Medium Datadog Utils

A Go library for Devops to retrieve important information from Datadog components

Moving Parts & Functions

Files:

• datadog_api.go -- Imports the Datadog API client, & uses it to make calls to our Datadog instances
• helpers.go -- A home for helper functions that allow us to work through some of the idiocyncracies in Datadog's API
• main.go -- Self-explanatory. It takes all the user inputs & calls all the things.

---

Helper Functions:

• CheckIgnoredWidgetDefs -- Iterates through a known list (as of 11-May 2021) of WidgetDefinition that we do not care about and that do not have the Requests array (or do not have the Q within Requests)
• SearchWidgetExpressionsArray -- Iterates through []string and checks whether a string is contained within each string
  • Takes two arguments:
    • expressionsArray -- Pointer to a []string
    • searchStr -- string that will be used in the search
  • We just use strings.Contains() for the search, but using this helper allows us to target an entire dashboard's array of expressions at once.

---

Non-Helper Functions:

• createApiClient() -- Called on init(), creates an instance of *datadog.APIClient that we can reuse across API-related methods. Returns:
  • APIClient
  • context (type -- context.Context)
• FindMetric() -- Performs a series of steps to fetch all dashboards, all widgets within each dashboard, and every string expression therein.
  • Takes a single argument: string metric (what we're searching for)
  • Returns a map[string][]string. Each key is the dashboardID, and the value is an array composed of strings: all of the expressions that contain a hit against the metric we searched for.
  • We concurrently fetch each dashboard object's Widgets array, and asynchronously iterate through each dashboard's widgets to retrieve the expressions.
  • Then, we [also concurrently] iterate through each dashboard's expressions to see if the metric name is contained therein
• GetDashboards() -- Retrieves a list of all dashboards in Datadog, and returns a map of dasboard ID's, with a nested map containing the following for each:
  • dashboard Title
  • dashboard Author
  • dashboard Last Modified
• GetDashboardDetails() -- Takes a single dashboard ID (in string format) and fetches info for that dashboard.
  • Returns map[string]string
  • The map contains:
    • Author
    • ID (dashboard ID)
    • URL
    • Widgets (a count of the number of widgets within the dashboard)
• GetWidgetExpressions() -- Takes a pointer to an array of datadog.Widget type, and will extract Expressions. These are string representations of the individual queries used in some widgets (like Timeseries).
  • Expressions are self-explanatory. These are the q portion within each request in the widget JSON (you can see those in Datadog)
  • Each of the above is stored in a map with the widget ID as the key, mapped to an array of either the expression string, or the requests type.

---

Example Usage

❯ go run datadog_utils --help
Usage of /var/folders/9d/w68r5sdd787gg_1djyk7jmkm0000gp/T/go-build1368975407/b001/exe/datadog_utils:
  -dash-details string
    	String: Get details for a specific dashboard by ID
  -dash-list
    	Bool: will list all DataDog dashboard IDs
  -find-metric string
    	String: Search our DataDog dashboards for the provided metric name to see if it is in use

1. --find-metric -- go run datadog_utils --find-metric="server.response.sent" Example output looks like this:

{
  "abc-n3k-26j": [
    "sum:server.response.sent{route:paymentwebhook.handlebraintreeevent,status-code:200,$env}.as_count()",
    "sum:server.response.sent{route:paymentwebhook.handlebraintreeevent,status-code:401,$env}.as_count()",
    "sum:server.response.sent{route:paymentwebhook.handlebraintreeevent,status-code:500,$env}.as_count()"
  ],
  "fzi-6va-aag": [
    "sum:server.response.sent{env:prod,route:mediummembership.addmembership,$provider,$environment} by {status-code}.as_count()",
    "sum:server.response.sent{env:prod,route:mediummembership.addmembership,status-code:200,$provider,$environment}.as_count()/sum:server.response.sent{env:prod,route:mediummembership.addmembership,$provider,$environment}.as_count()*100"
  ],
  "n43-std-5sh": [
    "sum:server.response.sent{service:medium-legacy,$env,$controller,$route,status-category:2xx,$status}.as_count()",
    "sum:server.response.sent{service:medium-legacy,$env,$controller,$route,status-category:3xx,$status}.as_count()",
    "sum:server.response.sent{service:medium-legacy,$env,$controller,$route,status-category:4xx,$status}.as_count()",
    "sum:server.response.sent{service:medium-legacy,$env,$controller,$route,status-category:5xx,$status}.as_count()"
  ]
}
{
  "Author": "adrian@medium.com",
  "ID": "abc-n3k-26j",
  "URL": "/dashboard/abc-n3k-26j/braintree",
  "Widgets": "14"
}
{
  "Author": "fiona@medium.com",
  "ID": "fzi-6va-aag",
  "URL": "/dashboard/fzi-6va-aag/trial-purchase-real-time-monitoring",
  "Widgets": "6"
}
{
  "Author": "lyra@medium.com",
  "ID": "n43-std-5sh",
  "URL": "/dashboard/n43-std-5sh/medium2",
  "Widgets": "25"
}

2. --dash-details -- `go run datadog_utils --dash-details="abc-n3k-26j"

Here are the high-level details: 
{
  "Author": "adrian@medium.com",
  "ID": "abc-n3k-26j",
  "URL": "/dashboard/abc-n3k-26j/braintree",
  "Widgets": "14"
}`

3. --dash-list -- go run datadog_utils --dash-list

• I'm not dumping output here, because it's massive and includes ALL dashboards.

---

Additional References & Documentation

• Datadog API documentation
• Dashboard Struct Breakdown
• Widget Struct Breakdown

---

Hierarchy of Datadog Dashboard (As Related to Fetching Expressions in Widgets)

Note: WidgetDefinition actually is a struct that contains every definition type. Datadog just adds nil to the ones not applicable to a specific widget. You can see all of them here, but ultimately this implementation is what necessitated the need for the CheckIgnoredWidgetDefs helper function. Each widget definition has its own Request Type. You can see GetWidgetExpressions() how we fetch the various parts of the Widget & nested info.

graph TD
    A[Dashboard]-->B[Widgets]
    B-->C[WidgetDefinition]
    C-->D[SpecificDefinitionType]
    D-->E[Requests Array]
    E-->F[SpecificRequestType]
    F-->G[Q]

Loading

---

Changelog

2021 - May 19 -- Initial Completion outside of Medium repo, for personal convenience / use