This repository contains a script to quickly set up demo data for Kibana development and demonstration purposes. Before running the script, make sure you are in the Kibana directory. You can run the script on any Kibana dev instance that uses the default credentials (elastic:changeme) with the following command:
curl -sSL https://elastic.github.io/kibana-demo-data | shThe script waits for the Kibana dev instance to be up and running, and then installs several sets of demo data automatically.
For convenience, you can start Kibana and ingest demo data in one command:
yarn start & curl -sSL https://elastic.github.io/kibana-demo-data | sh The script installs several demo data sets, each with an associated variable name that can be used for selective installation:
- Kibana Sample Data (
sample) - Custom Sample Data from the data folder of this repository (
custom) - Basic Security Sample Data (
security) - Observability (O11y) Sample Data (
o11y) - Makelogs Sample Data (
makelogs) - Searchkit Data (
searchkit) - Metrics Sample Data (
metrics) - Uses simian-forge to generate host metrics data - Edge Case Data with 50,000 dynamically generated unique fields across 10 indices (5,000 unique fields each) and 10 test documents for testing massive field scenarios (
edgecase) - Discover Demo Data (
discover) - A rich, 11-tab Discover session (classic, classic ad hoc data view, field statistics, ES|QL, ES|QL group by, ES|QL view, metrics experience, traces experience, patterns, change point, data sources) plus a dashboard embedding that session, backed by real data, created in four demo spaces, one per solution type:demo-classic(classic),demo-o11y(oblt),demo-security(security),demo-search(es). The "Data Sources" tab queries a real public dataset (AWS Open Data Bitcoin blockchain, in S3) registered via ES|QL Data Federation; the "ES|QL View" tab queries a reusable named ES|QL view. Every tab and the dashboard store an explicit time range, so they reliably show data regardless of the viewer's default time range. The spaces are created only if they don't already exist, and the Discover session/dashboard IDs are deterministic (demo-discover-session-<space-id>/demo-discover-dashboard-<space-id>) so automated tests can reference them directly and re-running the script is idempotent. Requires a recent Kibana checkout (usesnode scripts/synthtracefor traces data and the Discover session/Dashboard APIs), and an Elasticsearch version with ES|QL Data Federation and view support.
If you want to install only specific subsets of data, use the following command with the appropriate data set options (e.g., sample, custom, security, o11y, makelogs, metrics, edgecase or discover):
curl -sSL https://elastic.github.io/kibana-demo-data | sh -s <data_set>For example, to install Kibana Sample Data and Custom Sample Data together, run:
curl -sSL https://elastic.github.io/kibana-demo-data | sh -s sample customTo install Edge Case Data with 50,000 dynamically generated unique fields across 10 indices and 10 test documents:
curl -sSL https://elastic.github.io/kibana-demo-data | sh -s edgecaseTo install the Discover Demo Data (run from a Kibana checkout):
curl -sSL https://elastic.github.io/kibana-demo-data | sh -s discoverThe Discover installer is a standalone Node.js module (Node 18+, no npm dependencies). The shell script is a small launcher for the same module. Run either command from your Kibana checkout so synthtrace can seed traces:
node /path/to/kibana-demo-data/scripts/discover.mjs --es-url http://localhost:9200 --kibana-url http://localhost:5601
sh /path/to/kibana-demo-data/scripts/discover.sh --skip-tracesUse --help to list options. Connection flags (--es-url, --kibana-url,
--username, --password, --dev-prefix) override the corresponding
KIBANA_DEMO_ES_URL, KIBANA_DEMO_KIBANA_URL, KIBANA_DEMO_USERNAME,
KIBANA_DEMO_PASSWORD, and KIBANA_DEMO_DEV_PREFIX environment variables.
Kibana's development base path is detected automatically unless supplied explicitly.
--skip-traces skips synthtrace; the remaining data and saved objects are still created.
The Change Point tab uses 100 five-minute buckets of log traffic, with a sustained increase from roughly 10 to 40 requests per bucket halfway through the series. Its query sorts the buckets chronologically and uses a fixed time range matching the seeded logs, so the change remains visible when the session is reopened later. Re-run the installer to refresh existing demo data and saved tabs.
Flags also work when piping the launcher to sh. Export environment variables
before the pipeline so the installer inherits them:
export KIBANA_DEMO_PASSWORD=changeme
curl -fsSL https://elastic.github.io/kibana-demo-data/scripts/discover.sh | sh -s -- --skip-tracesFor a PR branch, point the launcher at the matching JavaScript source as well:
export KIBANA_DEMO_SCRIPTS_URL=https://raw.githubusercontent.com/elastic/kibana-demo-data/add-discover-demo-dataset/scripts
curl -fsSL "$KIBANA_DEMO_SCRIPTS_URL/discover.sh" | sh -s -- --skip-tracesA local launcher uses its adjacent discover.mjs; a piped launcher streams
it directly into Node without saving a JavaScript file. The streaming launcher
requires Bash to propagate failures from either curl or Node. The top-level dataset selector
(sh -s discover) continues to support environment variables; pass Discover-specific
flags to scripts/discover.sh or scripts/discover.mjs directly.
The streaming launcher executes remote code with your user's permissions. Use a
trusted HTTPS source for KIBANA_DEMO_SCRIPTS_URL; for reproducible runs, replace
the branch in the raw GitHub URL with a reviewed commit SHA. The launcher only
accepts HTTPS for the module and its redirects. This does not verify a signature
or checksum, and the initial curl ... | sh command is still a separate trust step.
Use a development cluster: the installer recreates demo-discover-logs and
demo-discover-metrics and updates the named demo objects. Use HTTPS for remote
Elasticsearch/Kibana connections; the HTTP defaults are for localhost development.
Pass credentials through environment variables to avoid shell history. When traces
are enabled, synthtrace still receives credentials in its --target argument,
which may be visible to local process inspection; use --skip-traces on shared
machines when that exposure is unacceptable.
Run the installer regression checks without contacting a cluster:
node --test scripts/discover.test.mjs