Guix Services (GNU Guix Reference Manual)

10.9.31 Guix Services

Guix Build Coordinator

The Guix Build Coordinator aids in distributing derivation builds among machines running an agent. The build daemon is still used to build the derivations, but the Guix Build Coordinator manages allocating builds and working with the results.

The Guix Build Coordinator consists of one coordinator, and one or more connected agent processes. The coordinator process handles clients submitting builds, and allocating builds to agents. The agent processes talk to a build daemon to actually perform the builds, then send the results back to the coordinator.

There is a script to run the coordinator component of the Guix Build Coordinator, but the Guix service uses a custom Guile script instead, to provide better integration with G-expressions used in the configuration.

Variable: Scheme Variable guix-build-coordinator-service-type: Service type for the Guix Build Coordinator. Its value must be a guix-build-coordinator-configuration object.

Data Type: guix-build-coordinator-configuration

Data type representing the configuration of the Guix Build Coordinator.

package (default: guix-build-coordinator): The Guix Build Coordinator package to use.
user (default: "guix-build-coordinator"): The system user to run the service as.
group (default: "guix-build-coordinator"): The system group to run the service as.
database-uri-string (default: "sqlite:///var/lib/guix-build-coordinator/guix_build_coordinator.db"): The URI to use for the database.
agent-communication-uri (default: "http://0.0.0.0:8745"): The URI describing how to listen to requests from agent processes.
client-communication-uri (default: "http://127.0.0.1:8746"): The URI describing how to listen to requests from clients. The client API allows submitting builds and currently isn’t authenticated, so take care when configuring this value.
allocation-strategy (default: #~basic-build-allocation-strategy): A G-expression for the allocation strategy to be used. This is a procedure that takes the datastore as an argument and populates the allocation plan in the database.
hooks (default: ’()): An association list of hooks. These provide a way to execute arbitrary code upon certain events, like a build result being processed.
guile (default: guile-3.0-latest): The Guile package with which to run the Guix Build Coordinator.

Variable: Scheme Variable guix-build-coordinator-agent-service-type: Service type for a Guix Build Coordinator agent. Its value must be a guix-build-coordinator-agent-configuration object.

Data Type: guix-build-coordinator-agent-configuration

Data type representing the configuration a Guix Build Coordinator agent.

package (default: guix-build-coordinator/agent-only)

The Guix Build Coordinator package to use.

user (default: "guix-build-coordinator-agent")

The system user to run the service as.

coordinator (default: "http://localhost:8745")

The URI to use when connecting to the coordinator.

authentication

Record describing how this agent should authenticate with the coordinator. Possible record types are described below.

systems (default: #f)

The systems for which this agent should fetch builds. The agent process will use the current system it’s running on as the default.

max-parallel-builds (default: 1)

The number of builds to perform in parallel.

max-1min-load-average (default: #f)

Load average value to look at when considering starting new builds, if the 1 minute load average exceeds this value, the agent will wait before starting new builds.

This will be unspecified if the value is #f, and the agent will use the number of cores reported by the system as the max 1 minute load average.

derivation-substitute-urls (default: #f)

URLs from which to attempt to fetch substitutes for derivations, if the derivations aren’t already available.

non-derivation-substitute-urls (default: #f)

URLs from which to attempt to fetch substitutes for build inputs, if the input store items aren’t already available.

Data Type: guix-build-coordinator-agent-password-auth

Data type representing an agent authenticating with a coordinator via a UUID and password.

uuid: The UUID of the agent. This should be generated by the coordinator process, stored in the coordinator database, and used by the intended agent.
password: The password to use when connecting to the coordinator.

Data Type: guix-build-coordinator-agent-password-file-auth

Data type representing an agent authenticating with a coordinator via a UUID and password read from a file.

uuid: The UUID of the agent. This should be generated by the coordinator process, stored in the coordinator database, and used by the intended agent.
password-file: A file containing the password to use when connecting to the coordinator.

Data Type: guix-build-coordinator-agent-dynamic-auth

Data type representing an agent authenticating with a coordinator via a dynamic auth token and agent name.

agent-name: Name of an agent, this is used to match up to an existing entry in the database if there is one. When no existing entry is found, a new entry is automatically added.
token: Dynamic auth token, this is created and stored in the coordinator database, and is used by the agent to authenticate.

Data Type: guix-build-coordinator-agent-dynamic-auth-with-file

Data type representing an agent authenticating with a coordinator via a dynamic auth token read from a file and agent name.

agent-name: Name of an agent, this is used to match up to an existing entry in the database if there is one. When no existing entry is found, a new entry is automatically added.
token-file: File containing the dynamic auth token, this is created and stored in the coordinator database, and is used by the agent to authenticate.

The Guix Build Coordinator package contains a script to query an instance of the Guix Data Service for derivations to build, and then submit builds for those derivations to the coordinator. The service type below assists in running this script. This is an additional tool that may be useful when building derivations contained within an instance of the Guix Data Service.

Variable: Scheme Variable guix-build-coordinator-queue-builds-service-type: Service type for the guix-build-coordinator-queue-builds-from-guix-data-service script. Its value must be a guix-build-coordinator-queue-builds-configuration object.

Data Type: guix-build-coordinator-queue-builds-configuration

Data type representing the options to the queue builds from guix data service script.

package (default: guix-build-coordinator): The Guix Build Coordinator package to use.
user (default: "guix-build-coordinator-queue-builds"): The system user to run the service as.
coordinator (default: "http://localhost:8746"): The URI to use when connecting to the coordinator.
systems (default: #f): The systems for which to fetch derivations to build.
systems-and-targets (default: #f): An association list of system and target pairs for which to fetch derivations to build.
guix-data-service (default: "https://data.guix.gnu.org"): The Guix Data Service instance from which to query to find out about derivations to build.
processed-commits-file (default: "/var/cache/guix-build-coordinator-queue-builds/processed-commits"): A file to record which commits have been processed, to avoid needlessly processing them again if the service is restarted.

Guix Data Service

The Guix Data Service processes, stores and provides data about GNU Guix. This includes information about packages, derivations and lint warnings.

The data is stored in a PostgreSQL database, and available through a web interface.

Variable: Scheme Variable guix-data-service-type: Service type for the Guix Data Service. Its value must be a guix-data-service-configuration object. The service optionally extends the getmail service, as the guix-commits mailing list is used to find out about changes in the Guix git repository.

Data Type: guix-data-service-configuration

Data type representing the configuration of the Guix Data Service.

package (default: guix-data-service): The Guix Data Service package to use.
user (default: "guix-data-service"): The system user to run the service as.
group (default: "guix-data-service"): The system group to run the service as.
port (default: 8765): The port to bind the web service to.
host (default: "127.0.0.1"): The host to bind the web service to.
getmail-idle-mailboxes (default: #f): If set, this is the list of mailboxes that the getmail service will be configured to listen to.
commits-getmail-retriever-configuration (default: #f): If set, this is the getmail-retriever-configuration object with which to configure getmail to fetch mail from the guix-commits mailing list.
extra-options (default: ’()): Extra command line options for guix-data-service.
extra-process-jobs-options (default: ’()): Extra command line options for guix-data-service-process-jobs.

Nar Herder

The Nar Herder is a utility for managing a collection of nars.

Variable: Scheme Variable nar-herder-type: Service type for the Guix Data Service. Its value must be a nar-herder-configuration object. The service optionally extends the getmail service, as the guix-commits mailing list is used to find out about changes in the Guix git repository.

Data Type: nar-herder-configuration

Data type representing the configuration of the Guix Data Service.

package (default: nar-herder)

The Nar Herder package to use.

user (default: "nar-herder")

The system user to run the service as.

group (default: "nar-herder")

The system group to run the service as.

port (default: 8734)

The port to bind the server to.

host (default: "127.0.0.1")

The host to bind the server to.

mirror (default: #f)

Optional URL of the other Nar Herder instance which should be mirrored. This means that this Nar Herder instance will download it’s database, and keep it up to date.

database (default: "/var/lib/nar-herder/nar_herder.db")

Location for the database. If this Nar Herder instance is mirroring another, the database will be downloaded if it doesn’t exist. If this Nar Herder instance isn’t mirroring another, an empty database will be created.

database-dump (default: "/var/lib/nar-herder/nar_herder_dump.db")

Location of the database dump. This is created and regularly updated by taking a copy of the database. This is the version of the database that is available to download.

storage (default: #f)

Optional location in which to store nars.

storage-limit (default: "none")

Limit in bytes for the nars stored in the storage location. This can also be set to “none” so that there is no limit.

When the storage location exceeds this size, nars are removed according to the nar removal criteria.

storage-nar-removal-criteria (default: '())

Criteria used to remove nars from the storage location. These are used in conjunction with the storage limit.

When the storage location exceeds the storage limit size, nars will be checked against the nar removal criteria and if any of the criteria match, they will be removed. This will continue until the storage location is below the storage limit size.

Each criteria is specified by a string, then an equals sign, then another string. Currently, only one criteria is supported, checking if a nar is stored on another Nar Herder instance.

ttl (default: #f)

Produce Cache-Control HTTP headers that advertise a time-to-live (TTL) of ttl. ttl must denote a duration: 5d means 5 days, 1m means 1 month, and so on.

This allows the user’s Guix to keep substitute information in cache for ttl.

negative-ttl (default: #f)

Similarly produce Cache-Control HTTP headers to advertise the time-to-live (TTL) of negative lookups—missing store items, for which the HTTP 404 code is returned. By default, no negative TTL is advertised.