##############################################################################
#
# Copyright (c) 2009 Zope Foundation and Contributors.
# All Rights Reserved.
#
# This software is subject to the provisions of the Zope Public License,
# Version 2.1 (ZPL). A copy of the ZPL should accompany this distribution.
# THIS SOFTWARE IS PROVIDED "AS IS" AND ANY AND ALL EXPRESS OR IMPLIED
# WARRANTIES ARE DISCLAIMED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
# WARRANTIES OF TITLE, MERCHANTABILITY, AGAINST INFRINGEMENT, AND FITNESS
# FOR A PARTICULAR PURPOSE.
#
##############################################################################
"""IObjectMover implementation.
"""
from __future__ import absolute_import
from __future__ import print_function
from zope.interface import implementer
from relstorage.adapters.interfaces import IObjectMover
from ..schema import Schema
from ..mover import AbstractObjectMover
from ..mover import metricmethod_sampled
[docs]
@implementer(IObjectMover)
class MySQLObjectMover(AbstractObjectMover):
_create_temp_store = Schema.temp_store.create()
[docs]
@metricmethod_sampled
def on_store_opened(self, cursor, restart=False):
"""Create the temporary table for storing objects"""
if restart:
# TRUNCATE is a DDL statement, even against a temporary
# table, and as such does an implicit transaction commit.
# Normally we want to avoid that, but here its OK since
# this method is called between transactions, as it were.
# TRUNCATE benchmarks (zodbshoot add) substantially faster
# (10157) than a DELETE (75xx) and moderately faster than
# a DROP/CREATE (9457). TRUNCATE is in the replication
# logs like a DROP/CREATE. (DROP TEMPORARY TABLE is *not*
# DDL and not transaction ending).
# We are up to 4 temp tables, doing this with a call to a stored
# proc saves round trips.
#
# It's possible that the DDL lock that TRUNCATE takes can be a bottleneck
# in some places, though?
self.driver.callproc_no_result(
cursor,
"clean_temp_state(false)"
)
else:
# InnoDB tables benchmark much faster for concurrency=2
# and 6 than MyISAM tables under both MySQL 5.5 and 5.7,
# at least on OS X 10.12. The OS X filesystem is single
# threaded, though, so the effects of flushing MyISAM tables
# to disk on every operation are probably magnified.
# note that the md5 column is not used if self.keep_history == False.
self._create_temp_store.execute(cursor)
stmt = """
CREATE TEMPORARY TABLE temp_read_current (
zoid BIGINT UNSIGNED NOT NULL PRIMARY KEY,
tid BIGINT UNSIGNED NOT NULL
) ENGINE InnoDB
"""
cursor.execute(stmt)
stmt = """
CREATE TEMPORARY TABLE temp_blob_chunk (
zoid BIGINT UNSIGNED NOT NULL,
chunk_num BIGINT UNSIGNED NOT NULL,
PRIMARY KEY (zoid, chunk_num),
chunk LONGBLOB
) ENGINE InnoDB
"""
cursor.execute(stmt)
stmt = """
CREATE TEMPORARY TABLE IF NOT EXISTS temp_locked_zoid (
zoid BIGINT UNSIGNED NOT NULL PRIMARY KEY
) ENGINE InnoDB
"""
cursor.execute(stmt)
AbstractObjectMover.on_store_opened(self, cursor, restart=restart)
[docs]
@metricmethod_sampled
def restore(self, cursor, batcher, oid, tid, data):
"""Store an object directly, without conflict detection.
Used for copying transactions into this database.
"""
if self.version_detector.requires_values_upsert_alias(cursor):
alias = 'AS NEW'
def VALUES(col):
return 'NEW.' + col
else:
alias = ''
def VALUES(col):
return f'VALUES({col})'
if self.keep_history:
suffix = f"""
{alias}
ON DUPLICATE KEY UPDATE
tid = {VALUES('tid')},
prev_tid = {VALUES('prev_tid')},
md5 = {VALUES('md5')},
state_size = {VALUES('state_size')},
state = {VALUES('state')}
"""
else:
suffix = f"""
{alias}
ON DUPLICATE KEY UPDATE
tid = {VALUES('tid')},
state_size = {VALUES('state_size')},
state = {VALUES('state')}
"""
self._generic_restore(batcher, oid, tid, data,
command='INSERT', suffix=suffix)
# Override this query from the superclass. The MySQL optimizer, up
# through at least 5.7.17 doesn't like actual subqueries in a DELETE
# statement. See https://github.com/zodb/relstorage/issues/175
_move_from_temp_hf_delete_blob_chunk_query = """
DELETE bc
FROM blob_chunk bc
INNER JOIN (SELECT zoid FROM temp_store) sq
ON bc.zoid = sq.zoid
"""
# We UPSERT for hf movement; no need to do a delete.
_move_from_temp_hf_delete_query = ''