<comma-separated list of optional services>
Expected results
Stack is added
Alternate
flow
flows
Not logged in (401)
Service doesn't exist (404)
Duplicate stack name (409)
NDS-181 now allows duplicate stack names

List stacks

- Preconditions:
  - Logged in
  - Stack has been added
- Test steps
  - ndslabsctl list stacks
- Expected results
  - Summary of stacks is displayed (name, services, status, SIDs)
- Alternate flow:
  - Not logged in (401)
  - No stacks (empty list)

Delete stack

- Preconditions:
  - Logged in
  - Stack exists
  - Stack not running
- Test steps
  - ndslabsctl delete stack <sid>
  - Stack is deleted (201)
- Alternate

flow

- flows
  - Not logged in (401)
  - Stack doesn't exist (404)
  - Stack is running (409)
  - Attached volumes are orphaned

Get stack details

- Preconditions:
  - Logged in
  - Stack exists
- Test steps
  - ndslabsctl get stack <sid>
- Expected results
  - Stack details, including endpoints and config, are displayed
- Alternate

flow

- flows
  - Not logged in (401)
  - Stack doesn't exist (404)

Start stack

- Preconditions
  - Logged in
  - Stack exists and started
- Test steps
  - ndslabsctl start <sid>
- Expected results
  - All stack services (service, replication controller, pods) are started and ready
- Alternate

flow

- flows
  - Not logged in (401)
  - Stack doesn't exist (404)
  - Stack not started or stopping (409)

Stop stack

- Preconditions
  - Stack started
- Test steps
  - ndslabsctl stop <sid>
- Expected results
  - Stack service, replication controller, pods are stopped (200)
- Alternative flow
  - Not logged in (401)
  - Stack does not exist (404)
  - Stack not started (200)

Configuration

List configuration options for spec

- Preconditions
  - Logged in
  - Service exists
- Test case:
  - ndslabsctl list configs <service key(s)>
- Expected results:
  - Possible service configuration options are displayed
- Alternative flow
  - Not logged in (401)
  - Service does not exist (404)
    - Problem: currently throws 500 error
    - Resolved – if one fails, all fail
- Issues:
  - Help text references "service ids", should be "service keys" for consistency
  - Resolved

Get stack service configuration

- Preconditions
  - Logged in
  - Stack exists
- Test case:
  - ndslabsctl get

configs <sid>

- - stack <stack id>
- Expected results:
  - Current service configuration options and settings are displayed

Problem: resulting in 500 now

- Alternative flow
  - Not logged in (401)
  - Stack

doesn't

- - does not exist (404)
    - Problem: currently throws 500 error
    - Works for me

Set stack service configuration option

- Preconditions
  - Logged in
  - Stack exists
  - Config exists
- Test case:
  - ndslabsctl set

<sid>

- - <stack service id> <name> <value>
- Expected results:
  - Specified configuration setting os overriden for the service
- Alternative flow
  - Not logged in (401)
  - Stack doesn't exist (404)
    - Problem: currently throws 500 error
    - Resolved
  - Config doesn't exist or is not overridable (409)
    - Problem: No output given, appears to be no-op?
    - Resolved

Other

View stack service logs

- Preconditions
  - Logged in
  - Stack exists
  - Stack started
  - Config exists
- Test case:
  - ndslabsctl logs <sid>
  - ndslabsctl logs <sid> --lines=<lines>
- Expected results:
  - Logs are displayed
  - Last n log lines are displayed
- Alternative flow
  - Not logged in (401)
  - Stack doesn't exist (404)
    - Problem: currently throws 500 error
  - Stack is not running (404)
    - Problem: No output given, appears to be no-op?

Console access

- Preconditions
  - Logged in
  - Stack exists
  - Stack started
- Test case
  - ndslabsctl console <ssid>
- Alternative flow
  - Not logged in (401)
  - Stack service doesn't exist (404)
  - Stack not running (404)

API Server Startup

Kubernetes token authentication

- Preconditions
  - Kubernetes started with token authentication enabled for kube-apiserver
  - Tokens are
- Test case
  - Configure apiserver.conf Kubernetes section
    - TokenPath=/run/secrets/kubernetes.io/serviceaccount/token
  - Start NDS Labs (e.g., ndslabs-up)
  - Tail the apiserver log and look for successful startup
- Expected results
  - Successful server startup (i.e., you do not see the error "Kubernertes API not availble"
- Alternative flow
  - Invalid token path
  - Token authentication not supported by kube-apiserver

Resource management

Add account with resource limits

Test steps
- ndslabsctl add account -f <account.json>
Expected results
- Prompted for admin password
- Account is added
- kubectl describe quota --namespace=demo returns specified quotas
- kubectl describe limitranges --namespace=demo returns specified limit ranges
Sample Data:

Code Block

{ 
    "id": "demo",
    "name": "demo project",
    "description": "demo project description",
    "namespace": "demo",
    "password": "12345",
	"storageQuota": 10,
	"resourceLimits": {
		"cpuMax": "2",
		"cpuDefault": "1",
		"memMax": "8Gi",
		"memDefault": "100Mi"
	}
}

5/19: OK (NDS-205)

Start stack with resource limits

Prerequisites:
- Test services have been added
- User is logged in
Test steps
- Add "Dataverse" service

Expected results

Services start as expected

For each pod, "kubectl get pod <pod> --namespace=<ns> -o wide" shows a "resources" section, such as:

Code Block
"resources": { "limits": { "cpu": "100m", "memory": "500Mi" }, "requests": { "cpu": "100m", "memory": "500Mi" } }

Resources and limits match those specified on the service spec

5/19: OK (NDS-205)

Start stack: Too big

- Prerequisites:
  - Test services have been added
  - User is logged in
- Test steps
  - Add "Test - Too Big" service
- Expected results
  - Stack service is in error state
- 5/19: OK (NDS-205)

Start stack: Memory Hog

- Prerequisites:
  - Test services have been added
  - User is logged in
- Test steps
  - Add "Test - Memory Hog" service
- Expected results
  - Stack service starts, but quickly enters error state
  - Log shows
    - Reason="OOMKilled"
- 5/19: OK (NDS-205)

Start stack: No such image

- Prerequisites:
  - Test services have been added
  - User is logged in
- Test steps
  - Add "Test - No Image" service
- Expected results
  - Stack service starts, but quickly enters error state
  - Log shows
    - Reason=Failed, Message=Failed to pull image "xyzzy": Error: image library/xyzzy not found
- 5/19: OK (NDS-205)

Start stack: Bad dependency

- Prerequisites:
  - Test services have been added
  - User is logged in
- Test steps
  - Add "Test - Bad dependency" service
- Expected results
  - "No image" service is in error state, "Bad dependency" service is in "stopped" state
  - Log shows
    - Reason=Failed, Message=Failed to pull image "xyzzy": Error: image library/xyzzy not found
- 5/19: OK (NDS-205)

Ingress/LoadBalancer support

Single-node, NodePort, no TLS

This first test case demonstrates how to configure a single-node instance using NodePort without TLS;

- Prerequisites
  - Kubernetes is running
  - Assumes single-node installation
- Test steps
  - Modify apiserver.yaml
    - CORS_ORIGIN_ADDR = http://<public ip>:30000
    - INGRESS = NodePort
    - PREFX=/api
  - Modify gui.yaml
    - APISERVER_HOST = <public ip>
    - APISERVER_PORT = 30001
    - UI_BASE_PATH = /ui
  - kubectl create -f gui.yaml
  - kubectl create -f apiserver.yaml
  - kubectl label nodes 127.0.0.1 ndslabs-role=compute
  - Open browser to: http://<public ip>:30000/ui/
    - Configure basic Clowder, launch stack
    - Confirm Clowder endpoint link points to NodePort (http://publicip:NodePort)

Single-node, LoadBalancer + TLS

- Prerequisites
  - Kubernetes is running
  - Assumes single-node installation
  - Wildcard CNAME entry for *.test.ndslabs.org pointing to single node IP
- Test steps
  - Generate wildcard TLS certificate and key

Code Block

openssl genrsa 2048 > certs/ndslabs.key
openssl req -new -x509 -nodes -sha1 -days 3650 -key certs/ndslabs.key > certs/ndslabs.cert
#[enter *.test.ndslabs.org for the Common Name]
cat certs/ndslabs.cert certs/ndslabs.key > certs/ndslabs.pem
chmod 400 certs/ndslabs.key certs/ndslabs.pem

Multi-node, LoadBalancer + TLS

- Prerequisites
  - Kubernetes is running
  - Assumes multi-node installation
  - Wildcard CNAME entry for *.test.ndslabs.org pointing to load balancer node
  - Assumes loadbalancer and default-backend services already deployed
- Test steps
  - Generate wildcard TLS certificate and key

Code Block

openssl genrsa 2048 > certs/ndslabs.key
openssl req -new -x509 -nodes -sha1 -days 3650 -key certs/ndslabs.key > certs/ndslabs.cert
#[enter *.test.ndslabs.org for the Common Name]
cat certs/ndslabs.cert certs/ndslabs.key > certs/ndslabs.pem
chmod 400 certs/ndslabs.key certs/ndslabs.pem

- - git clone https://github.com/craig-willis/ndslabs-startup
  - cd ndslabs-startup
  - ./ndslabs-up-multinode.sh
    - Domain name: test.ndslabs.org
    - Internal IP: accept default
  - Open browser to:
    - https://www.test.ndslabs.org/ui/ (Accept certificate)
    - Configure basic Clowder, launch stack
    - Confirm endpoint URI link points to ingress host (https://stackservice.test.ndslabs.org)
  - A few kubectl checks
    - kubectl get secrets --namespace=default
      - ndslabs-tls-secret
    - kubectl get ingress --namespace=default
      - ndslabs-ingress with rules for /api and /ui
    - kubectl get secret --namespace=demo
      - demo-tls-secret
    - kubectl get ingress --namespace=demo
      - stack-clowder-ingress with rule for clowder "/"
  - Stop the stack
    - Confirm ingress deleted
      - kubectl get ingress --namespace=demo
  - Delete project
    - ndslabsctl delete project demo
    - Confirm secret deleted
      - kubectl get secret --namespace=demo

Space shortcuts

Page tree

Page History

Versions Compared

Old Version 10

New Version Current

Key

Starting the API server

Global preconditions

Authentication/authorization

Login

Logout

Change password

Project administration

Account administration

Add account (admin only)

Delete account (admin only)

Delete project (admin only)

Get

account details (admin or current project only)

List

accounts

Authentication/authorization

Login

Logout

Service administration

Add service (admin only)

Delete service (admin only)

Change password

Service administration

Get service details

List

services

Volume administration

Add service

Delete service

Volume administration

Create volume

Attach volume

List volumes

Detach volume

Delete volume

Stack administration

Add stack

List stacks

Delete stack

Get stack details

Start stack

Stop stack

Configuration

List configuration options for spec

Get stack service configuration

Set stack service configuration option

Other

View stack service logs

Console access

API Server Startup

Kubernetes token authentication

Resource management

Add account with resource limits

Start stack with resource limits

Start stack: Too big

Start stack: Memory Hog

Start stack: No such image

Start stack: Bad dependency

Ingress/LoadBalancer support

Single-node, NodePort, no TLS

Single-node, LoadBalancer + TLS

Multi-node, LoadBalancer + TLS