During an incident, the on-call engineer opens a Dynatrace Managed dashboard and finds twenty tiles with no hierarchy: default charts, hosts from every environment mixed together, nothing about SLOs. Ten minutes go on finding the right view, and MTTR grows by the same amount.
This guide shows how to build useful classic dashboards on Dynatrace Managed: native tiles, Data Explorer charts with exact builtin: metric selectors, management zone filtering, then JSON export and import through the Configuration API so you can version them.
Goal: leave with three complete dashboards, described tile by tile (application health, hosts, operations view), and a way to provision them without clicking. The content targets Dynatrace Managed; classic dashboards and the Configuration API also exist in Dynatrace SaaS, where only the URL changes.
Prerequisites
- A Dynatrace Managed environment with OneAgent deployed on the hosts and services you care about.
- Management zones in place (see management zones and tagging): they filter every dashboard.
- For the API: a token with the
ReadConfigscope (read and export) andWriteConfig(create, update, validate). - The environment URL in Managed format:
https://{your-domain}/e/{your-environment-id}. - Existing SLOs if you want the SLO tile from example 3.
The building blocks of a Dynatrace Managed dashboard
A classic dashboard is a grid of tiles. Each tile has a type (tileType), a position (bounds: top, left, width, height in pixels) and, if needed, its own filter (tileFilter) that overrides the dashboard filter.
| Tile type | Use |
|---|---|
HEADER |
Section title, to structure the grid |
HOSTS, SERVICES, APPLICATIONS |
Overall health of an entity family, filterable by management zone |
DATA_EXPLORER |
Chart built in Data Explorer from a metric selector |
OPEN_PROBLEMS |
Open problems detected by Davis |
SLO |
Status and error budget of an SLO |
MARKDOWN |
Free text: runbook link, on-call contacts |
At dashboard level, dashboardMetadata.dashboardFilter sets the default management zone and timeframe (for example l_2_HOURS or l_72_HOURS). The dynamicFilters.filters field adds filters readers can change on the fly, such as SERVICE_TYPE, HOST_TAG_KEY:<tag> or KUBERNETES_CLUSTER.
Building a tile with Data Explorer
- Open Data Explorer from the menu and search for the metric (for example CPU usage %).
- Pick the aggregation, the split (Split by) and filters, or switch to the advanced editor and type the selector directly.
- Choose the visualisation (graph, single value, table, top list), then Pin to dashboard and the target dashboard.
- On the dashboard, edit the tile to give it a meaningful name and, if needed, its own management zone.
The selector follows Metrics v2 syntax: metric key, then transformations chained with :. A documented example: builtin:host.cpu.user:sort(value(max,descending)):limit(10) for the ten busiest hosts. To go further, see the data exploration module.
Example 1: application health
Goal: know within ten seconds whether an application’s services meet their SLIs (latency, errors, throughput). Dashboard filter: the application’s management zone, timeframe Last 2 hours.
| Tile | Type | Selector or setting |
|---|---|---|
| “Services” | HEADER |
Section title |
| Service health | SERVICES |
Filter: application management zone |
| Median response time | DATA_EXPLORER, graph |
builtin:service.response.time:splitBy("dt.entity.service"):median |
| p90 response time | DATA_EXPLORER, graph |
builtin:service.response.time:splitBy("dt.entity.service"):percentile(90) |
| Failure rate | DATA_EXPLORER, top list |
builtin:service.errors.total.rate:splitBy("dt.entity.service"):avg:sort(value(avg,descending)):limit(10) |
| Throughput | DATA_EXPLORER, graph |
builtin:service.requestCount.total:splitBy("dt.entity.service"):auto |
Key lines: builtin:service.response.time is stored in microseconds; Data Explorer converts the display. The failure rate builtin:service.errors.total.rate is already a percentage and only accepts auto and avg. The counter builtin:service.requestCount.total only accepts auto and value: do not apply avg to it. The p90 reflects real user experience better than a mean.
Example 2: hosts
Goal: spot saturated hosts before they raise a problem. Filter: the infrastructure team’s management zone, timeframe Last 24 hours.
| Tile | Type | Selector or setting |
|---|---|---|
| Host health | HOSTS |
Filter: management zone |
| Top 10 CPU | DATA_EXPLORER, top list |
builtin:host.cpu.usage:splitBy("dt.entity.host"):avg:sort(value(avg,descending)):limit(10) |
| Top 10 memory | DATA_EXPLORER, top list |
builtin:host.mem.usage:splitBy("dt.entity.host"):max:sort(value(max,descending)):limit(10) |
| Fullest disks | DATA_EXPLORER, table |
builtin:host.disk.usedPct:splitBy("dt.entity.host"):max:sort(value(max,descending)):limit(10) |
| Availability | DATA_EXPLORER, graph |
builtin:host.availability (default aggregation) |
The three resource metrics are percentages and accept avg, max and min. Use max for memory and disk: an average hides the peak that kills a process. In Data Explorer, add a threshold at 85% to colour breaches.
Example 3: operations view of the environment
Goal: the on-call landing page. No management zone filter, or a “Production” zone, timeframe Last 2 hours.
| Tile | Type | Setting |
|---|---|---|
| Open problems | OPEN_PROBLEMS |
Filter: Production management zone |
| Front-end availability SLO | SLO |
SLO assigned to the tile, with error budget |
| API latency SLO | SLO |
SLO assigned to the tile |
| Errors, all services | DATA_EXPLORER, single value |
builtin:service.errors.total.rate:splitBy():avg |
| Unavailable hosts | DATA_EXPLORER, top list |
builtin:host.availability:splitBy("dt.entity.host") |
| Runbook | MARKDOWN |
Link to the procedure and the on-call channel |
The OPEN_PROBLEMS tile shows what Davis has opened; it is the visual counterpart of the notifications sent by your alerting profiles. To watch the Managed cluster itself (nodes, storage, ActiveGates), keep a separate dashboard and rely on cluster health monitoring.
Exporting and importing a dashboard as JSON
Step 1: list and export
export DT_URL="https://{your-domain}/e/{your-environment-id}"
# List (ReadConfig scope): returns id, name, owner
curl -s "$DT_URL/api/config/v1/dashboards" \
-H "Authorization: Api-Token $DT_TOKEN" | jq '.dashboards[] | {id, name, owner}'
# Export one dashboard
curl -s "$DT_URL/api/config/v1/dashboards/<id>" \
-H "Authorization: Api-Token $DT_TOKEN" > app-health.json
The list returns every dashboard, whatever the access rights in the UI. The best way to get the exact structure of a DATA_EXPLORER tile is to export it after building it in the UI.
Step 2: a minimal valid JSON
{
"dashboardMetadata": {
"name": "Operations - Production",
"owner": "sre-team",
"shared": true,
"dashboardFilter": {
"timeframe": "l_2_HOURS",
"managementZone": { "id": "3438779970106539862", "name": "Production" }
}
},
"tiles": [
{
"name": "Hosts", "tileType": "HEADER", "configured": true,
"bounds": { "top": 0, "left": 0, "width": 304, "height": 38 },
"tileFilter": {}
},
{
"name": "Host health", "tileType": "HOSTS", "configured": true,
"chartVisible": true,
"bounds": { "top": 38, "left": 0, "width": 304, "height": 304 },
"tileFilter": {}
}
]
}
dashboardMetadata (with name and owner) and tiles are required. Before importing an export, remove id and the metadata block, which are for debugging: Dynatrace assigns the ID.
Step 3: validate, then create or update
# Validate without creating (WriteConfig scope): 204 = valid
curl -s -o /dev/null -w "%{http_code}\n" -X POST \
"$DT_URL/api/config/v1/dashboards/validator" \
-H "Authorization: Api-Token $DT_TOKEN" \
-H "Content-Type: application/json" -d @app-health.json
# Create
curl -s -X POST "$DT_URL/api/config/v1/dashboards" \
-H "Authorization: Api-Token $DT_TOKEN" \
-H "Content-Type: application/json" -d @app-health.json
# Update an existing dashboard
curl -s -X PUT "$DT_URL/api/config/v1/dashboards/<id>" \
-H "Authorization: Api-Token $DT_TOKEN" \
-H "Content-Type: application/json" -d @app-health.json
A 400 returns an ErrorEnvelope whose constraintViolations point at the faulty field. Management zone IDs are specific to each environment: adjust them when moving between environments.
Sharing and permissions
The shared field makes the dashboard visible to other users; sharingDetails (linkShared, published) controls link sharing and publishing. The owner keeps edit rights. A shared dashboard does not grant access to data: each reader only sees what their management zone permissions allow. Dashboards containing remote environment tiles can only be modified through the API by an admin (403 otherwise).
Provisioning with Terraform or Monaco
The Dynatrace Terraform provider manages classic dashboards with the dynatrace_json_dashboard resource, which takes the dashboard JSON. Its export tool excludes them by default: name them explicitly. In Monaco, dashboards are a special configuration type; since Monaco CLI 1.208, owner must be non-null and sharingDetails is no longer supported. Check the provider documentation for the exact syntax of your version.
How to check it works
- The validator call returns
204; the POST returns the ID of the new dashboard. - The list GET shows the dashboard with the right
owner. - In the UI, Dashboards menu: the dashboard appears, the management zone filter shows at the top, and each Data Explorer tile shows data for the selected timeframe.
- Signed in with an account limited to another management zone, the shared dashboard only shows permitted entities.
Best practices
- One dashboard, one question: “is the application healthy?”, not “everything about everything”.
- Put SLIs top left, details below.
- Prefer percentiles and maximums to averages for latency and saturation.
- Cap top lists at 10 entities with
limit(10): more is noise. - Version the JSON and apply it through a pipeline; a hand-edited dashboard drifts quickly.
- Add a Markdown tile with the runbook: it shortens time to first action.
Frequently asked questions
Are Dynatrace Managed dashboards the same as in SaaS?
Classic dashboards and the /api/config/v1/dashboards API exist in both. SaaS also offers the newer Dashboards app built on Grail and DQL; it does not apply to Managed.
Why is my Data Explorer tile empty?
Usually the tile or dashboard management zone excludes the entities, or the chosen aggregation is not supported by the metric. Test the selector alone in Data Explorer, then reapply the filter.
Can I copy a dashboard between two Managed environments?
Yes: GET export, remove id and metadata, replace management zone, SLO and entity IDs, then validate and POST to the target environment.
Conclusion
A good Dynatrace Managed dashboard is built around SLIs: latency, errors, throughput, saturation and open problems. With exact selectors, well-defined management zones and versioned JSON, your three dashboards become reproducible and useful when on-call needs them. Official reference: Dynatrace Managed Dashboards API.
span.observer is an independent community blog, not affiliated with Dynatrace LLC. Dynatrace is a trademark of its owner. Examples are provided for guidance: check them against the official documentation for your version.