PuppetDB 4.0: Upgrading from version 3
Included in Puppet Enterprise 2016.1.
This page describes changes that users will need to be aware of to make their code compliant with the changes in PuppetDB 3.0. Most of these changes are observable in released versions of the v4 query API, which has been marked ‘experimental’ since 2.0.0 but is the only API available in 3.0.
Note that this document focuses on API changes only, and only includes changes leading up to the release of 3.0. For a more complete description of the changes listed and changes in versions subsequent to 3.0, see the release notes.
Each change below is marked with the corresponding release version.
Changes affecting all endpoints
(3.0) The query API has been moved from
/pdb/query, so http://localhost:8080/v4/facts has become http://localhost:8080/pdb/query/v4/facts.
/commandsendpoint has been moved to its own API at
/pdb/cmd, so http://localhost:8080/v3/commands has become http://localhost:8080/pdb/cmd/v1.
server-timeendpoints have been moved out of the query API and into to a separate
metadataAPI. They are now accessed at
(3.0) All previously dash-separated field names (such as “receive-time”), subquery operators (such as “select-facts”), and query parameters (such as “order-by”) are now underscore-separated. This change does not apply to dash-separated endpoint names such as
(3.0) All endpoints with child rest queries (for example
/pdb/query/v4/environments/<env_name>/facts) now return a 404 if the parent data does not exist. In the past, an empty array was returned in this case, but this gave the impression that there was a parent but the particular child data was empty. To improve clarity, missing parent data now returns a 404 “Not Found” status code and a proper JSON error.
(3.0) The v4 catalogs endpoint has caused the response of the
resourcesfields to be expanded. For more information, see /pdb/query/v4/catalogs documentation.
(3.0) We have renamed the “name” key of the catalogs endpoint to “certname”, for consistency with other endpoints.
(3.0) The top-level v4 catalogs endpoint will now only return data about nodes that are active. This provides the same consistency as other non-historical data returns. Endpoints which specify the node explicitly will return results even for deactivated or expired nodes.
(2.0.0) The v4 catalogs endpoint does not contain a
metadatafield or an
api_versionfield. The contents of the v3
datafield compose the v4 response body, along with the new fields
environment. For more information, see the /pdb/query/v4/catalogs documentation.
(2.2.0) The v4 facts endpoint returns proper JSON rather than stringified JSON under the
valuefield in the case of a structured fact.
(2.2.0) Queries against fact values must use the appropriate type. Possible types are integer, float, Boolean, string, JSON, and null. Where the v3 API would return the same results for [”>”,”value”,”10”] and [”>”,”value”,10], only the second form will work on v4. The same applies for equality queries on Boolean values.
(2.2) The v4 factsets endpoint was added to facilitate the grouping of facts per node. For more information, see the /pdb/query/v4/factsets documentation.
(3.0) We added a
hashfield to the endpoint fields to support a unique identifer for factsets.
factsfield is now expanded as per our new expansion convention, so the data format has changed. For more information, see the /pdb/query/v4/factsets documentation.
/pdb/query/v4/factsets/<node>/factsendpoints will now return results even for deactivated or expired nodes.
aggregate-event-countsendpoint is no longer supported on HSQLDB.
aggregate-event-countsendpoint now accepts multiple
summarize_byparameters and returns an array of maps instead of a map. An additional
summarize_byfield has also been added to describe the parameter used. For more information, see the aggregate-event-counts documentation.
/metrics/v1 (formerly /v3/metrics)
(3.0) The former metrics endpoint has been split off into a separate service, and reversioned at v1. If you are currently accessing mbeans at http://localhost:8080/v3/metrics/mbeans, you will now access them at http://localhost:8080/metrics/v1/mbeans and so on, according to the metrics api documentation.
(3.0) PuppetDB’s mbeans (listed at /metrics/v1/mbeans) are no longer prefixed with “com.”
/pdb/cmd/v1 (formerly /v3/commands)
- (3.0) For users posting commands directly to the /pdb/cmd/v1 endpoint, the only valid command submission versions will be replace catalogs v6, store report v5, and replace facts v4.
/pdb/meta/v1/version (formerly /v3/version)
- (3.0) The version endpoint has been split from the query service and mounted
/pdb/meta/v1/server-time (formerly /v3/server-time)
- (3.0) The server-time endpoint has been split from the query service and mounted
New API features
/pdb/query/v4/factsetsThis endpoint returns a key-value hash for each certname. For more information, see the /pdb/query/v4/factsets documentation.
/pdb/query/v4/fact-pathsThis endpoint is similar to the existing fact-names endpoint in that one expected use is GUI autocompletion. For more information, see the /pdb/query/v4/fact-paths documentation.
/pdb/query/v4/fact-contentsThis endpoint allows fine-grained querying of structured facts. For more information, see the /pdb/query/v4/fact-contents documentation.
/pdb/query/v4/edgesThis endpoint allows querying edges inside a catalog. For more information, see the /pdb/query/v4/edges documentation
/pdb/query/v4/reports/<hash>/eventsThis convenience endpoint allows you to show events for a particular report by its hash. See the /pdb/query/v4/reports documentation
/pdb/query/v4/reports/<hash>/metricsThis endpoint allows you to show metrics for a particular report by its hash. See the /pdb/query/v4/reports documentation
/pdb/query/v4/reports/<hash>/logsThis endpoint allows you to show logs for a particular report by its hash. See the /pdb/query/v4/reports documentation
/pdb/query/v4/catalogs/<node>/[resources|edges]Both of these endpoints provide convenience for drilling into resources and edges data specific to a particular catalog. See /pdb/query/v4/catalogs documentation
Features affecting all endpoints
(3.0) Extract is available as a top-level query operator, useful for selecting only certain fields from a response. See the documentation on the extract operator for more information.
extractoperators have been changed to accept multiple fields, allowing more concise subquerying as explained here.
- (3.0) The v4 events endpoint does not require a query parameter, so
/pdb/query/v4/eventsis now a valid query. See the events endpoint documentation for more information.
(3.0) The response of the reports endpoint includes the new fields
metrics. For more information, see the documentation on the reports endpoint. For comparison, see an example of the new format, and an example of the old format.
(3.0) The reports endpoint takes a
latest_report?query to return only reports associated with the most recent puppet run for their nodes. Similar to the corresponding events query, there is no corresponding field in the response. For more information, see the documentation on the report query fields.
- (3.0) The v4 catalogs endpoint is queryable like the other
endpoints, whereas before it could only return a catalog for a
single host. The old query format (
/pdb/query/v4/catalogs/myhost) still works as before, but
/pdb/query/v4/catalogsreturns results too. For more information, see the catalog query examples.
(2.2.0) The new
select_fact_contentssubquery operator allows for filtering the results of other endpoints based on detailed queries about structured fact values. This is exhibited on the bottom of the subquery examples documentation.
(2.2.0) We have added the regexp array match operator
~>for querying fact paths on the
fact-paths endpoints. This is documented with the other operators. An example of usage is given at the bottom of the subquery examples page.
(3.0) We have added the
functionoperators, as well as support for the
countfunction. For more information, see the operators documentation.