Find file Copy path
Fetching contributors…
Cannot retrieve contributors at this time
1201 lines (874 sloc) 42.6 KB

Dashboard Developer Documentation

Frontend Development

Before you can start the dashboard from within a development environment, you will need to generate the frontend code and either use a compiled and running Ceph cluster (e.g. started by or the standalone development web server.

The build process is based on Node.js and requires the Node Package Manager npm to be installed.


  • Node 8.9.0 or higher
  • NPM 5.7.0 or higher

During Ceph's build we create a virtualenv with node and npm installed, which can be used as an alternative to installing node/npm in your system.

If you want to use the node installed in the virtualenv you just need to activate the virtualenv before you run any npm commands. To activate it run . build/src/pybind/mgr/dashboard/node-env/bin/activate.

Once you finish, you can simply run deactivate and exit the virtualenv.

Angular CLI:
If you do not have the Angular CLI installed globally, then you need to execute ng commands with an additional npm run before it.

Package installation

Run npm install in directory src/pybind/mgr/dashboard/frontend to install the required packages locally.

Setting up a Development Server

Create the proxy.conf.json file based on proxy.conf.json.sample.

Run npm start for a dev server. Navigate to http://localhost:4200/. The app will automatically reload if you change any of the source files.

Code Scaffolding

Run ng generate component component-name to generate a new component. You can also use ng generate directive|pipe|service|class|guard|interface|enum|module.

Build the Project

Run npm run build to build the project. The build artifacts will be stored in the dist/ directory. Use the -prod flag for a production build. Navigate to https://localhost:8443.

Code linting and formatting

We use the following tools to lint and format the code in all our TS, SCSS and HTML files:

We added 2 npm scripts to help run these tools:

  • npm run lint, will check frontend files against all linters
  • npm run fix, will try to fix all the detected linting errors

Running Unit Tests

Create unit-test-configuration.ts file based on unit-test-configuration.ts.sample in directory src/pybind/mgr/dashboard/frontend/src.

Run npm run test to execute the unit tests via Jest.

If you get errors on all tests, it could be because Jest or something else was updated. There are a few ways how you can try to resolve this:

  • Remove all modules with rm -rf dist node_modules and run npm install again in order to reinstall them
  • Clear the cache of jest by running npx jest --clearCache

Running End-to-End Tests

We use Protractor to run our frontend e2e tests.

Our script will check if Chrome or Docker is installed and run the tests if either is found.

Start all frontend e2e tests by running:

$ ./

You can force the script to use a specific device with the -d flag:

$ ./ -d <chrome|docker>

If you want to run the tests outside the ceph environment, you will need to manually define the dashboard url using -r:

When using docker, as your device, you might need to run the script with sudo permissions.

Further Help

To get more help on the Angular CLI use ng help or go check out the Angular CLI README.

Example of a Generator

# Create module 'Core'
src/app> ng generate module core -m=app --routing

# Create module 'Auth' under module 'Core'
src/app/core> ng generate module auth -m=core --routing
or, alternatively:
src/app> ng generate module core/auth -m=core --routing

# Create component 'Login' under module 'Auth'
src/app/core/auth> ng generate component login -m=core/auth
or, alternatively:
src/app> ng generate component core/auth/login -m=core/auth

Frontend Typescript Code Style Guide Recommendations

Group the imports based on its source and separate them with a blank line.

The source groups can be either from Angular, external or internal.


import { Component } from '@angular/core';
import { Router } from '@angular/router';

import { ToastsManager } from 'ng2-toastr';

import { Credentials } from '../../../shared/models/credentials.model';
import { HostService } from './services/host.service';

Frontend components

There are several components that can be reused on different pages. This components are declared on the components module: src/pybind/mgr/dashboard/frontend/src/app/shared/components.


This component should be used to provide additional information to the user.


  Some <strong>helper</strong> html text

Backend Development

The Python backend code of this module requires a number of Python modules to be installed. They are listed in file requirements.txt. Using pip you may install all required dependencies by issuing pip install -r requirements.txt in directory src/pybind/mgr/dashboard.

If you're using the ceph-dev-docker development environment, simply run ./ from the toplevel directory to install them.

Unit Testing

In dashboard we have two different kinds of backend tests:

  1. Unit tests based on tox
  2. API tests based on Teuthology.

Unit tests based on tox

We included a tox configuration file that will run the unit tests under Python 2 or 3, as well as linting tools to guarantee the uniformity of code.

You need to install tox and coverage before running it. To install the packages in your system, either install it via your operating system's package management tools, e.g. by running dnf install python-tox python-coverage on Fedora Linux.

Alternatively, you can use Python's native package installation method:

$ pip install tox
$ pip install coverage

To run the tests, run tox in the dashboard directory (where tox.ini is located).

We also collect coverage information from the backend code. You can check the coverage information provided by the tox output, or by running the following command after tox has finished successfully:

$ coverage html

This command will create a directory htmlcov with an HTML representation of the code coverage of the backend.

You can also run a single step of the tox script (aka tox environment), for instance if you only want to run the linting tools, do:

$ tox -e lint

API tests based on Teuthology

To run our API tests against a real Ceph cluster, we leverage the Teuthology framework. This has the advantage of catching bugs originated from changes in the internal Ceph code.

Our script will start a vstart Ceph cluster before running the Teuthology tests, and then it stops the cluster after the tests are run. Of course this implies that you have built/compiled Ceph previously.

Start all dashboard tests by running:

$ ./

Or, start one or multiple specific tests by specifying the test name:

$ ./ tasks.mgr.dashboard.test_pool.PoolTest

Or, source the script and run the tests manually:

$ source
$ run_teuthology_tests [tests]...
$ cleanup_teuthology

How to add a new controller?

A controller is a Python class that extends from the BaseController class and is decorated with either the @Controller, @ApiController or @UiApiController decorators. The Python class must be stored inside a Python file located under the controllers directory. The Dashboard module will automatically load your new controller upon start.

@ApiController and @UiApiController are both specializations of the @Controller decorator.

The @ApiController should be used for controllers that provide an API-like REST interface and the @UiApiController should be used for endpoints consumed by the UI but that are not part of the 'public' API. For any other kinds of controllers the @Controller decorator should be used.

A controller has a URL prefix path associated that is specified in the controller decorator, and all endpoints exposed by the controller will share the same URL prefix path.

A controller's endpoint is exposed by implementing a method on the controller class decorated with the @Endpoint decorator.

For example create a file under controllers directory with the following code:

from import Controller, ApiController, UiApiController, BaseController, Endpoint

class Ping(BaseController):
  def hello(self):
    return {'msg': "Hello"}

class ApiPing(BaseController):
  def hello(self):
    return {'msg': "Hello"}

class UiApiPing(BaseController):
  def hello(self):
    return {'msg': "Hello"}

The hello endpoint of the Ping controller can be reached by the following URL: https://mgr_hostname:8443/ping/hello using HTTP GET requests. As you can see the controller URL path /ping is concatenated to the method name hello to generate the endpoint's URL.

In the case of the ApiPing controller, the hello endpoint can be reached by the following URL: https://mgr_hostname:8443/api/ping/hello using a HTTP GET request. The API controller URL path /ping is prefixed by the /api path and then concatenated to the method name hello to generate the endpoint's URL. Internally, the @ApiController is actually calling the @Controller decorator by passing an additional decorator parameter called base_url:

@ApiController('/ping') <=> @Controller('/ping', base_url="/api")

UiApiPing works in a similar way than the ApiPing, but the URL will be prefixed by /ui-api: https://mgr_hostname:8443/ui-api/ping/hello. UiApiPing is also a @Controller extension:

@UiApiController('/ping') <=> @Controller('/ping', base_url="/ui-api")

The @Endpoint decorator also supports many parameters to customize the endpoint:

  • method="GET": the HTTP method allowed to access this endpoint.
  • path="/<method_name>": the URL path of the endpoint, excluding the controller URL path prefix.
  • path_params=[]: list of method parameter names that correspond to URL path parameters. Can only be used when method in ['POST', 'PUT'].
  • query_params=[]: list of method parameter names that correspond to URL query parameters.
  • json_response=True: indicates if the endpoint response should be serialized in JSON format.
  • proxy=False: indicates if the endpoint should be used as a proxy.

An endpoint method may have parameters declared. Depending on the HTTP method defined for the endpoint the method parameters might be considered either path parameters, query parameters, or body parameters.

For GET and DELETE methods, the method's non-optional parameters are considered path parameters by default. Optional parameters are considered query parameters. By specifying the query_parameters in the endpoint decorator it is possible to make a non-optional parameter to be a query parameter.

For POST and PUT methods, all method parameters are considered body parameters by default. To override this default, one can use the path_params and query_params to specify which method parameters are path and query parameters respectivelly. Body parameters are decoded from the request body, either from a form format, or from a dictionary in JSON format.

Let's use an example to better understand the possible ways to customize an endpoint:

from import Controller, BaseController, Endpoint

class Ping(BaseController):

  # URL: /ping/{key}?opt1=...&opt2=...
  @Endpoint(path="/", query_params=['opt1'])
  def index(self, key, opt1, opt2=None):
    # ...

  # URL: /ping/{key}?opt1=...&opt2=...
  def __call__(self, key, opt1, opt2=None):
    # ...

  # URL: /ping/post/{key1}/{key2}
  @Endpoint('POST', path_params=['key1', 'key2'])
  def post(self, key1, key2, data1, data2=None):
    # ...

In the above example we see how the path option can be used to override the generated endpoint URL in order to not use the method's name in the URL. In the index method we set the path to "/" to generate an endpoint that is accessible by the root URL of the controller.

An alternative approach to generate an endpoint that is accessible through just the controller's path URL is by using the __call__ method, as we show in the above example.

From the third method you can see that the path parameters are collected from the URL by parsing the list of values separated by slashes / that come after the URL path /ping for index method case, and /ping/post for the post method case.

Defining path parameters in endpoints's URLs using python methods's parameters is very easy but it is still a bit strict with respect to the position of these parameters in the URL structure. Sometimes we may want to explicitly define a URL scheme that contains path parameters mixed with static parts of the URL. Our controller infrastructure also supports the declaration of URL paths with explicit path parameters at both the controller level and method level.

Consider the following example:

from import Controller, BaseController, Endpoint

class Ping(BaseController):

  # URL: /ping/{node}/stats/{date}/latency?unit=...
  def latency(self, node, date, unit="ms"):
    # ...

In this example we explicitly declare a path parameter {node} in the controller URL path, and a path parameter {date} in the latency method. The endpoint for the latency method is then accessible through the URL: https://mgr_hostname:8443/ping/{node}/stats/{date}/latency .

For a full set of examples on how to use the @Endpoint decorator please check the unit test file: tests/ There you will find many examples of how to customize endpoint methods.

Implementing Proxy Controller

Sometimes you might need to relay some requests from the Dashboard frontend directly to an external service. For that purpose we provide a decorator called @Proxy. (As a concrete example, check the controllers/ file where we implemented an RGW Admin Ops proxy.)

The @Proxy decorator is a wrapper of the @Endpoint decorator that already customizes the endpoint for working as a proxy. A proxy endpoint works by capturing the URL path that follows the controller URL prefix path, and does not do any decoding of the request body.


from import Controller, BaseController, Proxy

class FooServiceProxy(BaseController):

  def proxy(self, path, **params):
    # if requested URL is "/foo/proxy/access/service?opt=1"
    # then path is "access/service" and params is {'opt': '1'}
    # ...

How does the RESTController work?

We also provide a simple mechanism to create REST based controllers using the RESTController class. Any class which inherits from RESTController will, by default, return JSON.

The RESTController is basically an additional abstraction layer which eases and unifies the work with collections. A collection is just an array of objects with a specific type. RESTController enables some default mappings of request types and given parameters to specific method names. This may sound complicated at first, but it's fairly easy. Lets have look at the following example:

import cherrypy
from import ApiController, RESTController

class Ping(RESTController):
  def list(self):
    return {"msg": "Hello"}

  def get(self, id):
    return self.objects[id]

In this case, the list method is automatically used for all requests to api/ping where no additional argument is given and where the request type is GET. If the request is given an additional argument, the ID in our case, it won't map to list anymore but to get and return the element with the given ID (assuming that self.objects has been filled before). The same applies to other request types:

Request type Arguments Method Status Code
GET No list 200
PUT No bulk_set 200
POST No create 201
DELETE No bulk_delete 204
GET Yes get 200
PUT Yes set 200
DELETE Yes delete 204

How to restrict access to a controller?

All controllers require authentication by default. If you require that the controller can be accessed without authentication, then you can add the parameter secure=False to the controller decorator.


import cherrypy
from . import ApiController, RESTController

@ApiController('ping', secure=False)
class Ping(RESTController):
  def list(self):
    return {"msg": "Hello"}

How to access the manager module instance from a controller?

We provide the manager module instance as a global variable that can be imported in any module. We also provide a logger instance in the same way.


import cherrypy
from .. import logger, mgr
from import ApiController, RESTController

class Servers(RESTController):
  def list(self):
    logger.debug('Listing available servers')
    return {'servers': mgr.list_servers()}

How to write a unit test for a controller?

We provide a test helper class called ControllerTestCase to easily create unit tests for your controller.

If we want to write a unit test for the above Ping controller, create a file under the tests directory with the following code:

from .helper import ControllerTestCase
from import Ping

class PingTest(ControllerTestCase):
    def setup_test(cls):
        Ping._cp_config['tools.authentication.on'] = False

    def test_ping(self):
        self.assertJsonBody({'msg': 'Hello'})

The ControllerTestCase class starts by initializing a CherryPy webserver. Then it will call the setup_test() class method where we can explicitly load the controllers that we want to test. In the above example we are only loading the Ping controller. We can also disable authentication of a controller at this stage, as depicted in the example.

How to listen for manager notifications in a controller?

The manager notifies the modules of several types of cluster events, such as cluster logging event, etc...

Each module has a "global" handler function called notify that the manager calls to notify the module. But this handler function must not block or spend too much time processing the event notification. For this reason we provide a notification queue that controllers can register themselves with to receive cluster notifications.

The example below represents a controller that implements a very simple live log viewer page:

from __future__ import absolute_import

import collections

import cherrypy

from import ApiController, BaseController, NotificationQueue

class LiveLog(BaseController):
    log_buffer = collections.deque(maxlen=1000)

    def __init__(self):
        super(LiveLog, self).__init__()
        NotificationQueue.register(self.log, 'clog')

    def log(self, log_struct):

    def default(self):
        ret = '<html><meta http-equiv="refresh" content="2" /><body>'
        for l in self.log_buffer:
            ret += "{}<br>".format(l)
        ret += "</body></html>"
        return ret

As you can see above, the NotificationQueue class provides a register method that receives the function as its first argument, and receives the "notification type" as the second argument. You can omit the second argument of the register method, and in that case you are registering to listen all notifications of any type.

Here is an list of notification types (these might change in the future) that can be used:

  • clog: cluster log notifications
  • command: notification when a command issued by MgrModule.send_command completes
  • perf_schema_update: perf counters schema update
  • mon_map: monitor map update
  • fs_map: cephfs map update
  • osd_map: OSD map update
  • service_map: services (RGW, RBD-Mirror, etc.) map update
  • mon_status: monitor status regular update
  • health: health status regular update
  • pg_summary: regular update of PG status information

How to write a unit test when a controller accesses a Ceph module?

Consider the following example that implements a controller that retrieves the list of RBD images of the rbd pool:

import rbd
from .. import mgr
from import ApiController, RESTController

class RbdImages(RESTController):
    def __init__(self):
        self.ioctx = mgr.rados.open_ioctx('rbd')
        self.rbd = rbd.RBD()

    def list(self):
        return [{'name': n} for n in self.rbd.list(self.ioctx)]

In the example above, we want to mock the return value of the rbd.list function, so that we can test the JSON response of the controller.

The unit test code will look like the following:

import mock
from .helper import ControllerTestCase

class RbdImagesTest(ControllerTestCase):
    def test_list(self, rbd_list_mock):
        rbd_list_mock.return_value = ['img1', 'img2']
        self.assertJsonBody([{'name': 'img1'}, {'name': 'img2'}])

How to add a new configuration setting?

If you need to store some configuration setting for a new feature, we already provide an easy mechanism for you to specify/use the new config setting.

For instance, if you want to add a new configuration setting to hold the email address of the dashboard admin, just add a setting name as a class attribute to the Options class in the file:

# ...
class Options(object):
  # ...


The value of the class attribute is a pair composed by the default value for that setting, and the python type of the value.

By declaring the ADMIN_EMAIL_ADDRESS class attribute, when you restart the dashboard plugin, you will automatically gain two additional CLI commands to get and set that setting:

$ ceph dashboard get-admin-email-address
$ ceph dashboard set-admin-email-address <value>

To access, or modify the config setting value from your Python code, either inside a controller or anywhere else, you just need to import the Settings class and access it like this:

from settings import Settings

# ...
tmp_var = Settings.ADMIN_EMAIL_ADDRESS

# ....

The settings management implementation will make sure that if you change a setting value from the Python code you will see that change when accessing that setting from the CLI and vice-versa.

How to run a controller read-write operation asynchronously?

Some controllers might need to execute operations that alter the state of the Ceph cluster. These operations might take some time to execute and to maintain a good user experience in the Web UI, we need to run those operations asynchronously and return immediately to frontend some information that the operations are running in the background.

To help in the development of the above scenario we added the support for asynchronous tasks. To trigger the execution of an asynchronous task we must use the following class method of the TaskManager class:

from import TaskManager
# ..., metadata, func, args, kwargs)
  • name is a string that can be used to group tasks. For instance for RBD image creation tasks we could specify "rbd/create" as the name, or similarly "rbd/remove" for RBD image removal tasks.
  • metadata is a dictionary where we can store key-value pairs that characterize the task. For instance, when creating a task for creating RBD images we can specify the metadata argument as {'pool_name': "rbd", image_name': "test-img"}.
  • func is the python function that implements the operation code, which will be executed asynchronously.
  • args and kwargs are the positional and named arguments that will be passed to func when the task manager starts its execution.

The method triggers the asynchronous execution of function func and returns a Task object. The Task provides the public method Task.wait(timeout), which can be used to wait for the task to complete up to a timeout defined in seconds and provided as an argument. If no argument is provided the wait method blocks until the task is finished.

The Task.wait is very useful for tasks that usually are fast to execute but that sometimes may take a long time to run. The return value of the Task.wait method is a pair (state, value) where state is a string with following possible values:

  • VALUE_DONE = "done"
  • VALUE_EXECUTING = "executing"

The value will store the result of the execution of function func if state == VALUE_DONE. If state == VALUE_EXECUTING then value == None.

The pair (name, metadata) should unequivocally identify the task being run, which means that if you try to trigger a new task that matches the same (name, metadata) pair of the currently running task, then the new task is not created and you get the task object of the current running task.

For instance, consider the following example:

task1 ="dummy/task", {'attr': 2}, func)
task2 ="dummy/task", {'attr': 2}, func)

If the second call to executes while the first task is still executing then it will return the same task object: assert task1 == task2.

How to get the list of executing and finished asynchronous tasks?

The list of executing and finished tasks is included in the Summary controller, which is already polled every 5 seconds by the dashboard frontend. But we also provide a dedicated controller to get the same list of executing and finished tasks.

The Task controller exposes the /api/task endpoint that returns the list of executing and finished tasks. This endpoint accepts the name parameter that accepts a glob expression as its value. For instance, an HTTP GET request of the URL /api/task?name=rbd/* will return all executing and finished tasks which name starts with rbd/.

To prevent the finished tasks list from growing unbounded, we will always maintain the 10 most recent finished tasks, and the remaining older finished tasks will be removed when reaching a TTL of 1 minute. The TTL is calculated using the timestamp when the task finished its execution. After a minute, when the finished task information is retrieved, either by the summary controller or by the task controller, it is automatically deleted from the list and it will not be included in further task queries.

Each executing task is represented by the following dictionary:

  'name': "name",  # str
  'metadata': { },  # dict
  'begin_time': "2018-03-14T15:31:38.423605Z",  # str (ISO 8601 format)
  'progress': 0  # int (percentage)

Each finished task is represented by the following dictionary:

  'name': "name",  # str
  'metadata': { },  # dict
  'begin_time': "2018-03-14T15:31:38.423605Z",  # str (ISO 8601 format)
  'end_time': "2018-03-14T15:31:39.423605Z",  # str (ISO 8601 format)
  'duration': 0.0,  # float
  'progress': 0  # int (percentage)
  'success': True,  # bool
  'ret_value': None,  # object, populated only if 'success' == True
  'exception': None,  # str, populated only if 'success' == False

How to use asynchronous APIs with asynchronous tasks?

The method as described in a previous section, is well suited for calling blocking functions, as it runs the function inside a newly created thread. But sometimes we want to call some function of an API that is already asynchronous by nature.

For these cases we want to avoid creating a new thread for just running a non-blocking function, and want to leverage the asynchronous nature of the function. The is already prepared to be used with non-blocking functions by passing an object of the type TaskExecutor as an additional parameter called executor. The full method signature of, metadata, func, args=None, kwargs=None, executor=None)

The TaskExecutor class is responsible for code that executes a given task function, and defines three methods that can be overridden by subclasses:

def init(self, task)
def start(self)
def finish(self, ret_value, exception)

The init method is called before the running the task function, and receives the task object (of class Task).

The start method runs the task function. The default implementation is to run the task function in the current thread context.

The finish method should be called when the task function finishes with either the ret_value populated with the result of the execution, or with an exception object in the case that execution raised an exception.

To leverage the asynchronous nature of a non-blocking function, the developer should implement a custom executor by creating a subclass of the TaskExecutor class, and provide an instance of the custom executor class as the executor parameter of the

To better understand the expressive power of executors, we write a full example of use a custom executor to execute the MgrModule.send_command asynchronous function:

import json
from mgr_module import CommandResult
from .. import mgr
from import ApiController, RESTController, NotificationQueue, \
                    TaskManager, TaskExecutor

class SendCommandExecutor(TaskExecutor):
    def __init__(self):
        super(SendCommandExecutor, self).__init__()
        self.tag = None
        self.result = None

    def init(self, task):
        super(SendCommandExecutor, self).init(task)

        # we need to listen for 'command' events to know when the command
        # finishes
        NotificationQueue.register(self._handler, 'command')

        # store the CommandResult object to retrieve the results
        self.result = self.task.fn_args[0]
        if len(self.task.fn_args) > 4:
            # the user specified a tag for the command, so let's use it
            self.tag = self.task.fn_args[4]
            # let's generate a unique tag for the command
            self.tag = 'send_command_{}'.format(id(self))

    def _handler(self, data):
        if data == self.tag:
            # the command has finished, notifying the task with the result
            self.finish(self.result.wait(), None)
            # deregister listener to avoid memory leaks
            NotificationQueue.deregister(self._handler, 'command')

class Test(RESTController):

    def _run_task(self, osd_id):
        task ="test/task", {}, mgr.send_command,
                               [CommandResult(''), 'osd', osd_id,
                                json.dumps({'prefix': 'perf histogram dump'})],
        return task.wait(1.0)

    def get(self, osd_id):
        status, value = self._run_task(osd_id)
        return {'status': status, 'value': value}

The above SendCommandExecutor executor class can be used for any call to MgrModule.send_command. This means that we should need just one custom executor class implementation for each non-blocking API that we use in our controllers.

The default executor, used when no executor object is passed to, is the ThreadedExecutor. You can check its implementation in the file.

How to update the execution progress of an asynchronous task?

The asynchronous tasks infrastructure provides support for updating the execution progress of an executing task. The progress can be updated from within the code the task is executing, which usually is the place where we have the progress information available.

To update the progress from within the task code, the TaskManager class provides a method to retrieve the current task object:


The above method is only available when using the default executor ThreadedExecutor for executing the task. The current_task() method returns the current Task object. The Task object provides two public methods to update the execution progress value: the set_progress(percentage), and the inc_progress(delta) methods.

The set_progress method receives as argument an integer value representing the absolute percentage that we want to set to the task.

The inc_progress method receives as argument an integer value representing the delta we want to increment to the current execution progress percentage.

Take the following example of a controller that triggers a new task and updates its progress:

from __future__ import absolute_import
import random
import time
import cherrypy
from import TaskManager, ApiController, BaseController

class DummyTask(BaseController):
    def _dummy(self):
        top = random.randrange(100)
        for i in range(top):
            # or TaskManager.current_task().inc_progress(100/top)
        return "finished"

    def default(self):
        task ="dummy/task", {}, self._dummy)
        return task.wait(5)  # wait for five seconds

How to deal with asynchronous tasks in the front-end?

All executing and most recently finished asynchronous tasks are displayed on "Background-Tasks" and if finished on "Recent-Notifications" in the menu bar. For each task a operation name for three states (running, success and failure), a function that tells who is involved and error descriptions, if any, have to be provided. This can be achieved by appending TaskManagerMessageService.messages. This has to be done to achieve consistency among all tasks and states.

Operation Object
Ensures consistency among all tasks. It consists of three verbs for each different state f.e. {running: 'Creating', failure: 'create', success: 'Created'}.
  1. Put running operations in present participle f.e. 'Updating'.
  2. Failed messages always start with 'Failed to ' and should be continued with the operation in present tense f.e. 'update'.
  3. Put successful operations in past tense f.e. 'Updated'.
Involves Function
Ensures consistency among all messages of a task, it resembles who's involved by the operation. It's a function that returns a string which takes the metadata from the task to return f.e. "RBD 'somePool/someImage'".

Both combined create the following messages:

  • Failure => "Failed to create RBD 'somePool/someImage'"
  • Running => "Creating RBD 'somePool/someImage'"
  • Success => "Created RBD 'somePool/someImage'"

For automatic task handling use TaskWrapperService.wrapTaskAroundCall.

If for some reason wrapTaskAroundCall is not working for you, you have to subscribe to your asynchronous task manually through TaskManagerService.subscribe, and provide it with a callback, in case of a success to notify the user. A notification can be triggered with NotificationService.notifyTask. It will use TaskManagerMessageService.messages to display a message based on the state of a task.

Notifications of API errors are handled by ApiInterceptorService.

Usage example:

export class TaskManagerMessageService {
  // ...
  messages = {
    // Messages for task 'rbd/create'
    'rbd/create': new TaskManagerMessage(
      // Message prefixes
      ['create', 'Creating', 'Created'],
      // Message suffix
      (metadata) => `RBD '${metadata.pool_name}/${metadata.image_name}'`,
      (metadata) => ({
        // Error code and description
        '17': `Name is already used by RBD '${metadata.pool_name}/${
    // ...
  // ...

export class RBDFormComponent {
  // ...
  createAction() {
    const request = this.createRequest();
    // Subscribes to 'call' with submitted 'task' and handles notifications
    return this.taskWrapper.wrapTaskAroundCall({
      task: new FinishedTask('rbd/create', {
        pool_name: request.pool_name,
      call: this.rbdService.create(request)
  // ...

Error Handling in Python

Good error handling is a key requirement in creating a good user experience and providing a good API.

Dashboard code should not duplicate C++ code. Thus, if error handling in C++ is sufficient to provide good feedback, a new wrapper to catch these errors is not necessary. On the other hand, input validation is the best place to catch errors and generate the best error messages. If required, generate errors as soon as possible.

The backend provides few standard ways of returning errors.

First, there is a generic Internal Server Error:

Status Code: 500
    "version": <cherrypy version, e.g. 13.1.0>,
    "detail": "The server encountered an unexpected condition which prevented it from fulfilling the request.",

For errors generated by the backend, we provide a standard error format:

Status Code: 400
    "detail": str(e),     # E.g. "[errno -42] <some error message>"
    "component": "rbd",   # this can be null to represent a global error code
    "code": "3",          # Or a error name, e.g. "code": "some_error_key"

In case, the API Endpoints uses @ViewCache to temporarily cache results, the error looks like so:

Status Code 400
    "detail": str(e),     # E.g. "[errno -42] <some error message>"
    "component": "rbd",   # this can be null to represent a global error code
    "code": "3",          # Or a error name, e.g. "code": "some_error_key"
    'status': 3,          # Indicating the @ViewCache error status

In case, the API Endpoints uses a task the error looks like so:

Status Code 400
    "detail": str(e),     # E.g. "[errno -42] <some error message>"
    "component": "rbd",   # this can be null to represent a global error code
    "code": "3",          # Or a error name, e.g. "code": "some_error_key"
    "task": {             # Information about the task itself
        "name": "taskname",
        "metadata": {...}

Our WebUI should show errors generated by the API to the user. Especially field-related errors in wizards and dialogs or show non-intrusive notifications.

Handling exceptions in Python should be an exception. In general, we should have few exception handlers in our project. Per default, propagate errors to the API, as it will take care of all exceptions anyway. In general, log the exception by adding logger.exception() with a description to the handler.

We need to distinguish between user errors from internal errors and programming errors. Using different exception types will ease the task for the API layer and for the user interface:

Standard Python errors, like SystemError, ValueError or KeyError will end up as internal server errors in the API.

In general, do not return error responses in the REST API. They will be returned by the error handler. Instead, raise the appropriate exception.