NetReach

Basic description

For this kind of task, kdoctor-controller will generate corresponding agent and other resources. Each agent Pod sends http requests to each other with the request address of each agent's Pod IP, service IP, ingress IP and so on, and obtains the success rate and average latency. It can specify the success condition to determine whether the result is successful or not. Detailed reports can be obtained through the aggregation API.

NetReach example

apiVersion: kdoctor.io/v1beta1
kind: NetReach
metadata:
  name: netreach
spec:
  agentSpec:
    hostNetwork: false
    kind: DaemonSet
    terminationGracePeriodMinutes: 60
  expect:
    meanAccessDelayInMs: 1500
    successRate: 1
  request:
    durationInSecond: 10
    perRequestTimeoutInMS: 1000
    qps: 10
  schedule:
    roundNumber: 1
    roundTimeoutMinute: 1
    schedule: 0 1
  target:
    clusterIP: true
    enableLatencyMetric: false
    endpoint: true
    ingress: true
    ipv4: true
    ipv6: true
    loadBalancer: true
    multusInterface: false
    nodePort: true
status:
  doneRound: 1
  expectedRound: 1
  finish: true
  history:
    - deadLineTimeStamp: "2023-07-28T09:59:12Z"
      duration: 15.462579243s
      endTimeStamp: "2023-07-28T09:58:27Z"
      expectedActorNumber: 2
      failedAgentNodeList:
        - kdoctor-worker
        - kdoctor-control-plane
      failureReason: some agents failed
      notReportAgentNodeList: []
      roundNumber: 1
      startTimeStamp: "2023-07-28T09:58:12Z"
      status: fail
      succeedAgentNodeList: []
  lastRoundStatus: fail

NetReach Definition

Metadata

Fields	Description	Structure	Validation
Name	Name of the NetReach Resource	String	Required

Spec

Fields	Description	Structure	Validation
agentSpec	Task Execution Agent Configuration	agentSpec	Optional
Schedule	Schedule Task Execution	schedule	Optional
Request	Request Configuration for Destination Address	request	Optional
Target	Request Target Settings	target	Optional
Expect	Task Success Condition Judgment	expect	Optional

AgentSpec

Fields	Description	Structure	Validation	Values	Default
Annotation	Annotation of Agent Workload	Map[string]String	Optional
kind	Type of Agent Workload	String	Optional	Deployment, DaemonSet	DaemonSet
deploymentReplicas	The expected number of replicas when the agent workload type is deployment	int	Optional	Greater than or equal to 0	0
Affinity	Agent Workload Affinity	labelSelector	Optional
env	Agent Workload Environment Variable	env	Optional
hostNetwork	Whether or not the agent workload uses the host network	Bool	Optional	True, false	False
Resources	Agent Workload Resource Usage Configuration	Resources	Optional		Limit cpu: 1000m,Memory:1024Mi
terminationGracePeriodMinutes	the minutes after a agent workload completes a task before it terminates	int	Optional	Greater than or equal to 0	60

Schedule

Fields	Description	Structure	Validation	Values	Defaults
roundNumber	Task Execution Rounds	int	Optional	A value greater than or equal to -1 indicates indefinite execution, with -1 representing permanent execution. A value greater than 0 represents the number of rounds to be executed	1
roundTimeoutMinute	Task timeout which needs to be greater than durationInSecond and task execution time	int	Optional	Greater than or equal to 1	60

Request

Fields	Description	Structure	Validation	Values	Defaults
durationInSecond	Duration of request send pressure for each round of tasks which is less than roundTimeoutMinute	int	Optional	Greater than or equal to 1	2
perRequestTimeoutInMS	Timeout per request, not greater than durationInSecond	int	Optional	Greater than or equal to 1	500
QPS	Requests per second per agent	int	Optional	Greater than or equal to 1	5

When using agent requests, all agents will make requests to the destination address, so the actual QPS received by the server is equal to the number of agents multiplied by the set QPS.

Target

Fields	Descriptions	Structures	Validations	Values	Defaults
ClusterIP	Test cluster service's cluster IP	Bool	Optional	True,false	True
Endpoint	Test cluster Pod endpoint	Bool	Optional	True,false	True
multusInterface	Test cluster Pod Multus multi-NIC IP	Bool	Optional	True,false	False
IPv4	Test IPv4	Bool	Optional	True,false	True
IPv6	Test IPv6	Bool	Optional	True,false	False
Ingress	Test Ingress Address	Bool	Optional	True,false	False
nodePort	Test Service Node Port	Bool	Optional	True,false	True
enableLatencyMetric	Statistics demo distribution, which increases memory usage when turned on	Bool	Optional	True,false	False

Expect

Task success condition. If the task result does not meet the expected condition, the task will fail.

Fields	Description	Structures	Validation	Values	Default
meanAccessDelayInMs	The average delay. If the final result exceeds this value, the task will be judged as failed	int	Optional	Greater than or equal to 1	5000
successRate	Success rate of the HTTP request. If the final result is less than this value, the task will fail	Float	Optional	0-1	1

status

Fields	Description	Structures	Values
doneRound	Number of completed task rounds	int
expectedRound	Number of rounds expected to be performed	int
Finish	Whether the task is complete or not	Bool	True, false
lastRoundStatus	lastRoundStatus	String	Notstarted, on-going, succeed, fail
History	Task History	Element is history array

History

Fields	Description	Structure	Values
roundNumber	Task Round Number	int
Status	Task Status	String	Notstarted, on-going, succeed, fail
startTimeStamp	Start of the current round of tasks	String
endTimeStamp	End of the current round of tasks	string
duration	Execution time of the current round of tasks	string
deadLineTimeStamp	Deadline of the current round of tasks	string
failedAgentNodeList	Agent whose tasks failed	Array of elements as string
succeedAgentNodeList	Agent whose task succeeded	Array of elements as string
notReportAgentNodeList	Agent who did not upload a task report	Array of elements as string