User facing APIs to control AiiDA from the verdi cli, scripts or plugins
Provides an API for postgres database maintenance tasks.
This API creates and drops postgres users and databases used by the
verdi quicksetup commandline tool. It allows convenient access to this
functionality from within python without knowing details about how postgres is
installed by default on various systems. If the postgres setup is not the
default installation, additional information needs to be provided.
- class aiida.manage.external.postgres.Postgres(dbinfo=None, **kwargs)#
Adds convenience functions to
- Provides convenience functions for
postgres = Postgres() postgres.create_dbuser('username', 'password') if not postgres.db_exists('dbname'): postgres.create_db('username', 'dbname')
- __init__(dbinfo=None, **kwargs)#
See documentation of
- __module__ = 'aiida.manage.external.postgres'#
Looks up if a database with the name exists, prompts for using or creating a differently named one.
dbname (str) – Name of the database to be created or reused.
tuple (dbname, created)
Looks up if a given user already exists, prompts for using or creating a differently named one.
dbuser (str) – Name of the user to be created or reused.
tuple (dbuser, created)
- copy_db(src_db, dest_db, dbuser)#
- create_db(dbuser, dbname)#
Create a database in postgres
- create_dbuser(dbuser, dbpass, privileges='')#
Create a database user in postgres
- create_dbuser_db_safe(dbname, dbuser, dbpass)#
Create DB and user + grant privileges.
Prompts when reusing existing users / databases.
Check wether a postgres database with dbname exists
dbname (str) – Name of the database to check for
(bool), True if database exists, False otherwise
- property dbinfo#
Alias for Postgres.dsn.
Find out if postgres user with name dbuser exists
dbuser (str) – database user to check for
(bool) True if user exists, False otherwise
Drop a database user in postgres
dbuser (str) – Name of the user to be dropped.
- classmethod from_profile(profile: Profile, **kwargs)#
Create Postgres instance with dbinfo from AiiDA profile data.
- Note: This only uses host and port from the profile, since the others are not going to be relevant for the
profile – AiiDA profile instance
kwargs – keyword arguments forwarded to PGSU constructor
Postgres instance pre-populated with data from AiiDA profile
- property host_for_psycopg2#
Return correct host for psycopg2 connection (as required by regular AiiDA operation).
- property port_for_psycopg2#
Return port for psycopg2 connection (as required by regular AiiDA operation).
- aiida.manage.external.postgres.manual_setup_instructions(dbuser, dbname)#
Create a message with instructions for manually creating a database
Components to communicate tasks to RabbitMQ.
- class aiida.manage.external.rmq.ProcessLauncher(loop: Optional[asyncio.events.AbstractEventLoop] = None, persister: Optional[plumpy.persistence.Persister] = None, load_context: Optional[plumpy.persistence.LoadSaveContext] = None, loader: Optional[plumpy.loaders.ObjectLoader] = None)#
A sub class of plumpy.ProcessLauncher to launch a Process.
It overrides the _continue method to make sure the node corresponding to the task can be loaded and that if it is already marked as terminated, it is not continued but the future is reconstructed and returned
- __module__ = 'aiida.manage.external.rmq'#
- async _continue(communicator, pid, nowait, tag=None)#
Continue the task.
Note that the task may already have been completed, as indicated from the corresponding the node, in which case it is not continued, but the corresponding future is reconstructed and returned. This scenario may occur when the Process was already completed by another worker that however failed to send the acknowledgment.
communicator – the communicator that called this method
pid – the pid of the process to continue
nowait – if True don’t wait for the process to finish, just return the pid, otherwise wait and return the results
tag – the tag of the checkpoint to continue from
- static handle_continue_exception(node, exception, message)#
Handle exception raised in _continue call.
If the process state of the node has not yet been put to excepted, the exception was raised before the process instance could be reconstructed, for example when the process class could not be loaded, thereby circumventing the exception handling of the state machine. Raising this exception will then acknowledge the process task with RabbitMQ leaving an uncleaned node in the CREATED state for ever. Therefore we have to perform the node cleaning manually.
exception – the exception object
message – string message to use for the log message
Try to store the values in the input dictionary.
For nested dictionaries, the values are stored by recursively.
Return the launch queue name with an optional prefix.
launch queue name
Return the message exchange name for a given prefix.
message exchange name
- aiida.manage.external.rmq.get_rmq_url(protocol=None, username=None, password=None, host=None, port=None, virtual_host=None, **kwargs)#
Return the URL to connect to RabbitMQ.
The default of the
hostis set to
localhostbecause on some computers localhost resolves first to IPv6 with address ::1 and if RMQ is not running on IPv6 one gets an annoying warning. For more info see: https://github.com/aiidateam/aiida-core/issues/1142
protocol – the protocol to use, amqp or amqps.
username – the username for authentication.
password – the password for authentication.
host – the hostname of the RabbitMQ server.
port – the port of the RabbitMQ server.
virtual_host – the virtual host to connect to.
kwargs – remaining keyword arguments that will be encoded as query parameters.
the connection URL string.
Return the task exchange name for a given prefix.
task exchange name