connection: pessimistic disconnect handling

Adds ``pool_pre_ping`` for pessimistic disconnect handling. Helps to
reconnect to database from stalled connections from long-running
pods. Closes reanahub#58. Addresses reanahub#133.

CHANGES.rst

@@ -7,6 +7,7 @@ Version 0.8.0 (UNRELEASED)
 - Adds new disk usage retrieval methods using canonical (bytes) and human-readable (KiB) units. (``User``, ``Workflow``)
 - Adds new properties ``started_at`` and ``finished_at`` to the ``Job`` model, updated on status change.
 - Adds ``get_priority`` workflow method, that combines both complexity and concurrency, to pass to the scheduler.
+- Adds database connection pool pre-ping to prevent stalled database connections problems.
 - Changes disk quota calculation functions to allow passing raw bytes to increase the used quota.
 - Removes support for Python 2.
 

reana_db/database.py

@@ -1,7 +1,7 @@
 # -*- coding: utf-8 -*-
 #
 # This file is part of REANA.
-# Copyright (C) 2018 CERN.
+# Copyright (C) 2018, 2019, 2020, 2021 CERN.
 #
 # REANA is free software; you can redistribute it and/or modify it
 # under the terms of the MIT License; see LICENSE file for more details.
@@ -19,7 +19,7 @@
 
 from reana_db.models import Base  # isort:skip  # noqa
 
-engine = create_engine(SQLALCHEMY_DATABASE_URI)
+engine = create_engine(SQLALCHEMY_DATABASE_URI, pool_pre_ping=False)
 Session = scoped_session(sessionmaker(autocommit=False, autoflush=False, bind=engine))
 Base.query = Session.query_property()