Save Postgres execution plan in database table

Postgres database doesn't provide a built-in decent way to follow statement execution plans at different points in time as it is in Oracle, for example. Imagine an ETL query takes two hours during the night while during the day it's only two minutes. Is it a different execution plan generated? In Postgres it's pretty hard to find that out unless you're sitting there during the night and following up on it.

One actually pretty simple way to get it done could be with your ETL tool and store the execution plan output together with the generation time in a database table. Here I'm using Kettle, but the same would most likely be possible with every other software.

First create a table in the target database:

CREATE TABLE save_xplan_output
(
  query_plan text,
  x_plan_date timestamp without time zone
)

Next, create a "Table Input" step and write the following query:

Then add a "Get System Info" step to get the date and time:

Next, add the target table for the plan:

After creating the hops between elements your transformation would look like this:

After executing the transformation you can check the results:

You could put this kind of transformation somewhere in the ETL flow next to the problematic statement and you could be almost sure that the statement would have the same statement. If the execution plan differs then you can try doing something about the table statistics, the query itself. If not, then see what's going on with the environment, parallel activities etc.
Thanks for your attention!