Skip to content

14.1 - Introduction to JSONPath

14.1.1 - Introduction to YAML

  • Ansible playbooks format similar to XML and JSON
  • Used to express data
  • Data in YAML files, at its most basic form is a series of Key-Value pairs, separated by a colon
  • For an array:
- entry1: value1
- entry2: value2
- entry3: value3
  • For a dictionary:
item:
  property1: value1
  property2: value2
  • Note: 2 spaces determine what properties come under what iteM
  • Can have lists containing dictionaries containing lists
  • To store information about different properties of a single item, use a dictionary
  • If the properties need further segregation, can use dictionaries within dictionary items
  • For multiple items of the same type, use an array

  • To store information for multiple items, expand each array item to a the dictionary for each

  • Key Notes:

  • Dictionary = Unordered
  • Lists / Arrays = Ordered
  • Comments added via #

14.1.2 - Introduction to JSONpath

  • Yaml vs Json:
  • Data can be expressed via both
  • To segregate data, methods differ:
    • YAML - Indentations and -'s
    • JSON - Indentation and {} for dictionaries, [] for array items
  • JSON data can be queried via JSON Path
  • To select an item, specify it as <item>
  • For a dictionary property: <item>.<property>
  • Note: Anything within { } denotes a dictionary
  • The top-level dictionary, which isn't named, is denoted by a $
  • A typical query is $.item.property etc
  • Any output from a JSONpath query is an array []
  • To query an array/list, use square brackets to reference the position, with positions starting at [0].
  • E.g. 1st element: $.[0]
  • For dictionaries in lists, combine the query use for all
  • For criteria: $.[CRITERIA] e.g.:
  • $.[?( @ > 40 )]
  • In this case, ?() signals to use a filter, the @ symbol signifies "each item"
  • Queries could be used to accommodate for changing positions
  • E.g. $.car.wheels[?(@.location == "rear-right")].model
    • Only entries that satisfy the citeria "location = rear-right" will be returned.

14.1.3 - JSONPath: Wildcard

  • Denoted by *, meaning "any", can be used to retrieve all/any properties of a particular dictionary
  • Can swap as a value when referencing an array position i.e. []

14.1.4 - JSONPath: Advanced List Queries

  • To get all names in an array's particular range, add [x:y], where x is the first element's position, y is the end position of the range +1
  • To iterate over a step, insert: [x:y:z], where z is the step rate
  • To get the last item: [-1:0]

14.1.5 - Advanced Kubectl Commands: Kubectl and JSONPath

  • Prerequisites:
  • JSONPath for beginners
  • JSONPath Practice tests - Kodekloud: General use and for Kubernetes
  • Why JSON Path?
  • Large data sets involved with production-grade clusters
    • 100s of nodes
    • 1000s of PODS, Deployments, ReplicaSets etc
  • More often than not, will want to quickly print information for large numbers of resources and particular information
  • Kubectl commands invokes the APIServer, which obtains the information requested in a JSON format and is redisplayed in a readable format by Kubectl
  • Particularly noticeable in kubectl get commands
  • For additional information, add -o wide flag
  • Example:
  • Suppose we want to see the following:
    • CPU count
    • Taints and Tolerations
    • Pod name and Images
  • There is no built-in kubectl command for this, but we can use kubectl and JSON path in combination to get the particular fields
  • To use JSON Path in Kubectl, consider the 4 steps:
  • Identify the kubectl command required e.g. kubectl get pods
  • Familiarize yourself with the JSON format output: add the -o json flag
  • From the JSON output, figure out the custom query you'd want to apply, e.g. for container images of a particular pod: .items[*].spec.containers[*].image
  • Combine the kubectl command with the JSON query i.e. kubectl get pods -o=jsonpath='{JSON_PATH_QUERY}'
  • Note:
  • For multiple queries, within the '', encompass each query by {}
  • To format this, use any of the following:
    • {"\n"} - New line (Add in between queries)
    • {"\t"} - Tab
  • Looping through ranges: '{range .items[*]} {Queries} {end}'
  • Can print custom columns via: -o=custom-columns=<COLUMN_NAME>:<JSON PATH>, <COLUMN>:<JSON PATH>
  • Recommended to view full query first then forming the JSON query
  • Use --sort-by property where necessary e.g. --sort-by=.metadata.name