如何在Kibana Docker镜像与RPM包中预置现成仪表盘？

Hey there! Let's tackle your two Kibana packaging questions—this is a common use case for distributing pre-configured analytics tools, and I’ve worked through this a bunch of times. Here’s a step-by-step breakdown:

The core idea is to pre-seed Kibana with your saved dashboards/reports (called "saved objects" in Kibana terms) so they load automatically on first launch. Below are the workflows for both Docker and RPM:

Docker Image Approach #

Step 1: Export Your Prebuilt Dashboards #

First, set up a temporary Kibana instance, create all your desired dashboards, visualizations, and index patterns. Then:

• Go to Stack Management > Saved Objects > Export
• Check the Include related objects box (critical to capture dependencies like visualizations or index patterns)
• Export the bundle as a file named prebuilt-dashboards.ndjson

Step 2: Build a Custom Docker Image #

Create a Dockerfile based on the official Kibana image, and add your saved objects + configuration to auto-import them:

# Use the exact Kibana version you used to export the objects
FROM docker.elastic.co/kibana/kibana:8.11.3

# Copy your saved objects bundle into the image
COPY prebuilt-dashboards.ndjson /usr/share/kibana/data/
# Copy a custom kibana.yml with import settings
COPY kibana.yml /usr/share/kibana/config/

Update your kibana.yml to enable automatic import on startup:

# Enable saved object imports
xpack.savedObjects.import.enabled: true
# Path to your prebuilt bundle
xpack.savedObjects.import.path: /usr/share/kibana/data/prebuilt-dashboards.ndjson
# Overwrite existing objects (safe for first-time launches)
xpack.savedObjects.import.overwrite: true

# Keep other default Kibana settings here, or copy from the official image's kibana.yml

Step 3: Build & Test the Image #

Run these commands to build and verify your custom image:

docker build -t my-custom-kibana .
docker run -p 5601:5601 my-custom-kibana

Once Kibana starts, open http://localhost:5601—your dashboards should be ready to use immediately.

RPM Package Approach #

Step 1: Export Your Saved Objects #

Same as the Docker workflow: export your dashboards and dependencies as prebuilt-dashboards.ndjson.

Step 2: Create a Custom RPM Spec File #

You’ll build a custom RPM that layers your prebuilt objects and configuration on top of the official Kibana RPM. Here’s a sample spec file:

Name: my-custom-kibana
Version: 8.11.3
Release: 1%{?dist}
Summary: Kibana with preconfigured dashboards
License: Elastic License
URL: https://elastic.co
Source0: https://artifacts.elastic.co/downloads/kibana/kibana-%{version}-x86_64.rpm
Source1: prebuilt-dashboards.ndjson
Source2: kibana.yml

BuildArch: x86_64
Requires: kibana = %{version}

%description
Custom Kibana package with prebuilt dashboards ready to use on first launch.

%prep
# Extract files from the official Kibana RPM
rpm2cpio %{SOURCE0} | cpio -idmv

%install
# Copy saved objects to Kibana's data directory
install -Dm 644 %{SOURCE1} %{buildroot}/usr/share/kibana/data/prebuilt-dashboards.ndjson
# Override the default kibana.yml with our import-enabled config
install -Dm 644 %{SOURCE2} %{buildroot}/etc/kibana/kibana.yml

%files
/usr/share/kibana/data/prebuilt-dashboards.ndjson
/etc/kibana/kibana.yml

Step 3: Prepare the Custom kibana.yml #

Use the same import configuration as the Docker workflow, plus any default Kibana settings you need:

xpack.savedObjects.import.enabled: true
xpack.savedObjects.import.path: /usr/share/kibana/data/prebuilt-dashboards.ndjson
xpack.savedObjects.import.overwrite: true

# Add other default settings from /etc/kibana/kibana.yml here

Step 4: Build & Install the RPM #

Use rpmbuild to package your custom Kibana:

# Place source files in the rpmbuild SOURCES directory
cp prebuilt-dashboards.ndjson kibana.yml ~/rpmbuild/SOURCES/
cp kibana-%{version}-x86_64.rpm ~/rpmbuild/SOURCES/

# Build the RPM
rpmbuild -ba my-custom-kibana.spec

Install and test the package:

rpm -ivh ~/rpmbuild/RPMS/x86_64/my-custom-kibana-8.11.3-1.el8.x86_64.rpm
systemctl start kibana

Navigate to Kibana in your browser—your prebuilt dashboards will be waiting.

• Version Lock: Always use the exact same Kibana version for exporting saved objects and building your package/image. Saved objects often break across major versions.
• Dependency Checks: Double-check that you exported all related objects (index patterns, searches, visualizations) — missing these will leave your dashboards empty or broken.
• Clean Environment Testing: Test your image/RPM in a fresh environment to ensure no hidden dependencies are missing.

内容的提问来源于stack exchange，提问作者sillydeedee