Catalog Configuration
Processors
The catalog has a concept of processors to perform catalog ingestion tasks,
such as reading raw entity data from a remote source, parsing it, transforming
it, and validating it. These processors are configured under the
catalog.processors
configuration key.
Static Location Configuration
The simplest configuration for the catalog, as shown in the default
@backstage/create-app
template, is to declaratively add locations pointing to
YAML files with static configuration.
Locations are added to the catalog under the catalog.locations
key:
catalog:
locations:
- type: url
target: https://github.com/backstage/backstage/blob/master/packages/catalog-model/examples/components/artist-lookup-component.yaml
The url
type locations are handled by a standard processor included with the
catalog (UrlReaderProcessor
), so no processor configuration is needed. This
processor does however need an integration to
understand how to retrieve a given URL. For the example above, you would need to
configure the GitHub integration to
read files from github.com.
The locations added through static configuration cannot be removed through the catalog locations API. To remove these locations, you must remove them from the configuration.
Syntax errors or other types of errors present in catalog-info.yaml
files will
be logged for investigation. Errors do not cause processing to abort.
When multiple catalog-info.yaml
files with the same metadata.name
property
are discovered, one will be processed and all others will be skipped. This
action is logged for further investigation.
Local File (type: file
) Configurations
In addition to url locations, you can use the file
location type to bring in content from the local file system. You should only use this for local development, test setups, and example data, not for production data.
You are also not able to use placeholders in them like $text
, $json
or $yaml
. You can however reference other files relative to the current file. See the full catalog example data set here for an extensive example.
Here is an example pulling in the all.yaml
file from the examples folder. Note the use of ../../
to go up two levels from the current execution path of the backend. This is typically packages/backend/
.
catalog:
locations:
- type: file
target: ../../examples/all.yaml
There might be cases where you need to test some file
configurations in a Docker container. In a case like this, as the backend is serving the frontend in a default setup, the path would be from the root. Also, you need to make sure to copy your files into your container.
Using the example above that would look like this:
catalog:
locations:
- type: file
target: ./examples/all.yaml
Integration Processors
Integrations may simply provide a mechanism to handle url
location type for an
external provider or they may also include additional processors out of the
box, such as the GitHub discovery
processor that scans a GitHub organization for
entity descriptor files.
Check the integrations documentation to see what is offered by each integration.
Custom Processors
To ingest entities from an existing system already tracking software, you can also write a custom processor to convert between the existing system and Backstage's descriptor format. This is documented in External Integrations.
Catalog Rules
By default, the catalog will only allow the ingestion of entities with the kind
Component
, API
, and Location
. In order to allow entities of other kinds to
be added, you need to add rules to the catalog. Rules are added either in a
separate catalog.rules
key or added to statically configured locations.
For example, given the following configuration:
catalog:
rules:
- allow: [Component, API, Location, Template]
locations:
- type: url
target: https://github.com/org/example/blob/master/org-data.yaml
rules:
- allow: [Group]
We are able to add entities of kind Component
, API
, Location
, or
Template
from any location, and Group
entities from the org-data.yaml
,
which will also be read as a statically configured location.
Note that if the catalog.rules
key is present it will replace the default
value, meaning that you need to add rules for the default kinds if you want
those to still be allowed.
The following configuration will reject any kind of entities from being added to the catalog:
catalog:
rules: []
Readonly mode
Processors provide a good way to automate the ingestion of entities when combined
with Static Location Configuration or a
discovery processor like
GitHub Discovery. To enforce the usage of
processors to locate entities we can configure the catalog into readonly
mode.
This configuration disables registering and deleting locations with the catalog APIs.
catalog:
readonly: true
Note that any plugin relying on the catalog API for creating, updating, and deleting entities will not work in this mode.
Deleting an entity by UUID, DELETE /entities/by-uid/:uid
, is allowed when using this mode. It may be rediscovered as noted in explicit deletion.
A common use case for this configuration is when organizations have a remote source that should be mirrored into Backstage. To make Backstage a mirror of this remote source, users cannot also register new entities with e.g. the catalog-import plugin.
Clean up orphaned entities
In short, entities can become orphaned through multiple means, such as when a catalog-info YAML file is moved from one place to another in the version control system without updating the registration in the catalog. For safety reasons, the default behavior is to just tag the orphaned entities, and keep them around. You can read more about orphaned entities here.
However, if you do wish to automatically remove the orphaned entities, you can use the following configuration, and everything with an orphaned entity tag will be eventually deleted.
catalog:
orphanStrategy: delete
Processing Interval
The processing loop is
responsible for running your registered processors on all entities, on a certain
interval. That interval can be configured with the processingInterval
app-config parameter.
catalog:
processingInterval: { minutes: 45 }
The value is a duration object, that has one or more of the fields years
,
months
, weeks
, days
, hours
, minutes
, seconds
, and milliseconds
.
You can combine them, for example as { hours: 1, minutes: 15 }
which
essentially means that you want the processing loop to visit entities roughly
once every 75 minutes.
Note that this is only a suggested minimum, and the actual interval may be longer. Internally, the catalog will scale up this number by a small factor and choose random numbers in that range to spread out the load. If the catalog is overloaded and cannot process all entities during the interval, the time taken between processing runs of any given entity may also be longer than specified here.
Setting this value too low risks exhausting rate limits on external systems that are queried by processors, such as version control systems housing catalog-info files.
Stitching strategy
Stitching finalizes the entity. It can be run in two modes:
immediate
- performs stitching in-band immediately when neededdeferred
- performs the stitching asynchronously
It can be configured with the stitchingStrategy
app-config parameter.
catalog:
stitchingStrategy: immediate
For the deferred
mode you can set up additional parameters to further tune the process,
by setting the following parameters:
pollingInterval
- the interval between polling for entities that need stitchingstitchTimeout
- the maximum time to wait for an entity to be stitched
These parameters accept a duration object, similar to the processingInterval
parameter.
catalog:
stitchingStrategy: deferred
pollingInterval: { seconds: 1 }
stitchTimeout: { minutes: 1 };
Subscribing to Catalog Errors
Catalog errors are published to the events plugin: @backstage/plugin-events-node
. You can subscribe to events and respond to errors, for example you may wish to log them.
The first step is to add the events backend plugin to your Backstage application. Navigate to your Backstage application directory and add the plugin package.
yarn --cwd packages/backend add @backstage/plugin-events-backend
Now you can install the events backend plugin in your backend.
backend.add(import('@backstage/plugin-events-backend'));
Logging Errors
If you want to log catalog errors you can install the @backstage/plugin-catalog-backend-module-logs
module.
Install the catalog logs module.
yarn --cwd packages/backend add @backstage/plugin-catalog-backend-module-logs
Add the module to your backend.
backend.add(import('@backstage/plugin-catalog-backend-module-logs'));
This will log errors with a level of warn
.
You should now see logs as the catalog emits events. Example:
[1] 2024-06-07T00:00:28.787Z events warn Policy check failed for user:default/guest; caused by Error: Malformed envelope, /metadata/tags must be array entity=user:default/guest location=file:/Users/foobar/code/backstage-demo-instance/examples/org.yaml
Custom Error Handling
If you wish to handle catalog errors with specific logic different from logging the errors the following should help you get started. For example, you may wish to send a notification or create a ticket for someone to investigate.
Create a backend module that subscribes to the catalog error events. The topic is experimental.catalog.errors
.
import { CATALOG_ERRORS_TOPIC } from '@backstage/plugin-catalog-backend';
import {
coreServices,
createBackendModule,
} from '@backstage/backend-plugin-api';
import { eventsServiceRef, EventParams } from '@backstage/plugin-events-node';
interface EventsPayload {
entity: string;
location?: string;
errors: Error[];
}
interface EventsParamsWithPayload extends EventParams {
eventPayload: EventsPayload;
}
const eventsModuleCatalogErrors = createBackendModule({
pluginId: 'events',
moduleId: 'catalog-errors',
register(env) {
env.registerInit({
deps: {
events: eventsServiceRef,
logger: coreServices.logger,
},
async init({ events, logger }) {
events.subscribe({
id: 'catalog',
topics: [CATALOG_ERRORS_TOPIC],
async onEvent(params: EventParams): Promise<void> {
const event = params as EventsParamsWithPayload;
const { entity, location, errors } = event.eventPayload;
// Add custom logic here for responding to errors
for (const error of errors) {
logger.warn(error.message, {
entity,
location,
});
}
},
});
},
});
},
});
Now install your module.
backend.add(eventsModuleCatalogErrors);