Take control of your internal daemons!
Pebble helps you to orchestrate a set of local service processes as an organized set. It resembles well known tools such as supervisord, runit, or s6, in that it can easily manage non-system processes independently from the system services, but it was designed with unique features that help with more specific use cases.
- General model
- Layer configuration examples
- Using Pebble
- Container usage
- Layer specification
- API and clients
- Roadmap/TODO
- Hacking / Development
- Contributing
Pebble is organized as a single binary that works as a daemon and also as a
client to itself. When the daemon runs it loads its own configuration from the
$PEBBLE directory, as defined in the environment, and also records in
that same directory its state and unix sockets for communication. If that variable
is not defined, Pebble will attempt to look for its configuration from a default
system-level setup at /var/lib/pebble/default. Using that directory is encouraged
for whole-system setup such as when using Pebble to control services in a container.
The $PEBBLE directory must contain a layers/ subdirectory that holds a stack of
configuration files with names similar to 001-base-layer.yaml, where the digits define
the order of the layer and the following label uniquely identifies it. Each
layer in the stack sits above the former one, and has the chance to improve or
redefine the service configuration as desired.
Below is an example of the current configuration format. For full details of all fields, see the complete layer specification.
summary: Simple layer
description: |
A better description for a simple layer.
services:
srv1:
override: replace
summary: Service summary
command: cmd arg1 "arg2a arg2b"
startup: enabled
after:
- srv2
before:
- srv3
requires:
- srv2
- srv3
environment:
VAR1: val1
VAR2: val2
VAR3: val3
srv2:
override: replace
startup: enabled
command: cmd
before:
- srv3
srv3:
override: replace
command: cmdThe override field (which is required) defines whether this
entry overrides the previous service of the same name (if any),
or merges with it. See the full layer specification
for more details.
Any of the fields can be replaced individually in a merged service configuration. To illustrate, here is a sample override layer that might sit on top of the one above:
summary: Simple override layer
services:
srv1:
override: merge
environment:
VAR3: val3
after:
- srv4
before:
- srv5
srv2:
override: replace
summary: Replaced service
startup: disabled
command: cmd
srv4:
override: replace
command: cmd
startup: enabled
srv5:
override: replace
command: cmdTo install the latest version of Pebble, run the following command (we don't currently ship binaries, so you must first install Go):
go install github.com/canonical/pebble/cmd/pebble@latest
Pebble is invoked using pebble <command>. To get more information:
- To see a help summary, type
pebble -h. - To see a short description of all commands, type
pebble help --all. - To see details for one command, type
pebble help <command>orpebble <command> -h.
A few of the commands that need more explanation are detailed below.
If Pebble is installed and the $PEBBLE directory is set up, running the daemon is easy:
$ pebble run
2022-10-26T01:18:26.904Z [pebble] Started daemon.
2022-10-26T01:18:26.921Z [pebble] POST /v1/services 15.53132ms 202
2022-10-26T01:18:26.921Z [pebble] Started default services with change 50.
2022-10-26T01:18:26.936Z [pebble] Service "srv1" starting: sleep 300
This will start the Pebble daemon itself, as well as starting all the services that
are marked as startup: enabled (if you don't want that, use --hold). Then
other Pebble commands may be used to interact with the running daemon, for example,
in another terminal window.
To override the default configuration directory, set the PEBBLE environment variable when running:
$ export PEBBLE=~/pebble
pebble run
2022-10-26T01:18:26.904Z [pebble] Started daemon.
...
You can view the status of one or more services by using pebble services:
$ pebble services srv1 # show status of a single service
Service Startup Current
srv1 enabled active
$ pebble services # show status of all services
Service Startup Current
srv1 enabled active
srv2 disabled inactive
The "Startup" column shows whether this service is automatically started when Pebble starts ("enabled" means auto-start, "disabled" means don't auto-start).
The "Current" column shows the current status of the service, and can be one of the following:
active: starting or runninginactive: not yet started, being stopped, or stoppedbackoff: in a backoff-restart looperror: in an error state
To start specific services, type pebble start followed by one or more service names:
$ pebble start srv1 srv2 # start two services (and any dependencies)
When starting a service, Pebble executes the service's command, and waits 1 second to ensure the command doesn't exit too quickly. Assuming the command doesn't exit within that time window, the start is considered successful, otherwise pebble start will exit with an error.
Similarly, to stop specific services, use pebble stop followed by one or more service names:
$ pebble stop srv1 # stop one service
When stopping a service, Pebble sends SIGTERM to the service's process group, and waits up to 5 seconds. If the command hasn't exited within that time window, Pebble sends SIGKILL to the service's process group and waits up to 5 more seconds. If the command exits within that 10-second time window, the stop is considered successful, otherwise pebble stop will exit with an error.
When you update service configuration (by adding a layer), the services changed won't be automatically restarted. To restart them and bring the service state in sync with the new configuration, use pebble replan.
The "replan" operation restarts startup: enabled services whose configuration have changed between when they started and now; if the configuration hasn't changed, replan does nothing. Replan also starts startup: enabled services that have not yet been started.
Here is an example, where srv1 is a service that has startup: enabled, and srv2 does not:
$ pebble replan
2023-04-25T15:06:50+02:00 INFO Service "srv1" already started.
$ pebble add lay1 layer.yaml # update srv1 config
Layer "lay1" added successfully from "layer.yaml"
$ pebble replan
Stop service "srv1"
Start service "srv1"
$ pebble add lay2 layer.yaml # change srv2 to "startup: enabled"
Layer "lay2" added successfully from "layer.yaml"
$ pebble replan
2023-04-25T15:11:22+02:00 INFO Service "srv1" already started.
Start service "srv2"
If you want to force a service to restart even if its service configuration hasn't changed, use pebble restart <service>.
Pebble takes service dependencies into account when starting and stopping services. Before the service manager starts a service, Pebble first starts the services that service depends on (configured with required). Conversely, before stopping a service, Pebble first stops services that depend on that service.
For example, if service nginx requires logger, pebble start nginx will start logger and then start nginx. Running pebble stop logger will stop nginx and then logger; however, running pebble stop nginx will only stop nginx (nginx depends on logger, not the other way around).
If multiple dependencies need to be started at once, they're started in order according to the before and after configuration: before is a list of services that must be started before this one (but it doesn't require them). Or if it's easier to specify the other way around, after is a list of services that must be started after this one.
If the configuration of requires, before, and after for a service results in a cycle or "loop", an error will be returned when attempting to start or stop the service.
Pebble's service manager automatically restarts services that exit unexpectedly. By default, this is done whether the exit code is zero or non-zero, but you can change this using the on-success and on-failure fields in a configuration layer. The possible values for these fields are:
restart: restart the service and enter a restart-backoff loop (the default behaviour).shutdown: shut down and exit the Pebble daemonignore: ignore the service exiting and do nothing further
In restart mode, the first time a service exits, Pebble waits the backoff-delay, which defaults to half a second. If the service exits again, Pebble calculates the next backoff delay by multiplying the current delay by backoff-factor, which defaults to 2.0 (doubling). The increasing delay is capped at backoff-limit, which defaults to 30 seconds.
The backoff-limit value is also used as a "backoff reset" time. If the service stays running after a restart for backoff-limit seconds, the backoff process is reset and the delay reverts to backoff-delay.
Separate from the service manager, Pebble implements custom "health checks" that can be configured to restart services when they fail.
Each check can be one of three types. The types and their success criteria are:
http: an HTTPGETrequest to the URL specified must return an HTTP 2xx status codetcp: opening the given TCP port must be successfulexec: executing the specified command must yield a zero exit code
Checks are configured in the layer configuration using the top-level field checks. Full details are given in the layer specification, but below is an example layer showing the three different types of checks:
checks:
up:
override: replace
level: alive
period: 30s
threshold: 1 # an aggressive threshold
exec:
command: service nginx status
online:
override: replace
level: ready
tcp:
port: 8080
test:
override: replace
http:
url: http://localhost:8080/test
Each check is performed with the specified period (the default is 10 seconds apart), and is considered an error if a timeout happens before the check responds -- for example, before the HTTP request is complete or before the command finishes executing.
A check is considered healthy until it's had threshold errors in a row (the default is 3). At that point, the check is considered "down", and any associated on-check-failure actions will be triggered. When the check succeeds again, the failure count is reset to 0.
To enable Pebble auto-restart behavior based on a check, use the on-check-failure map in the service configuration (this is what ties together services and checks). For example, to restart the "server" service when the "test" check fails, use the following:
services:
server:
override: merge
on-check-failure:
test: restart # can also be "shutdown" or "ignore" (the default)
You can view check status using the pebble checks command. This reports the checks along with their status (up or down) and number of failures. For example:
$ pebble checks
Check Level Status Failures
up alive up 0/1
online ready down 1/3
test - down 42/3
The "Failures" column shows the current number of failures since the check started failing, a slash, and the configured threshold.
If the --http option was given when starting pebble run, Pebble exposes a /v1/health HTTP endpoint that allows a user to query the health of configured checks, optionally filtered by check level with the query string ?level=<level> This endpoint returns an HTTP 200 status if the checks are healthy, HTTP 502 otherwise.
Each check can specify a level of "alive" or "ready". These have semantic meaning: "alive" means the check or the service it's connected to is up and running; "ready" means it's properly accepting network traffic. These correspond to Kubernetes "liveness" and "readiness" probes.
The tool running the Pebble server can make use of this, for example, under Kubernetes you could initialize its liveness and readiness probes to hit Pebble's /v1/health endpoint with ?level=alive and ?level=ready filters, respectively.
Ready implies alive, and not-alive implies not-ready. If you've configured an "alive" check but no "ready" check, and the "alive" check is unhealthy, /v1/health?level=ready will report unhealthy as well, and the Kubernetes readiness probe will act on that.
If there are no checks configured, the /v1/health endpoint returns HTTP 200 so the liveness and readiness probes are successful by default. To use this feature, you must explicitly create checks with level: alive or level: ready in the layer configuration.
When Pebble performs a (potentially invasive or long-running) operation such as starting or stopping a service, it records a "change" object with one or more "tasks" in it. The daemon records this state in a JSON file on disk at $PEBBLE/.pebble.state.
To see recent changes, for this or previous server runs, use pebble changes. You might see something like this:
$ pebble changes
ID Status Spawn Ready Summary
1 Done today at 14:33 NZDT today at 14:33 NZDT Autostart service "srv1"
2 Done today at 15:26 NZDT today at 15:26 NZDT Start service "srv2"
3 Done today at 15:26 NZDT today at 15:26 NZDT Stop service "srv1" and 1 more
To drill down and see the tasks that make up a change, use pebble tasks <change-id>:
$ pebble tasks 3
Status Spawn Ready Summary
Done today at 15:26 NZDT today at 15:26 NZDT Stop service "srv1"
Done today at 15:26 NZDT today at 15:26 NZDT Stop service "srv2"
The daemon's service manager stores the most recent stdout and stderr from each service, using a 100KB ring buffer per service. Each log line is prefixed with an RFC-3339 timestamp and the [service-name] in square brackets.
Logs are viewable via the logs API or using pebble logs, for example:
$ pebble logs
2022-11-14T01:35:06.979Z [srv1] Log 0 from srv1
2022-11-14T01:35:08.041Z [srv2] Log 0 from srv2
2022-11-14T01:35:09.982Z [srv1] Log 1 from srv1
To view existing logs and follow (tail) new output, use -f (press Ctrl-C to exit):
$ pebble logs -f
2022-11-14T01:37:56.936Z [srv1] Log 0 from srv1
2022-11-14T01:37:57.978Z [srv2] Log 0 from srv2
2022-11-14T01:37:59.939Z [srv1] Log 1 from srv1
^C
You can output logs in JSON Lines format, using --format=json:
$ pebble logs --format=json
{"time":"2022-11-14T01:39:10.886Z","service":"srv1","message":"Log 0 from srv1"}
{"time":"2022-11-14T01:39:11.943Z","service":"srv2","message":"Log 0 from srv2"}
{"time":"2022-11-14T01:39:13.889Z","service":"srv1","message":"Log 1 from srv1"}
If you want to also write service logs to Pebble's own stdout, run the daemon with --verbose:
$ pebble run --verbose
2022-10-26T01:41:32.805Z [pebble] Started daemon.
2022-10-26T01:41:32.835Z [pebble] POST /v1/services 29.743632ms 202
2022-10-26T01:41:32.835Z [pebble] Started default services with change 7.
2022-10-26T01:41:32.849Z [pebble] Service "srv1" starting: python3 -u /path/to/srv1.py
2022-10-26T01:41:32.866Z [srv1] Log 0 from srv1
2022-10-26T01:41:35.870Z [srv1] Log 1 from srv1
2022-10-26T01:41:38.873Z [srv1] Log 2 from srv1
...
Pebble works well as a local service manager, but if running Pebble in a separate container, you can use the exec and file management APIs to coordinate with the remote system over the shared unix socket.
Pebble's "exec" feature allows you to run arbitrary commands on the server. This is intended for short-running programs; the processes started with exec don't use the service manager.
For example, you could use exec to run pg_dump and create a PostgreSQL database backup:
$ pebble exec pg_dump mydb
--
-- PostgreSQL database dump
--
...
The exec feature uses WebSockets under the hood, and allows you to stream stdin to the process, as well as stream stdout and stderr back. When running pebble exec, you can specify the working directory to run in (-w), environment variables to set (--env), and the user and group to run as (--uid/--user and --gid/--group).
You can also apply a timeout with --timeout, for example:
$ pebble exec --timeout 1s -- sleep 3
error: cannot perform the following tasks:
- exec command "sleep" (timed out after 1s: context deadline exceeded)
Pebble provides various API calls and commands to manage files and directories on the server. The simplest way to use these is with the commands below, several of which should be familiar:
$ pebble ls <path> # list file information (like "ls")
$ pebble mkdir <path> # create a directory (like "mkdir")
# TODO -- the following commands are coming soon
$ pebble rm <path> # remove a file or directory (like "rm")
$ pebble push <local> <remote> # copy file to server (like "cp")
$ pebble pull <remote> <local> # copy file from server (like "cp")
Below is the full specification for a Pebble configuration layer. Layers are added statically using a file in $PEBBLE/layers, or dynamically via the layers API or pebble add.
# (Optional) A short one line summary of the layer
summary: <summary>
# (Optional) A full description of the configuration layer
description: |
<description>
# (Optional) A list of services managed by this configuration layer
services:
<service name>:
# (Required) Control how this service definition is combined with any
# other pre-existing definition with the same name in the Pebble plan.
#
# The value 'merge' will ensure that values in this layer specification
# are merged over existing definitions, whereas 'replace' will entirely
# override the existing service spec in the plan with the same name.
override: merge | replace
# (Required in combined layer) The command to run the service. The
# command is executed directly, not interpreted by a shell.
#
# Example: /usr/bin/somecommand -b -t 30
command: <commmand>
# (Optional) A short summary of the service.
summary: <summary>
# (Optional) A detailed description of the service.
description: |
<description>
# (Optional) Control whether the service is started automatically when
# Pebble starts. Default is "disabled".
startup: enabled | disabled
# (Optional) A list of other services in the plan that this service
# should start after.
after:
- <other service name>
# (Optional) A list of other services in the plan that this service
# should start before.
before:
- <other service name>
# (Optional) A list of other services in the plan that this service
# requires in order to start correctly.
requires:
- <other service name>
# (Optional) A list of key/value pairs defining environment variables
# that should be set in the context of the process.
environment:
<env var name>: <env var value>
# (Optional) Username for starting service as a different user. It is
# an error if the user doesn't exist.
user: <username>
# (Optional) User ID for starting service as a different user. If both
# user and user-id are specified, the user's UID must match user-id.
user-id: <uid>
# (Optional) Group name for starting service as a different user. It is
# an error if the group doesn't exist.
group: <group name>
# (Optional) Group ID for starting service as a different user. If both
# group and group-id are specified, the group's GID must match group-id.
group-id: <gid>
# (Optional) Defines what happens when the service exits with a zero
# exit code. Possible values are: "restart" (default) which restarts
# the service after the backoff delay, "shutdown" which shuts down and
# exits the Pebble server, and "ignore" which does nothing further.
on-success: restart | shutdown | ignore
# (Optional) Defines what happens when the service exits with a nonzero
# exit code. Possible values are: "restart" (default) which restarts
# the service after the backoff delay, "shutdown" which shuts down and
# exits the Pebble server, and "ignore" which does nothing further.
on-failure: restart | shutdown | ignore
# (Optional) Defines what happens when each of the named health checks
# fail. Possible values are: "restart" (default) which restarts
# the service once, "shutdown" which shuts down and exits the Pebble
# server, and "ignore" which does nothing further.
on-check-failure:
<check name>: restart | shutdown | ignore
# (Optional) Initial backoff delay for the "restart" exit action.
# Default is half a second ("500ms").
backoff-delay: <duration>
# (Optional) After each backoff, the backoff delay is multiplied by
# this factor to get the next backoff delay. Must be greater than or
# equal to one. Default is 2.0.
backoff-factor: <factor>
# (Optional) Limit for the backoff delay: when multiplying by
# backoff-factor to get the next backoff delay, if the result is
# greater than this value, it is capped to this value. Default is
# half a minute ("30s").
backoff-limit: <duration>
# (Optional) The amount of time afforded to this service to handle
# SIGTERM and exit gracefully before SIGKILL terminates it forcefully.
# Default is 5 seconds ("5s").
kill-delay: <duration>
# (Optional) A list of health checks managed by this configuration layer.
checks:
<check name>:
# (Required) Control how this check definition is combined with any
# other pre-existing definition with the same name in the Pebble plan.
#
# The value 'merge' will ensure that values in this layer specification
# are merged over existing definitions, whereas 'replace' will entirely
# override the existing check spec in the plan with the same name.
override: merge | replace
# (Optional) Check level, which can be used for filtering checks when
# calling the checks API or health endpoint.
#
# For the health endpoint, ready implies alive. In other words, if all
# the "ready" checks are succeeding and there are no "alive" checks,
# the /v1/health API will return success for level=alive.
level: alive | ready
# (Optional) Check is run every time this period (time interval)
# elapses. Must not be zero. Default is "10s".
period: <duration>
# (Optional) If this time elapses before a single check operation has
# finished, it is cancelled and considered an error. Must be less
# than the period, and must not be zero. Default is "3s".
timeout: <duration>
# (Optional) Number of times in a row the check must error to be
# considered a failure (which triggers the on-check-failure action).
# Default 3.
threshold: <failure threshold>
# Configures an HTTP check, which is successful if a GET to the
# specified URL returns a 20x status code.
#
# Only one of "http", "tcp", or "exec" may be specified.
http:
# (Required) URL to fetch, for example "https://example.com/foo".
url: <full URL>
# (Optional) Map of HTTP headers to send with the request.
headers:
<name>: <value>
# Configures a TCP port check, which is successful if the specified
# TCP port is listening and we can successfully open it. Nothing is
# sent to the port.
#
# Only one of "http", "tcp", or "exec" may be specified.
tcp:
# (Required) Port number to open.
port: <port number>
# (Optional) Host name or IP address to use. Default is "localhost".
host: <host name>
# Configures a command execution check, which is successful if running
# the specified command returns a zero exit code.
#
# Only one of "http", "tcp", or "exec" may be specified.
exec:
# (Required) Command line to execute. The command is executed
# directly, not interpreted by a shell.
command: <commmand>
# (Optional) A list of key/value pairs defining environment
# variables that should be set when running the command.
environment:
<name>: <value>
# (Optional) Username for starting command as a different user. It
# is an error if the user doesn't exist.
user: <username>
# (Optional) User ID for starting command as a different user. If
# both user and user-id are specified, the user's UID must match
# user-id.
user-id: <uid>
# (Optional) Group name for starting command as a different user.
# It is an error if the group doesn't exist.
group: <group name>
# (Optional) Group ID for starting command as a different user. If
# both group and group-id are specified, the group's GID must
# match group-id.
group-id: <gid>
# (Optional) Working directory to run command in.
working-dir: <directory>The Pebble daemon exposes an API (HTTP over a unix socket) to allow remote clients to interact with the daemon. It can start and stop services, add configuration layers the plan, and so on.
There is currently no official documentation for the API at the HTTP level (apart from the code itself!); most users will interact with it via the Pebble command line interface or by using the Go or Python clients.
The Go client is used primarily by the CLI, but is importable and can be used by other tools too. See the reference documentation and examples at pkg.go.dev.
We try to never change the underlying HTTP API in a backwards-incompatible way, however, in rare cases we may change the Go client in a backwards-incompatible way.
In addition to the Go client, there's also a Python client for the Pebble API that's part of the Python Operator Framework used by Juju charms (documentation here).
This is a preview of what Pebble is becoming. Please keep that in mind while you explore.
Here are some of the things coming soon:
- Support
$PEBBLE_SOCKETand default$PEBBLEto/var/lib/pebble/default - Define and enforce convention for layer names
- Dynamic layer support over the API
- Configuration retrieval commands to investigate current settings
- Status command that displays active services and their current status
- General system modification commands (writing configuration files, etc)
- Better log caching and retrieval support
- Consider showing unified log as output of
pebble run(use-v) - Automatically restart services that fail
- Support for custom health checks (HTTP, TCP, command)
- Terminate all services before exiting run command
- Log forwarding (syslog and Loki)
- Other in-progress PRs
- Other requested features
See HACKING.md for information on how to run and hack on the Pebble codebase during development. In short, use go run ./cmd/pebble.
We welcome quality external contributions. We have good unit tests for much of the code, and a thorough code review process. Please note that unless it's a trivial fix, it's generally worth opening an issue to discuss before submitting a pull request.
Before you contribute a pull request you should sign the Canonical contributor agreement -- it's the easiest way for you to give us permission to use your contributions.
... and enjoy the rest of the year!