Datadog

Query and interact with Datadog using the dogshell CLI.

When to Use

Use this skill when asked to:

Query metrics or APM data from Datadog
Check monitor status or SLOs
Investigate performance issues using Datadog
List or review dashboards
Create or update dashboards, notebooks, or monitors

Setup

If dogshell isn’t installed, set it up via mise:

    mise use python
pip install datadog

Reading Data

Command	Purpose
`dog metric query`	Query metric timeseries
`dog monitor show <id>`	View monitor details
`dog monitor show_all`	List all monitors
`dog dashboard show_all`	List dashboards
`dog downtime show_all`	List scheduled downtimes
`dog tag show <host>`	Show tags for a host

Creating/Updating Resources

Command	Purpose
`dog monitor post`	Create a new monitor
`dog monitor update <id>`	Update existing monitor
`dog dashboard post`	Create a new dashboard
`dog downtime post`	Schedule a downtime

Common Tasks

Query a Metric

    dog metric query --start $(date -v-1H +%s) --end $(date +%s) "avg:trace.rack.request.duration{env:production}"

  

Time range options:

Last hour: --start $(date -v-1H +%s) --end $(date +%s)
Last day: --start $(date -v-1d +%s) --end $(date +%s)
Last week: --start $(date -v-7d +%s) --end $(date +%s)

Check Monitor Status

    dog monitor show_all
dog monitor show 12345

Create a Monitor

    dog monitor post "metric alert" "avg(last_5m):avg:trace.rack.request.duration{env:production} > 1" --name "High Request Latency" --message "Request latency exceeded 1s"

  

Useful Metric Queries

Rails/Ruby performance metrics:

avg:trace.rack.request.duration{env:production} - Request duration
avg:trace.active_record.sql.duration{env:production} - DB query time
sum:trace.rack.request.hits{env:production}.as_count() - Request count
avg:trace.sidekiq.job.duration{env:production} - Background job duration
p95:trace.rack.request.duration{env:production} - 95th percentile latency

Infrastructure:

avg:system.cpu.user{*} - CPU usage
avg:system.mem.used{*} - Memory usage
avg:kubernetes.cpu.usage.total{*} - K8s CPU
avg:kubernetes.memory.usage{*} - K8s memory

Query Syntax

Metrics follow the pattern: aggregation:metric.name{tags}

Aggregations: avg, sum, min, max, p50, p75, p90, p95, p99

Tags filter results: {env:production,service:api} or {*} for all.

Use .as_count() or .as_rate() to change how counters are displayed.