The majority of common CKAN administration tasks are carried out using the paster script.
Paster is run on the command line on the server running CKAN. This section covers:
paster-understanding
. Understanding paster syntax and getting help.paster-tasks
. How to carry out common CKAN admin tasks using paster.
At its simplest, paster commands can be thought of like this:
paster <ckan commands>
But there are various extra elements to the commandline that usually need adding. We shall build them up:
Paster is used for many things aside from CKAN. You usually need to tell paster that you want to enable the CKAN commands:
paster --plugin=ckan <ckan commands>
You know you need to do this if you get the error Command 'user' not known
for a valid CKAN command.
(Alternatively, CKAN commands are enabled by default if your current directory is the CKAN source directory)
Paster needs to know where your CKAN config file is (so it knows which database and search index to deal with etc.):
paster --plugin=ckan <ckan commands> --config=<config file>
If you forget to specify --config
then you will get error AssertionError: Config filename '/home/okfn/development.ini' does not exist.
(Paster defaults to looking for development.ini in the current directory.)
For example, to initialise a database:
paster --plugin=ckan db init --config=/etc/ckan/std/std.ini
You often need to run paster within your CKAN virtual environment (pyenv). If CKAN was installed as 'source' then you can activate it as usual before running the paster command:
. ~/pyenv/bin/activate
paster --plugin=ckan db init --config=/etc/ckan/std/std.ini
The alternative, which also suits a CKAN 'package' install, is to simply give the full path to the paster in your pyenv:
/var/lib/ckan/std/pyenv/bin/paster --plugin=ckan db init --config=/etc/ckan/std/std.ini
If CKAN is deployed with Apache on this machine, then you should run paster as the same user, which is usually www-data
. This is because paster will write to the same CKAN logfile as the Apache process and file permissions need to match.
For example:
sudo -u www-data /var/lib/ckan/std/pyenv/bin/paster --plugin=ckan db init --config=/etc/ckan/std/std.ini
Otherwise you will get an error such as: IOError: [Errno 13] Permission denied: '/var/log/ckan/std/std.log'
.
To get a full list of paster commands (i.e. including CKAN commands):
paster --plugin=ckan --help
And to get more detailed help on each command (e.g. on db
):
paster --plugin=ckan --help db
It is essential to run the correct paster. The program may be installed globally on a server, but in nearly all cases, the one installed in the CKAN python virtual environment (pyenv) is the one that should be used instead. This can be done by either:
Activating the virtual environment:
. pyenv/bin/activate
Giving the path to paster when you run it:
pyenv/bin/paster ...
The position of paster parameters matters.
--plugin
is a parameter to paster, so needs to come before the CKAN command. To do this, the first parameter to paster is normally --plugin=ckan
.
Note
The default value for --plugin
is setup.py
in the current directory. If you are running paster from the directory where CKAN's setup.py
file is located, you don't need to specify the plugin parameter..
Meanwhile, --config
is a parameter to CKAN, so needs to come after the CKAN command. This specifies the CKAN config file for the instance you want to use, e.g. --config=/etc/ckan/std/std.ini
Note
The default value for --config
is development.ini
in the current directory. If you are running a package install of CKAN (as described in install-from-package
), you should explicitly specify std.ini
.
The position of the CKAN command itself is less important, as longs as it follows --plugin
. For example, both the following commands have the same effect::
paster --plugin=ckan db --config=development.ini init
paster --plugin=ckan db init --config=development.ini
If you want to run a "paster shell", which can be useful for development, then the plugin is pylons. e.g. paster --plugin=pylons shell
.
Often you will want to run this as the same user as the web application, to ensure log files are written as the same user. And you'll also want to specify a config file (note that this is not specified using the --config
parameter, but simply as the final argument). For example:
sudo -u www-data paster --plugin=pylons shell std.ini
The following tasks are supported by paster.
celeryd Control celery daemon. check-po-files Check po files for common mistakes color Create or remove a color scheme. create-test-data Create test data in the database. dataset Manage datasets. datastore Perform commands to set up the datastore. db Perform various tasks on the database. front-end-build Creates and minifies css and JavaScript files less Compile all root less documents into their CSS counterparts minify Create minified versions of the given Javascript and CSS files. notify Send out modification notifications. plugin-info Provide info on installed plugins. profile Code speed profiler ratings Manage the ratings stored in the db rdf-export Export active datasets as RDF. search-index Creates a search index for all datasets sysadmin Gives sysadmin rights to a named user. tracking Update tracking statistics. trans Translation helper functions user Manage users.
For the full list of tasks supported by paster, you can run:
paster --plugin=ckan --help
Usage:
celeryd <run> - run the celery daemon
celeryd run concurrency - run the celery daemon with
argument 'concurrency'
celeryd view - view all tasks in the queue
celeryd clean - delete all tasks in the queue
Usage:
check-po-files [options] [FILE] ...
After running this command, you'll need to regenerate the css files. See less
for details.
Usage:
color - creates a random color scheme
color clear - clears any color scheme
color <'HEX'> - uses as base color eg '#ff00ff' must be quoted.
color <VALUE> - a float between 0.0 and 1.0 used as base hue
color <COLOR_NAME> - html color name used for base color eg lightblue
As the name suggests, this command lets you load test data when first setting up CKAN. See create-test-data
for details.
Usage:
dataset DATASET_NAME|ID - shows dataset properties
dataset show DATASET_NAME|ID - shows dataset properties
dataset list - lists datasets
dataset delete [DATASET_NAME|ID] - changes dataset state to 'deleted'
dataset purge [DATASET_NAME|ID] - removes dataset from db entirely
Make sure that the datastore URLs are set properly before you run these commands.
Usage:
datastore set-permissions SQL_SUPER_USER
Where:
SQL_SUPER_USER is the name of a postgres user with sufficient
permissions to create new tables, users, and grant
and revoke new permissions. Typically, this would
be the "postgres" user.
Lets you initialise, upgrade, and dump the CKAN database.
Before you can run CKAN for the first time, you need to run "db init" to create the tables in the database and the default authorization settings:
paster --plugin=ckan db init --config=/etc/ckan/std/std.ini
If you forget to do this then CKAN won't serve requests and you will see errors such as this in the logs:
ProgrammingError: (ProgrammingError) relation "user" does not exist
You can delete everything in the CKAN database, including the tables, to start from scratch:
paster --plugin=ckan db clean --config=/etc/ckan/std/std.ini
The next logical step from this point is to do a "db init" step before starting CKAN again.
You can 'dump' (save) the exact state of the database to a file on disk and at a later point 'load' (restore) it again, or load it on another machine.
To write the dump:
paster --plugin=ckan db dump --config=/etc/ckan/std/std.ini std.pg_dump
To load it in again, you first have to clean the database of existing data (be careful not to wipe valuable data), followed by the load:
paster --plugin=ckan db clean --config=/etc/ckan/std/std.ini std.pg_dump
paster --plugin=ckan db load --config=/etc/ckan/std/std.ini std.pg_dump
When you upgrade CKAN software by any method other than the package update described in install-from-package
, before you restart it, you should run 'db upgrade', which will do any necessary migrations to the database tables:
paster --plugin=ckan db upgrade --config=/etc/ckan/std/std.ini
For information on using db
to create dumpfiles, see database-dumps
.
Usage:
front-end-build
Usage:
less
Usage:
paster minify [--clean] PATH
For example:
paster minify ckan/public/base
paster minify ckan/public/base/css/*.css
paster minify ckan/public/base/css/red.css
If the --clean option is provided any minified files will be removed.
Usage:
notify replay - send out modification signals. In "replay" mode,
an update signal is sent for each dataset in the database.
As the name suggests, this commands shows you the installed plugins, their description, and which interfaces they implement
Provide a ckan url and it will make the request and record how long each function call took in a file that can be read by runsnakerun.
Usage:
profile URL
The result is saved in profile.data.search. To view the profile in runsnakerun:
runsnakerun ckan.data.search.profile
You may need to install the cProfile python module.
Manages the ratings stored in the database, and can be used to count ratings, remove all ratings, or remove only anonymous ratings.
For example, to remove anonymous ratings from the database:
paster --plugin=ckan ratings clean-anonymous --config=/etc/ckan/std/std.ini
This command dumps out all currently active datasets as RDF into the specified folder:
paster rdf-export /path/to/store/output
Rebuilds the search index. This is useful to prevent search indexes from getting out of sync with the main database.
For example:
paster --plugin=ckan search-index rebuild --config=/etc/ckan/std/std.ini
This default behaviour will clear the index and rebuild it with all datasets. If you want to rebuild it for only one dataset, you can provide a dataset name:
paster --plugin=ckan search-index rebuild test-dataset-name --config=/etc/ckan/std/std.ini
Alternatively, you can use the -o or --only-missing option to only reindex datasets which are not already indexed:
paster --plugin=ckan search-index rebuild -o --config=/etc/ckan/std/std.ini
If you don't want to rebuild the whole index, but just refresh it, use the -r or --refresh option. This won't clear the index before starting rebuilding it:
paster --plugin=ckan search-index rebuild -r --config=/etc/ckan/std/std.ini
There are other search related commands, mostly useful for debugging purposes:
search-index check - checks for datasets not indexed
search-index show DATASET_NAME - shows index of a dataset
search-index clear [DATASET_NAME] - clears the search index for the provided dataset or for the whole ckan instance
Gives sysadmin rights to a named user. This means the user can perform any action on any object.
For example, to make a user called 'admin' into a sysadmin:
paster --plugin=ckan sysadmin add admin --config=/etc/ckan/std/std.ini
Usage:
tracking update [start_date] - update tracking stats
tracking export FILE [start_date] - export tracking stats to a csv file
Usage:
trans js - generate the javascript translations
trans mangle - mangle the zh_TW translations for testing
Lets you create, remove, list and manage users.
For example, to create a new user called 'admin':
paster --plugin=ckan user add admin --config=/etc/ckan/std/std.ini
To delete the 'admin' user:
paster --plugin=ckan user remove admin --config=/etc/ckan/std/std.ini