[ 
https://issues.apache.org/jira/browse/AIRFLOW-139?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15324535#comment-15324535
 ] 

Rafael commented on AIRFLOW-139:
--------------------------------

So if I understand correctly, autocommit is no longer supported in Postgres 
9.5.2. Therefore the `autocommit=True` trick won't work for me, even with the 
patch.

My workaround is perform the vacuum manually with a little python script:

{code}
import click
import psycopg2

@click.command()
@click.argument('host')
@click.argument('dbname')
@click.argument('user')
@click.argument('password')
@click.argument('table')
def vacuum(host, dbname, user, password, table):
    conn_string = "dbname={} host={} user={} password={}"
    conn_string = conn_string.format(dbname, host, user, password)

    conn = psycopg2.connect(conn_string)
    conn.set_isolation_level(0)
    cursor = conn.cursor()
    vacuum_cmd = "VACUUM {}".format(table)
    cursor.execute(vacuum_cmd)
    conn.close()

if __name__ == "__main__":
    vacuum()
{code}

> Executing VACUUM with PostgresOperator
> --------------------------------------
>
>                 Key: AIRFLOW-139
>                 URL: https://issues.apache.org/jira/browse/AIRFLOW-139
>             Project: Apache Airflow
>          Issue Type: Bug
>    Affects Versions: Airflow 1.7.0
>            Reporter: Rafael
>
> Dear Airflow Maintainers,
> h1. Environment
> * Airflow version: *v1.7.0*
> * Airflow components: *PostgresOperator*
> * Python Version: *Python 3.5.1*
> * Operating System: *15.4.0 Darwin*
> h1. Description of Issue
> I am trying to execute a `VACUUM` command as part of DAG with the 
> `PostgresOperator`, which fails with the following error:
> {quote}
> [2016-05-14 16:14:01,849] {__init__.py:36} INFO - Using executor 
> SequentialExecutor
> Traceback (most recent call last):
>   File "/usr/local/bin/airflow", line 15, in <module>
>     args.func(args)
>   File 
> "/usr/local/Cellar/python3/3.5.1/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/airflow/bin/cli.py",
>  line 203, in run
>     pool=args.pool,
>   File 
> "/usr/local/Cellar/python3/3.5.1/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/airflow/models.py",
>  line 1067, in run
>     result = task_copy.execute(context=context)
>   File 
> "/usr/local/lib/python3.5/site-packages/airflow/operators/postgres_operator.py",
>  line 39, in execute
>     self.hook.run(self.sql, self.autocommit, parameters=self.parameters)
>   File 
> "/usr/local/Cellar/python3/3.5.1/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/airflow/hooks/dbapi_hook.py",
>  line 109, in run
>     cur.execute(s)
> psycopg2.InternalError: VACUUM cannot run inside a transaction block
> {quote}
> I could create a small python script that performs the operation, as 
> explained in [this stackoverflow 
> entry](http://stackoverflow.com/questions/1017463/postgresql-how-to-run-vacuum-from-code-outside-transaction-block).
>  However, I would like to know first if the `VACUUM` command should be 
> supported by the `PostgresOperator`.
> h1. Reproducing the Issue
> The operator can be declared as follows:
> {quote}
> conn = ('postgres_default')
> t4 = PostgresOperator(
>     task_id='vacuum',
>     postgres_conn_id=conn,
>     sql=("VACUUM public.table"),
>     dag=dag
>     )
> {quote}



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to